x\
0
(5

-§....

This is to certify that the
dissertation entitled

Variable Input and the Acquisition of Plurallty in Two Varieties
in Spanish

presented by

Karen Lynn Miller

has been accepted towards fulﬁllment
of the requirements for the

Doctoral degree in Uﬂuistics

 

 

02, Cl?

 

Maior Prof ssor’s Signature

3- lu~o7f

 

Date

MSU is an Afﬁrmative Action/Equal Opportunity Institution

- -—-.--- -.-- ->-.-t—-r§.— —.- . -

 

 

 

LIBRARY

Michigan State

University

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p2/CIRC/Date0uerindd-p1

 

VARIABLE INPUT AND THE ACQUISITION OF PLURALITY IN TWO
VARIETIES OF SPANISH

By

Karen Lynn Miller

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Linguistics and Germanic, Slavic, Asian and African Languages

2007

ABSTRACT

VARIABLE INPUT AND THE ACQUISITION OF PLURALITY IN TWO
VARIETIES OF SPANISH

By

Karen Lynn Miller

This dissertation presents a series of production and comprehension experiments
designed to test how variable and ambiguous input affects the acquisition of grammatical
morphology in children. Acquisition of plural morphology was examined in two varieties
of Spanish, one where the plural morpheme is sometimes omitted in adult speech and the
other where the plural morpheme is consistently produced by adult speakers. The results
show that clear differences exist in the way that children acquire grammatical
morphology that is consistently produced in the input vs. grammatical morphology that is
variable and ambiguous in the input. That is, in production children exposed to consistent
input, produce the plural morpheme consistently in their own speech, while children
exposed to variable input are variable in their own production. In comprehension, the
results show that children exposed to consistent input associate the plural morpheme to
an interpretation of ‘more than one’ by at least 4 years of age, while children exposed to
variable and ambiguous input are delayed in their comprehension of plural morphology.
Speciﬁcally, these children do not assign an interpretation of ‘more than one’ to the
plural morpheme until approximately 7 years of age. The results of this dissertation
strongly suggest that variable and ambiguous input delays the acquisition of grammatical

morphology that is affected by that variability.

Cepyright by
KAREN LYNN MILLER
2007

ACKNOWLEDGMENTS

The topic of this doctoral dissertation came about purely by accident. In
December of 2003 I was carrying out a pilot study in Chile that examined Spanish-
speaking children’s ability to assign a wide-scope interpretation to plural indeﬁnites. To
my great surprise, several Chilean children were consistently treating the plural indeﬁnite
as singular, associating it to an interpretation of ‘one’. After long discussion with my
advisor, Cristina Schmitt, we hypothesized that this pattern might be due to the process of
syllable-ﬁnal /s/ lenition in Chilean Spanish, which affects the plural morpheme and
creates an ambiguous input for Chilean children. At that moment, it occurred to me that
language acquisition research must examine not only the effect of the ﬁequencies in the
input but also the effect of different types of input (consistent vs. ambiguous) on language
development. Moreover, upon reading Kroch (1994) and Yang (2001), I began to
understand how this topic is related more generally to language change. As a result of this
ﬁrst pilot study and several discussions with Cristina, subsequent research was carried
out over the next two years in Chile and also in Mexico City, where no such lenition
process takes place. The results of this research are presented here in this dissertation.

There are many people who I would like to thank for helping me carry out and
develop the research presented in this dissertation. First and foremost, I would like to
thank my advisor, Cristina Schmitt, to whom I will be eternally indebted for the
enormous amount of time she spent teaching and training me during my graduate studies
at MSU. The research presented in this dissertation would not have developed in the way
that it has if not for Cristina’s very insightful guidance and expertise and I am extremely

flattered that she took interest in my research and education at MSU from the start.

iv

Cristina is the best teacher I have ever had and has had a tremendous impact on my
learning over the last several years. I would also like to thank Alan Munn, who was
extremely generous with his time and expertise, both as a committee member and as my
teacher at MSU. I am very grateful for his frank and honest comments on my graduate
work and greatly appreciate his helpful suggestions on the experimental studies that were
run. I thank Ana Teresa Pe'rez-Leroux, Barbara Abbot, Patricia Lunn, Yen-Hwei Lin and,
Marcin Morzycki, who served as members on my doctoral committee, for their very
useful comments and discussion of the experimental studies being carried out. I am also
indebted to John Grinstead and Antoinette Hawayek for helping me set up the
experimental studies in Mexico City. They both very generously dedicated their time and
expertise and without their help, this dissertation would not have been possible.

There were many undergraduate and graduate students who spent long hours
working on this project. I thank Katerina French for the many tedious hours she spent
meticulously transcribing and coding the Chilean and Mexican data and for her useful
and very insightful comments about the production data. I also thank Andrew Sanford
and Pascale Schnitzer for their help transcribing and coding data. I thank Rodrigo
Cardenas, Cynthia Corona, Marena Garcia, Edgardo Mansilla, Erika Mendoza, and
Heriberto Sierra for their help in carrying out the experimental studies with children in
Chile and Mexico. I also appreciate the comments and suggestions made to me by the
members of the Michigan State University Language Acquisition Lab, and the audiences
at BUCLD 30, GLOW 2006, the Hispanic Linguistic Symposium 2004, 2005, the Calvin

College Foreign Language Forum, NWAV 2004, 2006, and X IASCL.

This research would have been impossible without the cooperation of many
schools and daycares in Mexico and Chile. For this reason, I thank the parents,
administrators and children at the following schools in Punta' Arenas, CHILE: Colegio
Aleman, Colegio Britanico, Colegio Pierre Faure, Junta Nacional de Jardines Infantiles
de Chile (JUNJI), Jardin Bambi, Jardin Las Charitas, Escuela 18 de Septiembre, Colegio
Miguel de Cervantes, and the Universidad de Magallanes, and also in Mexico: Centro de
Desarrollo Infantil (CENDI) and the Universidad Auténoma Metropolitana (UAM),
Campus Iztapalapa of México, D.F.

In order to carry out the experimental research presented in this dissertation, I had
to travel quite a bit to Chile and Mexico and had to pay a number of undergraduate and
graduate research assistants for their work. This would have been impossible for me if
not for the ﬁnancial support from the following grants: NSF Doctoral Dissertation
Improvement Grant #0446769, NSF Grant #BCS-0126502, the Michigan State
University Graduate Student Research Enhancement Award, the Tinker Field Research
Grant from the Center for Latin American and Caribbean Studies at MSU, and the
Walker Hill International Award ﬁom International Studies and Programs at MSU.

Finally, I would like to thank my family for being enthusiastic supporters of my
graduate studies. I especially appreciate my mother, Julie Swain, for her encouragement,
unique insight and great sense of humor, which made the process of writing and editing

this dissertation (and graduate life in general) even more enjoyable.

vi

TABLE OF CONTENTS

LIST OF TABLES ................................................................ ' .......................................... x
LIST OF FIGURES ...................................................................................................... xiii
CHAPTER 1: INTRODUCTION .................................................................................... 1
1.0 Introduction ........................................................................................................ l
1.1 Statement of Problem .......................................................................................... 8
1.2 Theoretical Importance ..................................................................................... 18
1.3 Applied Importance .......................................................................................... 19
1.4 Organization of the Chapters ............................................................................. 21
CHAPTER 2: A LINGUISTIC DESCRIPTION OF PLURAL MORPHOLOGY .......... 23
2.0 Introduction ...................................................................................................... 23
2.1 Spanish Plural Morphology ............................................................................... 25
2.2 The Phonological Form of the Spanish Plural .................................................... 37
2.2.1 Mexico City Spanish .................................................................................. 38
2.2.2 Chilean Spanish ......................................................................................... 39
2.2.3 Syllable Final /s/ Lenition in Other Varieties of Spanish ............................. 46

2.3 The Underlying Syntactic Representation of the Spanish Plural ........................ 49
2.4 The Underlying Semantic Representation of the Spanish Plural ........................ 55
2.5 Implications for Interpretation by Spanish Speakers .......................................... 61
CHAPTER 3: PREVIOUS LANGUAGE ACQUISITION RESEARCH ....................... 69
3.0 Introduction ...................................................................................................... 69
3.1 Acquisition of Plural Morphology ..................................................................... 74
3.1.1 Production of Plural Morphology ............................................................... 74
3.1.2 Placement of the Plural Morpheme in Production ....................................... 84
3.1.3 Comprehension of Plural Morphology ........................................................ 87

3.2 Acquisition of Variable Input ............................................................................ 92
3.3 Summary of Previous Research ....................................................................... 108
CHAPTER 4: PRODUCTION OF PLURAL MORPHOLOGY ................................... 110
4.0 Introduction .................................................................................................... l 10
4.1 Method and Design ......................................................................................... 112
4.2 Subjects .......................................................................................................... 1 15
4.3 Procedure ........................................................................................................ 115
4.4 Analysis .......................................................................................................... 116
4.5 Results ............................................................................................................ 124
4.5.1 The Effect of Style on Plural Morpheme Production ................................ 134
4.5.2 The Effect of Syntactic Category on Plural Morpheme Production ........... 138
4.5.3 The Effect of Order on Plural Morpheme Production ............................... 140
4.5.4 The Effect of Number Information on Plural Morpheme Production ........ 143

vii

4.6 Summary of Results ........................................................................................... 146

4.7 Discussion .......................................................................................................... 147
CHAPTER 5: COMPREHENSION OF PLURAL MORPHOLOGY ........................... 151
5.0 Introduction ............................................................... . ........................................ 15 1
5.1 Study 1. Act-out Task: Indeﬁnite Noun Phrase ................................................. 154
5.1.1 Background ................................................................................................. 154
5.1.2 Method and Design ..................................................................................... 157
5.1.3 Subjects ....................................................................................................... 160

5. 1 .4 Procedure .................................................................................................... 163
5.1.5 Results ......................................................................................................... 165
5.1.6 Highlighting Task ....................................................................................... 182
5.1.7 Feminine vs. Masculine Nouns ................................................................... 183
5.1.8 Discussion ................................................................................................... 185
5.1.9 Summary of Findings .................................................................................. 190
5.2 Study 2. Picture Matching Task: Indeﬁnite Noun Phrases ................................ 192
5.2.1 Background ................................................................................................. 192
5.2.2 Method and Design ..................................................................................... 193
5.2.3 Subjects ....................................................................................................... l 94
5.2.4 Procedure .................................................................................................... 195
5.2.5 Results ......................................................................................................... 196
5.2.6 Feminine vs. Masculine Nouns ................................................................... 200
5.2.7 Discussion ................................................................................................... 201
5.2.8 Summary of Findings .................................................................................. 202
5.3 Study 3. Picture Matching Task: Indeﬁnite Noun Phrases ................................ 203
5.3.1 Background .................. 203
5.3.2 Method and Design ..................................................................................... 204
5.3.3 Subjects ....................................................................................................... 207
5.3.4 Procedure .................................................................................................... 208
5.3.5 Results ......................................................................................................... 209
5.3.6 Feminine vs. Masculine Nouns ................................................................... 213
5.3.7 Discussion ................................................................................................... 214
5.3.8 Summary of Findings .................................................................................. 215
5.4 Study 4. Elicitation Task: Bare Nouns and Indeﬁnite Noun Phrases ................. 216
5.4.1 Background ................................................................................................. 21 7
5.4.2 Method and Design ..................................................................................... 217
5.4.3 Subjects ....................................................................................................... 219
5.4.4 Procedure .................................................................................................... 220
5.4.5 Results ......................................................................................................... 220
5.4.6 Discussion ................................................................................................... 227
5.4.7 Summary of Findings .................................................................................. 228
5.5 Study 5. Picture Matching Task: Bare Singulars vs. Bare Plurals ...................... 229
5.5.1 Background ................................................................................................. 229
5.5.2 Method and Design ..................................................................................... 230
5.5.3 Subjects ....................................................................................................... 233
5.5.4 Procedure .................................................................................................... 234

viii

5.5.5 Results ..................................................................................................... 234

5.5.6 Discussion ................................................................................................ 241
5.5.7 Summary of Findings ............................................................................... 242
5.6 Study 6. Act-out Task: Inalienable Possession ................................................ 242
5.6.1 Background ........................................................ ' ...................................... 2 43
5.6.2 Method and Design .................................................................................. 244
5.6.3 Subjects ................................................................................................... 245
5.6.4 Procedure ................................................................................................. 246
5.6.5 Results ..................................................................................................... 247
5.6.6 Discussion ................................................................................................ 250
5.6.7 Summary of Findings ............................................................................... 252
5.7 Study 7. Act-out Task: Deﬁnite Noun Phrases ................................................ 252
5.7.1 Background .............................................................................................. 253
5.7.2 Method and Design .................................................................................. 254
5.7.3 Subjects ................................................................................................... 256
5.7.4 Procedure ................................................................................................. 256
5.7.5 Results ..................................................................................................... 257
5.7.6 Discussion ................................................................................................ 261
5.7.7 Summary of Findings ............................................................................... 262
5.8 Summary and Discussion ................................................................................ 262
CHAPTER 6: SUMMARY AND CONCLUSIONS .................................................... 265
6.1 Goal of Dissertation ........................................................................................ 265
6.2 Summary and Discussion of Findings ............................................................. 267
6.3 Conclusions .................................................................................................... 271
6.4 Future Research .............................................................................................. 273
APPENDICES ............................................................................................................ 275
APPENDIX A: Production: Study 1. Repetition Task .............................................. 276
APPENDIX B: Study 2. Picture Matching Task: Indeﬁnite Noun Phrases ............... 278
APPENDIX C: Study 3. Picture Matching Task: Indeﬁnite Noun Phrases ............... 282
APPENDIX D: Study 4. Elicitation Task: Bare Nouns and Indeﬁnite Noun Phrases 292
APPENDIX E: Study 5. Picture Matching Task: Bare Noun Phrases ....................... 297
APPENDIX F: Study 6. Act-out Task: Deﬁnites in Inalienable Possession .............. 307
APPENDIX G: Study 7. Act-out Task: Deﬁnite Noun Phrases ................................ 309
BIBLIOGRAPHY ....................................................................................................... 313

ix

LIST OF TABLES

Table l. Verbal Morphology: Chilean Spanish Present Tense. ....................................... 28
Table 2. Spanish Determiners. ....................................................................................... 30
Table 3. Number Marking on Spanish Demonstratives. ................................................. 34
Table 4. The Spanish Plural Morpheme. ........................................................................ 38
Table 5. Plural Morpheme Lenition by Syntactic Category in Chilean Spanish. ............. 42

Table 6. Plural Morpheme Lenition in Mono- and Polysyllabic Noun Premodiﬁers. ...... 43

Table 7. The Effect of Phonological Environment on /s/ Lenition in Chilean Spanish. ...44

Table 8. Social Distribution of Syllable Final /s/ in Chilean Spanish. ............................. 45
Table 9. Hypercorrection in Dominican Spanish. ........................................................... 47
Table 10. Chilean Adult Speech Samples ....................................................................... 64
Table 11. Variation in Auxiliary Production. ................... 71

Table 12. Percentage of Systematic Responses in Hudson Kam and Newport (2005)... 106

Table 13. /s/ Lenition by Phonological Environment in Chilean Spanish. .................... 113
Table 14. Factor Groups in Analysis of Chilean Production Data ................................. 119
Table 15. Ordering of Lexical Items within the DP. ..................................................... 123
Table 16. Factor Groups and Factor Weights for Chilean Production Data ................... 126
Table 17. Percentage of Plural Responses for Production Data. ................................... 128
Table 18. Percentage of Plural Variants in Production Data. ........................................ 131
Table 19. Number of Plural Tokens by Subject ........................................................... 133
Table 20. The Effect of Style on Plural Morpheme Production. ................................... 136
Table 21. The Effect of Syntactic Category on Plural Morpheme Production. .............. 139

Table 22. The Effect of Linear Order on Plural Morpheme Production. ....................... 141

Table 23.

Table 24.

Table 25.

Table 26.

Table 27.

Table 28.

Table 29.

Table 30.

Table 31.

Table 32.

Table 33.

Table 34.

Table 35.

Table 36.

Table 37.

Table 38.

Table 39.

Table 40.

Table 41.

Table 42.

Table 43.

Table 44.

The Effect of Order on Plural Morpheme Production. .................................. 142
Coding of Number Information in Production Data. ..................................... 144
The Effect of Number Information on Plural Morpheme Production. ........... 145
Free Speech Samples of Plural Morphology Produced by Chilean Adults. 149
Percentage of Plural Responses for Production Data. ................................... 157
Lenition of Syllable Final /s/ by Phonological Environment. ........................ 158
Study 1: Distribution of Children. ................................................................ 162
Study 1: Highlighting Task. ......................................................................... 164
Study 1: Plural Responses in Control Conditions. ........................................ 166
Study 1: Plural Responses in Target Conditions. .......................................... 168
Study 1: Unos: Systematic Responders. ....................................................... 171
Study 1: Algunos: Systematic Responders................................... ................. 174
Study 12A1gunos de los: Systematic Responders. ......................................... 177
Study 1: Un: Systematic Responders. ........................................................... 178
Study 1: Uno de los: Systematic Responders. ............................................... 181
Study 1: Highlighting Task: Plural Responses. ............................................. 182
Study 1: Plural Responses in Feminine and Masculine Trials. ...................... 185
Study 2: Distribution of Children. ................................................................ 195
Study 2: Plural Responses in Target Conditions. .......................................... 197
Study 2: Unos: Systematic Responders. ....................................................... 198
Study 2: Un: Systematic Responders. ........................................................... 199
Study 2: Plural Responses in Feminine and Masculine Trials. ...................... 200

xi

Table 45. Study 3: Distribution of Children. ................................................................ 208
Table 46. Study 3: Plural Responses in Target Conditions. .......................................... 210
Table 47. Study 3: Unos: Systematic Responders. ...................................... 212
Table 48. Study 3: Un: Systematic Responders. ........................................................... 213
Table 49. Study 3: Plural Responses in Feminine and Masculine Trials. ...................... 214
Table 50. Study 4: Target Words Elicited. ................................................................... 219
Table 51. Study 4: Distribution of Children. ................................................................ 220
Table 52. Study 4: Percentage of Noun Types Produced. ............................................. 225
Table 53. Study 5: Plural Responses in Target Conditions. .......................................... 235
Table 54. Study 5: Bare Plural: Systematic Responders. .............................................. 237
Table 55. Study 5: Bare Singular: Systematic Responders. .......................................... 237
Table 56. Comparison of Ages in Study 3 and Study 5. ............................................... 239
Table 57. Study 6: Distribution of Children. ................................................................ 246
Table 58. Study 6: Plural Responses in Target Conditions. .......................................... 247
Table 59. Study 6: Las: Systematic Responders. .......................................................... 249
Table 60. Study 6: La: Systematic Responders. ........................................................... 250
Table 61. Study 7: Plural Responses in Target Conditions. .......................................... 258
Table 62. Study 7: Las: Systematic Responders. .......................................................... 260
Table 63. Study 7: La: Systematic Responders. ........................................................... 260
Table 64. APPENDIX A: Study 1 Repetition Task. ..................................................... 276
Table 65. APPENDIX F: Study 6 Act-out Task. .......................................................... 307
Table 66. APPENDIX G: Study 7 Act-out Task. ......................................................... 309

xii

 

LIST OF FIGURES

Figure l. The Dynamics of Language Acquisition (adapted from Yang 2001). ................ 2
Figure 2. Linguistic Input for Plural Morphology in Chilean Spanish. ............................. 6
Figure 3. Linguistic Input for Plural Morphology in Mexican Spanish. ............................ 7

Figure 4. Mereological Representation of a Nominal Root (adapted ﬁ'om Ojeda 1998). .57

Figure 5. Mereological Representation of a Plural Noun (adapted from Ojeda 1998). 57

Figure 6. Mereological Representation of a Singular Noun (adapted from Ojeda 1998). 58

Figure 7. The Wug Test (Berko 1958). .......................................................................... 76
Figure 8. Experimental Display in Munn et a1. (2006) .................................................... 90
Figure 9. Experimental Display in Pérez-Leroux (2005). ............................................... 94
Figure 10. Study 2: Sample Target Trial. ..................................................................... 194
Figure 11. Study 3: Sample Target Trial. ................................................... . ................. 205
Figure 12. Study 4: Sample Target Trial. ..................................................................... 218
Figure 13. Study 4: Percentage of Nouns with Plural Variant [s] .................................. 221
Figure 14. Study 5: Sample Target Trial. ..................................................................... 232
Figure 15. Comparison of Plural Responses in Study 3 and Study 5 ............................. 240
Figure 16. Study 6: Sample Target Trial ...................................................................... 244
Figure 17. Study 7: Diagram of Sample Target Trial. .................................................. 254
Figure 18. Study 7: Sample Experimental Story ........................................................... 257
Figure 19. APPENDIX B: Study 2 Barcos ‘ships’. ...................................................... 278
Figure 20. APPENDIX B: Study 2 Monedas ‘coins’. ................................................... 278
Figure 21. APPENDIX B: Study 2 Botellas ‘bottles’. .................................................. 279

xiii

Figure 22.
Figure 23.
Figure 24.
Figure 25.
Figure 26.
Figure 27.
Figure 28.
Figure 29.
Figure 30.
Figure 31.
Figure 32.
Figure 33.
Figure 34.
Figure 35.
Figure 36.
Figure 37.
Figure 38.
Figure 39.
Figure 40.
Figure 41.
Figure 42.
Figure 43.

Figure 44.

APPENDIX B:

APPENDIX B:

APPENDIX B:

APPENDIX B:

APPENDIX B:

APPENDIX C:

APPENDIX C:

APPENDIX C:

APPENDIX C:

APPENDIX C:

APPENDIX C:

APPENDIX C:

APPENDIX C:

APPENDIX C:

APPENDIX C:

APPENDIX D:

APPENDIX E:

APPENDIX E:

APPENDIX E:

APPENDIX E:

APPENDIX E:

APPENDIX E:

APPENDIX E:

Study 2 Monos (Changos for Mexico City) ‘monkeys’ ....... 279
Study 2 Burros ‘donkeys’ ................................................... 280
Study 2 Martillos ‘hammers’ ......... ' ..................................... 2 80
Study 2 Bolitas (Canicas in Mexico City) ‘marbles’ ........... 281
Study 2 Manzanas ‘apples’. ............................................... 281
Study 3 Target Story: Introduction to Characters. ............... 282
Study 3 Target Story: Narrative l. ...................................... 283
Study 3 Target Story: Narrative 2 ....................................... 284
Study 3 Target Story: Narrative 3. ...................................... 285
Study 3 Target Story: Narrative 4. ...................................... 286
Study 3 Control Story: Introduction to Characters. ............. 287
Study 3 Control Story: Narrative 1. .................................... 288
Study 3 Control Story: Narrative 2. .................................... 289
Study 3 Control Story: Narrative 3. .................................... 290
Study 3 Control Story: Narrative 4. .................................... 291
Study 4 Elicitation Task. .................................................... 292
Study 5 Target Story: Introduction to Characters. ............... 297
Study 5 Target Story: Narrative 1. ...................................... 298
Study 5 Target Story: Narrative 2. ...................................... 299
Study 5 Target Story: Narrative 3. ...................................... 300
Study 5 Target Story: Narrative 4. ...................................... 301
Study 5 Control Story: Introduction to Characters. ............. 302
Study 5 Control Story: Narrative 1. .................................... 303

xiv

Figure 45. APPENDIX E: Study 5 Control Story: Narrative 2. .................................... 304
Figure 46. APPENDIX E: Study 5 Control Story: Narrative 3. .................................... 305

Figure 47. APPENDIX E: Study 5 Target Story: Narrative 4.. ..................................... 306

XV

CHAPTER 1

INTRODUCTION

1 .0 Introduction

The goal of this dissertation is to present experimental research that provides a
deeper understanding of the role of input on ﬁrst language acquisition. The process of
ﬁrst language acquisition is often represented in terms of the following equation: Innate
Component + Input = Ll. In the equation, the term ‘innate component’ refers to an innate
mechanism or language acquisition device (Language Acquisition Device (LAD),
Chomsky 1986) that allows humans to acquire language and not other species. Chomsky
(1986) deﬁnes two aspects to language: I-Language and E-Language. I-Language is the
mentally represented linguistic knowledge that a native speaker of a language has
(competence) and E-Language refers to the external observable behavior of the speaker
(performance). E-Language is inﬂuenced by extra-linguistic factors (e. g. style, social
status, gender) and the constraints that govern production and comprehension. In the
above equation, the term ‘input’ refers to the E-Language of the adult speakers in the
child’s speech community while the term ‘Ll ’ refers to the child’s target I-Language (or
grammar). The equation can thus be rewritten as: Innate Component + Adult E-
Language(s)l = Child I-Language. While I assume that the ‘innate component’ is
generally invariable across typically-developing human populations (all typically-
developing humans have the ability to acquire language), we know that the ‘input’ (i.e.
the speech of the speakers with whom the child interacts, Adult E-languages) varies

within and across speakers to the extent that no two humans will ever be exposed to

 

1 ‘Adult E-Language’ may not be the correct term given that it is most likely the case that
children receive linguistic input not only from adults but also from other children.

identical linguistic input, which implies that there will never be two identical ‘Lls’ but

instead extensive overlap between ‘Lls’ to allow for communication between

individuals. For this reason, the term ‘L l ’ in the equation refers to the unique l-Language

of each individual learner. This process of language acquisition can be represented as in

Figure l, which was modiﬁed and adapted from Yang (2001).

Figure l. The Dynamics of Language Acquisition (adapted from Yang 2001).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Adult #1 Adult #2 Adult #3 Adult #4
I-language I-language l-language l-language
Adult #1 Adult #2 Adult #3 Adult #4
E-language E-language E-Ianguage E-language

Child #1 Child #2 Child #3
I-language r l-language I-language
Child #1 Child #2 Child #3
E-language E-language E-language

 

 

 

 

 

 

 

 

 

 

The dotted lines on either side of the Adult boxes represent the idea that the

number of speakers with whom a child interacts will vary but is most likely more than 4

people (e.g. school/daycare community, neighbors, relatives, parents, church community,

 

pediatrician, family ﬁ'iends, etc.). The dotted lines extending from one Child’s E-
Language to another Child’s I-Language represent the possibility that children rely on the
input or E-Language, not only of adults, but also of other child learners when
constructing their own Lls. Figure 1 represents the following ideas: (1) that children must
rely on the E-Language of other speakers to construct their own I-Language (2) that the
E-Language that speakers provide for children does not always match the speakers’ own
I-Language and (3) since the linguistic experiences of each child differ, it holds that no
two I-Languages will ever be identical, neither between two children nor between any
adult and child (Yang 2001).

It should be noted that Figure l is an oversimpliﬁcation of the actual complex
nature of the input in several ways including that (1) it does not include any information
about whether children are selective in which speakers they rely on for input information
but rather assumes that they weigh all input the same (i.e., do learners rely equally on
adult vs. child input, kinship vs. non-kinship input, male vs. female input for L1
construction?), (2) the diagram is too general to represent the fact that the E-Language of
speakers is variable. In other words, in Figure l, the fact that Child #1 and Child #3 both
receive input from Adult #3 does not guarantee that the input to both children from Adult
#3 will be identical in terms of certain linguistic features or forms nor does it guarantee
that Child #1 and Child #3 will both equally treat the input of Adult #3 as reliable for
constructing their own grammar. Moreover, given linguistic variation, it is very likely
that two different utterances from Adult #3 to Child #2 could differ with respect to
certain grammatical features, depending on the social contexts in which the two

utterances were produced. This dissertation is primarily concerned with the idea that the

input of the adult speaker (Adult E-Language) is sometimes variable with respect to
certain grammatical forms, with certain morphological forms sometimes present in an
utterance and other times absent in subsequent utterances, depending on a variety of
extra-linguistic factors. It should also be noted that while Figure 1 shows a representation
for both the I-language and the E—language of each adult and child speaker, experimental
tasks on production and comprehension of plural morphology can only provide evidence
about the speaker’s E-language (performance). Based on the ﬁndings for the children’s
E-language, I will attempt to make inferences about the development of their I-language
at the time of testing.

Given the very complex nature of the linguistic input, as the discussion above
indicates, we start very small by examining only the acquisition of the plural morpheme
associated to an interpretation of ‘more than one’ in two varieties of Spanish. The ﬁrst
variety is Chilean Spanish, where the plural morpheme is affected by a phonological
process of lenition that affects all syllable-ﬁnal /s/. Irnportantly, this lenition process is
(1) not categorical but constrained by linguistic and extra-linguistic factors in the adult
speech and (2) creates a fair amount of omissions. As a result, Chilean adult speakers
sometimes produce the plural morpheme (as [s] or [h]) and sometimes omit it (zero form)
when referring to plural sets, creating an ambiguous input with respect to plural
morphology, in other words, with respect to whether plural morphology must be
represented or not. The second variety is Mexican Spanish (Mexico City) where no such
lenition process occurs and instead there is a categorical process of assimilation. The
plural morpheme is pronounced as [s] after all vowels, voiceless consonants and pauses

and as [2] after all voiced consonants. Unlike Chilean Spanish, in Mexican Spanish the

plural morpheme is systematically produced on all elements within the noun phrase in the
adult speech. Plural morpheme production in Chilean and Mexican Spanish is discussed
in more detail in Chapter 2 of this dissertation.

Although language acquisition research has taken input into consideration, very
few studies have examined the effect of variable input on language acquisition and those
that do exist have examined the acquisition of phonological variable rules by children,
but the acquisition of grammatical features that are affected by variable rules in the adult
speech has not received much attention. In a recent paper, however, Wilson and Henry
(1998) argue that any theory of language acquisition must account for the fact that the
input into the emerging linguistic system is variable, even within a monolingual context,
and that a key part of the language acquisition device (LAD) is designed to enable it to
cope with this variability. With this in mind, Figure 1 can be rewritten as Figure 2 to

represent the ambiguous input with respect to plural morphology in Chilean Spanish.

Figure 2. Linguistic Input for Plural Morphology in Chilean Spanish.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Adult #1 Adult #2 Adult #3 Adult #4
l-language I-language l-language _ I-language
PL+ PL+ PL+ PL+
Adult #1 Adult #2 Adult #3 Adult #4
E-language E-language E-language E-language

 

 

 

 

 

 

 

 

 

 

 

 

 

‘. 7. -’ //
\\\>.\ \l/(Z'
\ /

 

Child #1

I-language
????

 

 

 

 

Focusing now on just one child. Figure 2 shows that while the various adult I-languages
have an underlying representation of plural (represented by PL+), the adults sometimes
produce (+) and sometimes omit (-) the plural morpheme in their own speech or E-
Language when referring to plural sets. The child only has access to the speakers’ E-
Language when constructing his own l-Language. The underlying question of this
dissertation then is what type of l-Language will the Chilean child construct given the
variable and ambiguous nature of the input he is presented with, one with plural
morphology or one without plural morphology? In Figure 2, the ‘?’ sign in the child’s 1-

Language box represents this underlying question. On the other hand, the input for

 

Mexican children (Mexico City) is consistent in the sense that adults always produce the

plural morpheme when referring to plural sets. This can be represented as in Figure 3.

Figure 3. Linguistic Input for Plural Morphology in Mexican Spanish.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Adult #1 Adult #2 Adult #3 Adult #4
I-language l-language l-language I-language
PL+ PL+ PL+ PL+
Adult #1 Adult #2 Adult #3 Adult #4
E-language E-language E-language E-language

r +/

//
\5.\.l/:‘//f

Child #1

1- language
????

\\
\7‘

 

 

 

 

 

 

 

Given the differences between the input to Mexican and Chilean children, this
dissertation asks a related question to the one posed above: Will development of plural
morphology differ between Chilean (variable input) vs. Mexican children (consistent
input), and if so, how will it differ? Comparing two varieties of Spanish, rather than two
completely different languages, provides an ideal way of examining one grammatical

form while holding other forms in the two varieties constant.

It may be the case that variable and ambiguous input will cause Chilean children
to initially construct a grammar (I-Language) that is different from their parents’
grammar and also from the grammar of Mexican children. It has been proposed in the
language change literature that the linguistic input that adults provide for children (E-
language) does not always reﬂect the adult grammar (I-language) and this may result in
language change, as children construct a grammar (I-language) that is slightly different
from their parents’ grammar (Yang 2001 ). On the other hand, it may be the case that the
child acquires the variable rules early on and as a result the variable input has no effect
on the acquisition of the grammatical features under consideration. Instead, as long as
plural morphology is overtly present in the input some of the time, the Chilean child will
develop plural morphology in the same way as the Mexican child. The ﬁrst part of
Chapter 1 will address in more detail the underlying question of this dissertation and the

second part will discuss the theoretical and applied implications of this research.

1.1 Statement of Problem
The underlying question of this dissertation is how does variability in the input

affect child language acquisition of grammatical morphology when the variability
includes omission of the morpheme and is constrained not only by linguistic
(phonological, grammatical) but also extra-linguistic (socioeconomic status (SES), age,
sex) factors. Speciﬁcally, as noted above, this dissertation examines the development of
plural morphology in two varieties of Spanish. The ﬁrst variety is Chilean Spanish, where
the plural morpheme is affected by a phonological process of lenition that affects all
syllable-ﬁnal /s/. In Chilean Spanish the plural can surface as [s], [h] or it has a zero

form. For example the word gatos ‘cats.PL’ can surface as [gatos], [gatoh], or [gato]. The

third variant [gato] is identical in form to the singular gato ‘cat.SG’ ([gato]), which
creates an ambiguous input. The second variety is Mexican Spanish (Mexico City) where
the plural morpheme is systematically produced as [s] or [z] in all plural noun phrases in
the adult speech (6. g. [gatos]). The variant [5] surfaces before vowels, voiceless
consonants and pauses while [2] surfaces before voiced consonants.

In response to the question above, it is reasonable to think that as long as the
morpheme is available in the adult speech some of the time, children would acquire the
morpheme and its grammatical features early on and even use it systematically, or at least
more systematically than the adult. In fact, research by Singleton and Newport (2004)
and Hudson Karn and Newport (2005) lend support to this idea by showing that children
are able to acquire inconsistent forms in the input and, in addition, they regularize such
forms in their own production Singleton and Newport (2004) examined the development
of morphology by a signing child (Simon) whose only linguistic input was that of his
non-native signing parents who demonstrated inconsistency and errors in their own use of
certain grammatical forms. Nonetheless, by age 7, Simon’s production of these
grammatical forms was much more regular and consistent than his parents and was even
at the level of other signing children of the same age whose parents were native signers.
However, the input that children were exposed to in the Singleton and Newport (2004)
and the Hudson Kam and Newport (2005) studies is different from the input that Chilean
children receive with respect to the plural morpheme in that their children were exposed
to inconsistent input, that is neither linguistically nor extra-linguistically predictable. We
ﬁnd this type of input coming ﬁom the speech of adult non-native speakers to their

children, as is the case for Simon. These studies suggest that when a particular form is

inconsistently produced in the input, in other words, the form is not linguistically
constrained and hence there are no patterns to be learned with respect to this form,
children will attempt to regularize the form by either producing it systematically or never
producing it at all in their E-language, which suggests that they have regularized the
feature in their I-language. Both systematic use and systematic non-use are forms of
regularization. However, what about the case of Chilean Spanish plural morphology
where we ﬁnd that the distribution of the plural morpheme is variable, in other words, it
is linguistically and extra-linguistically constrained. Will Chilean children behave like
Simon and regularize the variable input?

It seems unlikely that children will treat variable input in the same way as
inconsistent input. Studies have shown that when the input is variable, 4-7 year old
children can acquire at least some of the variable rules, making child production of the
particular form under consideration variable as well (Kovac and Adarnsonv 1981, Labov
1989, Roberts 1994, 1997, Smith et al. 2006). In other words, children are able to detect
patterns in the input and act upon these patterns in their acquisition of the variable form.
However, these studies also suggest that linguistic variability is learned by children
before extra-linguistic variability (i.e. variation based on SES, age, gender) (Kovac and
Adarnson 1981, Roberts 1994, 1997, Smith et al. 2006). For example, Roberts (1994,
1997) found that, unlike adults, neither the addressee of the child nor the formality of the
speech task affected child use of variable forms while grammatical and phonological
constraints did. For this reason, child production patterns, although similar, were not
identical to adult production patterns. This suggests that as long as production of plural

morphology by adult Chilean speakers is linguistically constrained in the input, Chilean

10

children will be able to detect patterns and reproduce some of these patterns in their own
speech. Therefore, we would predict that presented with variable input, children would
not regularize the input, (always produce the form or never produce the form) but rather
production of the form would be constrained by variable rules that the child has decoded
from the adult input, at least variable rules that are linguistically constrained. For this
reason, a second answer to the question posed above would be that Chilean child
production of plural morphology would show only variable behavior. However, even if
Chilean children show variability in their production of plural morphology, this would
not necessarily mean that the child had linked the plural morpheme to an underlying
representation for number. In other words, it does not guarantee that the child’s 1-
language has a representation for the plural morpheme. Dominican Spanish also has
syllable ﬁnal /s/ lenition and recent research on Dominican Spanish-speaking children
suggests that children may not have a representation for the plural morpheme, which may
be due to the high level of plural morpheme omission in adult speech (Pérez-Leroux
2005). This study will be discussed in more detail in Chapter 3 of this dissertation.

It appears that matters are a bit more complicated as we have not yet taken into
consideration the comprehension of plural morphology. It may be the case that
comprehension of the plural morpheme and production of the plural morpheme occur at
different stages in the child’s development. If this were the case, describing the E-
language and I-language of the child becomes more complicated. In order to understand
how the development of the comprehension of plural morphology might unfold in
children, we need ﬁrst to discuss in a bit more detail the distribution of the plural variants

in order to understand the leamability problem facing the child. Returning to the above

11

description of the variable behavior of plural morphology in Chilean Spanish we ﬁnd that
there are two forms of the plural morpheme in all varieties of Spanish, [s] and [es].
Spanish words that end with a vowel have [s] as the plural mOrpheme and Spanish words
that end in a consonant have [es]. Given that most Spanish words end in a vowel, [s] is
the most common form. In Chilean Spanish, the [s] in both of these forms undergoes a
process of lenition and can be pronounced as [h] or can be omitted This means that in
Chilean Spanish the plural morpheme can surface in adult speech as [s], [h], [zero form]
(omission), [es], [eh] and [e].2 If we compare the distribution of the Chilean variants of
the plural morpheme to the distribution of the English allomorphs of the plural
morpheme, the complexity facing Chilean children becomes a bit clearer. In English the
plural allomorphs, [s], [2], [oz], are in complementary distribution and always occur in
certain phonological contexts. The allomorph [s] always occurs after voiceless
consonants (pets [pats], cats [kats], paths [pzeﬂs]), [2] after a vowel or voiced consonant
(kids [kIdz], dogs [dagz], bees [biz]), and [oz], after sibilants (kisses [klsaz], ﬁshes
[fI]oz]). However, this is not the case for Chilean plural morphology where the
distribution of the various forms ([8], [h], zero form) are not in complementary
distribution and, in addition, are determined by several interacting factors, including
phonological environment, syntactic position, and social factors, such as the formality of
the speech act or age and gender of the speaker’. In other words, the Chilean Spanish

word gatos (‘cats’) could surface as [gatos], [gatoh] or [gato], depending on a variety of

 

2 Previous research on Chilean Spanish discusses only the [s] or [h] variants of the plural
morpheme; however, the data that I collected for this dissertation also shows evidence for
at least one other variant, an unreleased glottal stop.

3 The constraints governing syllable ﬁnal /s/ lenition in Chilean Spanish will be discussed
in more detail in Chapter 2.

12

interacting linguistic and extra-linguistic constraints. For example, the plural morpheme
might have a zero form more often in direct object sentence ﬁnal position than in subject
position or it may surface more as [s] when the speaker is addressing a middle class
stranger vs. a middle-class friend. The plural may be produced as [h] more often on
deﬁnite determiners than on indeﬁnite determiners, depending on the speech style of the
speaker at the time of the utterance. Ultimately, both the English and Chilean learner
have the same learning task of relating the various phonological surface forms of the
plural morpheme found in the adult E-Language to each other and also to the same
underlying syntactic position when constructing their own I-Language. It appears that the
constraints governing the distribution of the plural morpheme in Chilean Spanish are
more numerous and complex than those governing the distribution of the English
allomorphs and, for this reason, we might predict that English children would acquire the
English constraints before Chilean children acquire the Chilean constraints. Mexican
plural morphology is similar to English plural morphology in that the distribution of the
various plural forms ([3], [2], [es]) is categorical and not variable and there is not process
of lenition that results in a zero form.

If we start by ﬁrst looking at research that has dealt solely with the acquisition of
plural variants that are categorical and in complementary distribution, as in English and
Mexican City Spanish, we ﬁnd evidence that 4-year—old children show productive use of
some of the allomorphs but not all of them (Berko 1958, Perez-Perreira 1989).
Speciﬁcally, of the two Spanish variants [s] and [es], Mexican 4-year-old children use [5]
productively but not [es]. Instead, when presented with a novel word in a Berko (195 8)

style task that required the [es] variant of the plural morpheme, Mexican children were

13

found to either repeat the word without a plural morpheme or use the [s] variant. In the
same way, English-speaking children were found to use the [s] and [z] variants
productively but not [oz]. With respect to the comprehension of the English allomorphs, a
recent study has noted a trend (although it did not reach signiﬁcance) that 24 month old
English-speaking children perform better in comprehension tasks on the plural allomorph
[s] than on the allomorphs [z] or [oz] (Kouider et al. 2006). This suggests that acquisition
of plural allomorphs that are categorical and in complementary distribution, may not
occur all at once, but rather some variants may be acquired before others.

Even in languages like English and Mexican Spanish, where the plural variants
are in complementary distribution and are categorical in nature, it appears that the plural
forms are not all acquired at once. Hence, we might predict that plural forms that are
variable, not in complementary distribution and also that include a zero form for the
plural morpheme, as in Chilean Spanish, will likewise not be acquired all at once. In
addition, if it is true that the constraints that govern variable rules are acquired
throughout development, rather than all at once, with some linguistic constraints being
learned before others, and social constraints perhaps being acquired last, as research by
Kovac and Adamson (1981), Labov (1989), Roberts (1994, 1997), Smith et al. (2006)
have indicated, then we would expect that within a certain age ﬂame Chilean child
behavior with respect to the acquisition of plural morphology may vary between children
based on their linguistic and extra-linguistic experiences (i.e. the amount of variable input
and zero forms of the plural morpheme they are exposed to). While production of the
plural forms by children may appear variable, certain grammatical or sociolinguistic

factors may not affect the variability in children as they do in adults. Furthermore, if

14

Chilean children have not acquired all of the linguistic and extra-linguistic constraints
governing the variability of the plural morpheme in the adult speech, then they may not
know that the various forms of the plural morpheme are related to the same underlying
grammatical feature, in other words, that they are related to the plural. For this reason, it
seems reasonable to predict that while Chilean children may appear variable in their
production of plural morphology (although variability may not pattern I 00% with
adults), they may be systematic in their comprehension of plural morphology, always
treating the plural morpheme (either [s] or [h ]) as plural or always treating it as
singular, depending on their linguistic experiences at the time of testing.

The above prediction suggests that production of plural morphology could
precede comprehension of plural morphology in Chilean children. Given that we predict
that Chilean children will reproduce adult patterns in their production, yet may not
associate the various forms of the plural morpheme with an underlying representation of
plural ﬁ'om the very beginning, it is possible that Chilean children will produce the plural
variants in their own speech (as either or both [3] and [h]) but not associate them to an
underlying representation for plural and, hence, an interpretation of ‘more than one’ in
comprehension tasks. Intuitively, we might think that as long as the child produces
certain morphological forms, they have already associated these forms to an
interpretation of ‘more than one’. However, research has found that in English, where the
plural morpheme has no zero form and the variants are in complementary distribution,
children appear to produce plural morphology before associating it to the meaning of
‘more than one’ in comprehension tasks. English-speaking children start producing the

plural morpheme at about l;10 years of age (Ferenz and Prasada 2002), yet Kouider et al.

15

(2006) found that English-speaking children do not associate the plural morpheme to an
interpretation of ‘more than one’ until about 3 years of age. In addition, Diaz Campos
(2005) has found that children acquire variable phonological forms on a word-by-word
basis, which suggests that variable forms will be present in child speech very early in
language acquisition. This does not mean, however, that children will associate the
variable forms with each other or with any underlying representation from the very
beginning.

Given the underlying question of this dissertation the following predictions are
made for 4—7 year old Mexican and Chilean Spanish-speaking children: (1) production of
the plural morpheme will be variable in Chilean Spanish child language but not in
Mexican Spanish child language, (2) the variability found in Chilean child production
will be similar to adult production but not identical, (3) in comprehension Chilean
children will be consistent in their comprehension of plural morphology, either always
associating or never associating the plural morpheme to an interpretation of ‘more than
one’; Mexican children will always associate the plural morpheme to an interpretation of
‘more than one’ (4) comprehension of plural morphology will vary between Chilean
children based on their linguistic and extra-linguistic experiences, and (5) production of
the plural morpheme will precede comprehension of plural morphology in both Mexican
and Chilean Spanish.

In this dissertation I propose the Variability Delay Hypothesis, shown in (l),
which hypothesizes that, all other things being equal, a grammatical morpheme, which is
subject to variable rules and is ambiguous (i.e. sometimes has a zero form) in the adult

speech, will be acquired later than a grammatical form that is consistently produced in the

16

adult speech. This hypothesis is adapted from Yang’s (2000, 2002) Variation Model of
language acquisition, which proposes that the cumulative effect of the input combined
with a theory of a restricted search space can explain language acquisition. According to
Yang, children make hypotheses within the limits of UG that are punished or rewarded
depending on their ability to account for particular properties of the input. If the input is
unambiguous and frequent, acquisition happens early. If input is ambiguous, the child

may take longer to set a parameter.

(1) Variabiligz Delay Hypothesis (based on Yang 2000, 2002): Variability in
the input will delay child comprehension of grammatical morphemes when the variability
causes an ambiguity in the input (involves a zero form") and is constrained not only by

linguistic (phonological, grammatical) but also extra-linguistic (SES, age, sex) factors.

While there has been almost no research addressing whether variable input
involving zero forms aﬁects comprehension of grammatical morphemes, the Variability
Delay Hypothesis is partially supported by a recent study by Johnson (2005) who found
that AAVE-speaking children lagged behind their MAE-speaking counterparts in their
ability to indirectly assign quantity to subject noun phrases based on the presence or
absence of the English 3rd person singular marker in sentences like “The cats_sleep” vs.

“The cat__sleeps”, where the nominal plural morpheme is disguised by the initial [5] on

 

4 Syllable-ﬁnal /s/ lenition is generally discussed in the literature in terms of omission or
production of the plural morpheme. However, I have been describing omission of the
plural morpheme in terms of a “zero form” under the reasoning that if children have not
yet acquired plural morphology, they would not associate the absence of the plural
morpheme with an omission.

l7

the verb. Johnson (2005) suggested that difficulty for AAVE children may have been due
to the fact that AAVE adult speakers tend to omit the 3rd person singular marker in their
own speech. In other words, the AAVE children receive an input where the 3rd person
singular marker is ambiguous, it is sometimes present and sometimes absent in the input,

while the MAE children receive consistent evidence for the 3rd person singular marker.

1.2 Theoretical Importance

While we might assume that the LAD is designed to cope with variability in the
input, we do not have a clear understanding of how it deals with variable input (i.e. does
it regularize variable input or not?) nor how different levels or types of variability affect
language acquisition. Are certain constraints governing variability more difﬁcult to
acquire than others? Is variability affecting grammatical features more difficult to acquire
than other forms of variability? Will variability cause any delay in the development of
grammatical features? While research has suggested that frequency of forms is important
in language acquisition, it may be that ﬁequency is not so important if the form is
unreliable (i.e. ambiguous) or is variable in the input. In other words, it seems reasonable
that a child would learn a less frequent but more reliable form before leaming a more
frequent but less reliable form. For example, it may be that, while the plural morpheme is
more frequent in Chilean Spanish than in English, because it is marked on nouns,
determiners, and adjectives in Spanish, it may be acquired earlier in English, because it is
not variable nor omitted in English, rather it is systematically produced on all plural
nouns. The research presented in this dissertation will provide a clearer understanding of
how both the properties of the input and the frequency of those properties affect language

development.

18

Finally, while there are several studies on the production and comprehension of
plural morphology in English, there are very few studies that have examined the
production of plural morphology in Spanish-speaking children and, as far as we know,
there are no studies that have examined comprehension of nominal plural morphology in
Spanish-speaking children. In English the plural morpheme is only produced on the noun,
while in Spanish the plural morpheme occurs on all elements within the noun phrase.
Examining plural morphology production and comprehension in Mexican Spanish-
speaking children will provide much needed data on the development of plural
morphology in a language where the plural morpheme is consistently produced on all
elements within the noun phrase. Examining comprehension and production of plural
morphology in Chilean Spanish-speaking children will reveal how variable input affects

the acquisition of nominal plural morphology.

1.3 Applied lrnportance

The results of the research presented in this dissertation will have important social
impact on matters concerning education and language testing in children from socially
disadvantaged groups since many tests of early children’s language abilities do not
adequately take social variation nor its effects on learning into consideration. If
differences in the input affect child comprehension and production of plural morphology,
it will also affect their performance on standardized tests evaluating plural usage. While
several varieties of Spanish have syllable ﬁnal /s/ lenition, many standardized tests
measuring language competence in Spanish-speaking children fail to take into
consideration this variation and instead children are tested on both their production and

comprehension of the plural morpheme. In Chile some of these tests include: el Test de

19

Desarrollo Psicomotor: 2-5 Aﬁos (Psychomotor Development Test for 2-5 year olds)
(Haeussler and Marchant 1988), el Test para la comprensio'n auditiva del lenguaje de E.
Carrow: Aplicacio’n en Chile (Auditory Language Comprehension Test of E. Carrow:
Application in Chile (Pavez 2004) and el Test Exploratorio de Grdmatica Espaﬁola de A.
Toronto: Aplicacio’n en Chile (Exploratory Spanish Grammar Test of A. Toronto:
Application in Chile) (Pavez 2003). Since 1997 Chile has opened a variety of Escuelas
Especiales de Lenguaje (EEL) (Special Language Schools) in order to provide special
services and specialized education to children with Speciﬁc Language Impairment (SLI).
These schools are funded by the Chilean government and are invaluable to the children
and families that they serve. Children must be diagnosed as SLI before they can be
admitted to these schools. Unfortunately, some of the standardized tests designated by the
Chilean Ministry of Education (Decreto #1300 del Ministerio de Educacio'n) for
diagnosing children with Speciﬁc Language Impairment (SLI) include test items that
examine child comprehension and production of plural morphology. Obviously, if
children are acquiring a variety of Spanish where the plural morpheme is variable and
sometimes has a zero form in the adult speech, we might expect similar patterns in child
production. Moreover, if variability turns out to affect child comprehension of plural
morphology, then testing Chilean children on their comprehension of plural morphology
is inappropriate and could unintentionally discriminate against certain groups of children.
With respect to this, it is important to note that Cepeda (1995) found working-class
Chilean adult speakers omitting syllable ﬁnal /s/ more often than middle-class speakers,

which suggests that, if plural morpheme omission affects child production and

20

comprehension of plural morphology, then Chilean working-class children may be the
most likely to suffer the consequences of inappropriate testing methods.

Within the United States, research on Speciﬁc Language Impairment (SLI) in
US. Spanish-speaking children found that they omitted the plural morpheme in
elicitation tasks and the study suggested that omission of the plural morpheme may be a
useful diagnostic for uncovering SLI in Spanish-speaking children (Bedore and Leonard
2001). Unfortunately, this research only examined Mexican Spanish-speaking children
and failed to take into account the fact that in the majority of Spanish dialects (e. g. Chile,
Cuba, Dominican Republic) the plural morpheme is oﬁen omitted in the speech of adult
speakers. As a result, this creates the potential for misdiagnosing typically developing
Spanish-speaking children as language disordered. In this sense the work relates directly
to that of Seymour et al. (2004) and de Villiers et al. (2004) on standardized tests for

AAVE-speaking children.

1.4 Organization of the Chapters

This dissertation is organized as follows: Chapter 2 provides an account of the
syntax and semantics of Spanish plural morphology and an outline of syllable-fmal /s/
lenition in Chilean Spanish. Chapter 3 discusses previous research on the acquisition of
plural morphology in Spanish and English both in areas of comprehension and
production. In addition, I discuss recent research on the acquisition of inconsistent input
and variable input and discuss the ﬁndings of these studies in light of the research
presented in this dissertation. Chapter 4 presents the production experiments that were
carried out with Chilean and Mexican children and adults. In this chapter I present results

from a Free Speech Task, Naming Task and Repetition Task and discuss whether and to

21

what degree child production matches adult production in Mexican and Chilean Spanish
and also the degree to which Chilean production matches Mexican production. Chapter 5
presents the comprehension experiments that were carried out with Chilean and Mexican
children and adults. I compare Mexican vs. Chilean child comprehension and production
of the plural morpheme in indeﬁnite and deﬁnite noun phrases and also present ﬁndings
on Chilean child comprehension and production of bare plural and bare singular noun
phrases. Chapter 6 concludes with a summary of the results of the experiments presented
in Chapters 4 and 5 and a discussion of the implications of such ﬁndings for the study of

the effect of variable input on ﬁrst language acquisition.

22

CHAPTER 2

A LINGUISTIC DESCRIPTION OF PLURAL MORPHOLOGY

2.0 Introduction

The experimental studies presented in this dissertation look at children’s ability to
associate the plural morpheme to an interpretation of ‘more than one’. The ability to
associate the plural morpheme to other properties, such as distributivity, was not tested.
For this reason, the discussion in this chapter of the syntactic and semantic properties of
the plural morpheme will focus on how they are related to the interpretation of ‘more
than one’.

Any study of the acquisition of morphology must be concerned not only with
which morpho-phonological forms are associated to a particular morpheme but also with
the syntactic and semantic properties of the morpheme. In other words, we need to
establish what plural ([+PL] is semantically and how the presence or absence of this
feature in the syntax is interpreted. Furthermore, in languages like Spanish, where
singular and plural morphology show up in determiners, nouns and adjectives, we need to
address whether one of these elements is the realization of an interpretable feature or
whether all the realizations of plural vs. singular morphology are the result of agreement.
In this chapter, I will ignore most of the debates on both the semantics of plurals in
general and the syntax of agreement. I will concentrate instead on providing a general
picture of number inflectional morphology.

I will make the following main claims: (1) While singular has no overt morpho-
phonology, Spanish plural morphology can be realized morpho-phonologically in

different ways in different dialects because /s/ undergoes two main types of processes

23

across dialects: a process of assimilation and a process of lenition. The process of lenition
will cause a certain amount of ambiguity for listeners. (2) Semantically, I will follow
Ojeda (1998) in saying that plurality in Spanish is represented as [+PL] and that plural
nouns only include in their denotation plural sets but not singleton sets. (3) Syntactically,
I will claim that Spanish nominal plural morphology is always a manifestation of
agreement with an interpretable head. No overt manifestation of plural morphology is the
spell-out of a semantically interpretable feature. Instead, all plural morphology is a
manifestation of agreement with a syntactic functional head for number that takes a DP as
its sister. (4) I will assume that bare singular count nouns are neither singular nor plural
because they lack this functional projection for number.

This chapter is organized as follows: Section 2.1 will discuss Spanish number
marking in general, outlining the syntactic categories that are marked with plural
morphology. Section 2.2 will present the phonological forms of the Spanish plural
morpheme in Mexican Spanish vs. Chilean Spanish and will also provide a background
on some recent research on syllable-ﬁnal /s/ lenition in other varieties of Spanish. Section
2.3 will present the syntactic analysis of Spanish plural morphology that is being adopted
in this dissertation. Section 2.4 will provide the semantic analysis of Spanish plurality
that I will be assuming. Finally, Section 2.5 will discuss the implications of the semantic
and syntactic analyses for the interpretation assigned to the plural morpheme by Spanish

speakers.

24

2.1 Spanish Plural Morphology
The Spanish plural morpheme occurs on all elements within the determiner
phrase: the determiner, the noun and all adjectives. Examples are shown in (l) and (2)

below.

(1) a. Las vacas perdidas nunca regresaron a casa.
The.F.PL cows.F.PL lost.F.PL never came.3.PL home

‘The lost cows never came home.’

b. La vaca perdida nunca regreso a casa.
The.F.SG cow.F.SG lost.F.SG never carne.3.SG home

‘The lost cow never returned home.’

(2) a. Los toros perdidos nunca regresaron a casa.
The.M.PL bulls.M.PL lost.M.PL never came.3.PL home

‘The lost bulls never came home.’
b. El toro perdido nunca regreso a casa.
The.M.SG bull.M.SG lost.M.SG never came.3.SG home

‘The lost bull never came home.’

The examples in (l) and (2) show that the determiner, noun and adjective all agree in

number and gender. (1a) and (2a) illustrate that all of the elements within the determiner

25

phrase ‘the lost cows’ are marked with a plural morpheme unlike English, which mainly
exhibits plural morphology only on the noun. The Spanish plural morpheme is
phonetically realized in different ways depending on the variety of Spanish under
investigation (see Section 2.2).

While the Spanish plural morpheme occurs on all elements within the determiner
phrase, it is not always the sole indicator of number in the linguistic input to children.
There are two other potential indicators of plurality: verbal morphology and the form of
the determiner itself.

As is well-known, Spanish subjects agree with verbs in number, as shown in (1)
and (2), where the singular (regreso’ ‘retumed.3.SG’) and plural (regresaron
‘returned.3.PL’) forms of the verb regresar (‘return.NONFrN’) agree with singular and
plural subjects, respectively. However, Spanish also allows, and in many cases prefers,
the use of null subjects. In such cases where the subject is null, only the verb supplies
number information about the nominal subject. This fact is important as it suggests that
verbal morphology may provide reliable information about nominal number (nouns in
subject position but not object position) for children. Examples of constructions involving

null subjects are shown in (3) and (4).
(3) Regresaron a casa.

Came.3.PL home

‘They came home.’

26

(4)

Regresé a casa.
Came.3.SG home

‘It came home.’

The example in (3) involves a verb in the 3rd person plural and the subject is interpreted

as plural. The example in (4) contains a verb in the 3rd person singular and the subject is

interpreted as singular.

While syllable-fmal /s/ lenition may affect nominal number marking and create

potential ambiguity in the input, syllable ﬁnal /s/ lenition almost never affects verbal

number marking, because verbal morphology is generally not represented phonologically

as /s/, and, when it does affect verbal morphology, it generally does not create any

number ambiguities in the input, as illustrated in (5) and (6).

(5)

(6)

La nifla esta saltando.
The.F.SG girl.1=.so is.3.so jumping

‘The girl is jumping.’

Las niﬁas estan saltando.

The.F.PL girl.F.PL are.3.PL jumping

‘The girls are jumping.’

27

Possible Pronunciations

[la] [niﬁal [esta]

[las], [lah], [la]
[niﬁas], [niflah], [niﬁa]

[estan]

The examples in (5) and (6) illustrate that the singular and plural forms of the subject
determiner and subject noun may overlap in varieties of Spanish that have syllable-ﬁnal
/s/ lenition (i.e. [la] and [niﬁa] can both be used in semantically plural and singular noun
phrases). On the other hand, the singular and plural forms of the verb do not overlap
because the 3rd person plural verbal morpheme occurs as /n/, not /s/. This suggests that
verbal morphology may be a more reliable indicator of number than nominal morphology
in varieties of Spanish with syllable ﬁnal /s/ lenition. While nominal plural morphology is
sometimes present and sometimes absent on semantically plural nouns or determiners,
verbal plural morphology will always be present when the nominal subject is plural,
which might affect what children exposed to such a dialect initially rely on when
assigning number to subject noun phrases. Table 1 provides an overview of Spanish

present tense verbal morphology.

Table l. Verbal Morphology: Chilean Spanish Present Tense.

 

Person/Number estar (‘to be’) Verbal Morpheme

1St singular estoy [estoi] -oy [oi]

1St plural estamos [estamos], [estamoh], [estamo] -mos [mos], [mob], [mo]
2“d singular estas [estas], [estah], [esta] -s [s], [h], zero

2Ind plural estan [estan] -n [n]

3rd singular esta [esta] -zero

3rd plural estan [estan] -n [n]

 

 

28

Table 1 shows the various verbal morphemes and possible pronunciations in Chilean
Spanish. The data illustrate that, for the most part, syllable ﬁnal /s/ lenition does not
create number ambiguity in verbal morphology. The only pessible ambiguity would be
associated to person and would occur between 2nd person singular vs. 3rd person singular
where there is overlapping of forms. For this reason, it appears that in Chilean Spanish
verbal number morphology is more reliable than nominal number morphology.

The second indicator of number is the form of the determiner. While Spanish
plural determiners carry a plural morpheme, the form of certain determiners guarantees a
‘more than one’ interpretation regardless of whether the listener takes into consideration
the plural morpheme or not. This, of course, is the case for quantiﬁers such as todos
‘all.PL/every.PL’, for example, and some determiners even have different forms in the
singular vs. plural in addition to the difference associated with the presence or absence of
the plural morpheme. For example, singular masculine determiners are often different in

form from plural masculine determiners. This is illustrated in Table 2.

29

Table 2. Spanish Determiners.

 

Singular Plural Gender
Deﬁnite e1 los ' Masculine
la las Feminine
Indeﬁnite un unos Masculine
una unas Feminine
Indeﬁnite algt'm algunos Masculine
alguna algunas Feminine

 

Note in Table 2 that the only difference between the plural and singular feminine
determiners is the presence vs. the absence of the plural morpheme. However, the forms
of the masculine determiners are different when plural vs. singular.

While plural masculine determiners are generally different in form from singular
masculine determiners, there are cases where plural and singular masculine determiners
only differ by the presence or absence of the plural morpheme. One such case is in

constructions involving noun drop, which is illustrated in (7).

(7) 3. Un monito pequeﬁo esta encima de la mesa.

A/one.M.SG monkey.M.SG small.M.SG is.3.SG on top of the table

‘A/one small monkey is on the table.’

30

b. Uno grande esta encima de la cama.
A/one.M.SG big.M.SG is.3.SG on top of the bed

‘A big one is on top of the bed.’

c. Unos grandes estan encima de la casa.
Some.M.PL big.M.PL are.3.PL on top of the house

‘Some big ones are on top of the house.’

The examples in (7) demonstrate that when the noun is dropped, the indeﬁnite plural and
singular masculine determiners only differ by the presence or absence of the plural
morpheme. The data presented in (7) are important for the present dissertation because
they demonstrate that the masculine plural vs. singular indeﬁnite determiners, in addition
to feminine plural vs. singular indeﬁnite determiners, sometimes differ only by the
presence or absence of the plural morpheme. This means that Spanish-speaking children
must be able to associate the plural morpheme to an interpretation of ‘more than one’ in
order to correctly comprehend constructions like those presented in (7) as plural.

However, note the examples of deﬁnite noun phrases with noun drop presented in (8).
(8) a. El elefante gran—dc esta encima de la mesa.

The.M.SG elephant.M.SG big.M.SG is.3.SG on top of the table

‘The big elephant is on the table.’

31

b. El pequeﬁo esta encima de la cama.
The.M.SG small.M.SG is.3.SG on top of the bed

‘The small one is on the bed.’

c. Los pegueﬁos estan encima de la casa.
The.M.PL small.M.PL are.3.PL on top of the house

‘The small ones are on the house.’

The examples in (8) illustrate that the forms of the deﬁnite plural vs. singular determiners
remain different, even when the noun is dropped. With respect to language acquisition,
while the Spanish-speaking child must rely on the presence vs. absence of the plural
morpheme at least some of the time to correctly assign number to masculine indeﬁnite
noun phrases, this is not the case for masculine deﬁnite noun phrases, which are different

in form when plural vs. singular.S

 

5 It should be noted that Spanish allows a neuter determiner lo ‘the.SG’ to occur before
adjectives in constructions like (i) and (ii). Hence, one might be tempted to think that
masculine plural vs. singular deﬁnite determiners at times differ only with respect to the
plural morpheme. However, this is not the case, as the plural counterparts of these
constructions do not exist, as (iii) illustrates.

(i) Lo irnportante es que llegaste.
The.SG important.SG is that arrived
‘The important thing is that you arrived.’

(ii) Lo bueno es que llegaste.
The.SG good.SG is that arrived
‘The good thing is that you arrived’

(iii) *Los importantes son que llegaste y que no te perdiste.
The.pl important are that arrived and that NEG REFL lost.2.SG.PAST
‘The important things are that you arrived and that you didn’t get lost.’

32

In Spanish direct and indirect object pronouns are also marked for number as

shown in (9)-(10).

(9) a. Lo/ la compre'.
It.M.so/it.r.so bought. 1 .SG.PAST

‘I bought it.’

b. Los/ las compré.

Them.M.PL/them.F.PL bought. 1 .SG.PAST

‘I bought them.’
(10) a. Le mande' una carta.
Her/him.NEUT.SG sent.1.SG.PAST a letter
‘I sent her/him a letter.’
b. Les mandé una carta.

Them.NEUT.PL sent. 1 .SG.PAST a letter

‘I sent them a letter.’

Unlike the case for deﬁnite masculine determiners, the only difference between the plural
and singular object pronouns is the presence or absence of the plural morpheme.
Spanish demonstratives are also marked with number. The form of the masculine

demonstrative is different for singular vs. plural, while the only difference between the

33

plural vs. singular feminine demonstratives is the presence or absence of the plural

morpheme. This is shown in Table 3.

Table 3. Number Marking on Spanish Demonstratives.

 

 

Singular Plural Gender
este ‘this.M.SG’ estos ‘these.M.PL’ Masculine
esta ‘this.F.SG’ estas ‘these.F.PL’ Feminine
ese ‘that.M.SG’ esos ‘those.M.PL’ Masculine
esa ‘that.F.SG’ esas ‘those.F.PL’ Feminine

 

Finally, Spanish allows both bare plural and bare singular count nouns in

complement position, as in (11), although the distribution of bare singular count nouns is

more restricted than for bare plural count nouns.

(l l) a. Tengo pelota.
I have ball.F.SG

‘I have a ball/some balls.’
b. Tengo pelotas.

I have balls.F.PL

‘I have balls.’

34

The translations for (11a) illustrate that bare singular count nouns are associated to both
an interpretation of ‘more than one’ and ‘one’ while (1 lb) shows that bare plurals are
only associated to only an interpretation of ‘more than one’.‘

While bare Singulars and bare plurals are only allowed in complement position,
the distribution of bare Singulars is much more restricted than for bare plurals. Bosque
(1996) observes that Spanish bare Singulars, in the dialects where they are found, are
allowed as verbal complements in certain contexts: as objects of intensional predicates, as
in (12), as objects under negation, as in (13), and in constructions that denote inherent

properties of a particular entity or where the object has a unique interpretation, as in (14).

(12) a. Ando buscando casa
Go.l.SG.PRES looking house

‘I am looking for a house.’

b. Quiero auto.
Want.l.SG.PRES car

‘I want a car.’

c. Necesito lépiz.

Need. l.SG.PRES pencil

‘I need a pencil’

35

(13) No tengo lapiz.
NEG have.1.SG.PRES pencil

‘I don’t have a pencil.’

(14) Llevaba chaqueta.
Wore. 1 .SG.PAST jacket

‘He was wearing a jacket.’

In our own ﬁeldwork, we found that Chilean Spanish also allows non-intensional verbs to

take bare singular objects, as in (15).

(15) a. Me compré auto.
REFL bought. 1 .SG.PAST car

‘I bought myself a car.’

b. Me consegui perro.
REFL got. 1 .SG.PAST dog

‘I got myself a dog.’

c. Hicimos muralla.

Made. 1 .PL.PAST wall

‘We put up a wall (ﬁre wall).’

36

(1. Me falta cuchillo.
REFL miss.3.so.r>ru=.s knife

‘I’m missing a knife.’

One noticeable characteristic linking together several of these predicates is possession.
The above predicates allowing bare singular objects all seem to denote some type of
possession, lack of possession or intention of possession (Miller and Schmitt 2004).

In summary, Spanish plural marking has the following characteristics: (1) plural
marking occurs on all elements within the determiner phrase, (2) Spanish has subject-
verb agreement and hence the verb also provides number information about nominal
subjects (3) plural masculine determiners generally differ in form ﬁom singular
masculine determiners, (4) plural feminine determiners only differ from singular
feminine determiners by the presence or absence of the plural morpheme, (5) Plural and
singular object pronouns only differ with respect to the presence or absence of the plural
morpheme, and (6) bare plural count nouns, which are associated to an interpretation of
‘more than one’, and bare singular count nouns, which are associated to both an
interpretation of ‘one’ and ‘more than one’, only differ by the presence or absence of the

plural morpheme.

2.2 The Phonological Form of the Spanish Plural

The Spanish plural morpheme is realized as -/s/ and -/es/. —/s/ occurs post-
vocalically and -/es/ occurs after consonants due to an epenthetic [e] which breaks up
post-vocalic consonantal clusters. Since most words in Spanish end in a vowel, -/s/ is the

most frequent form of the plural morpheme. Examples are provided in Table 4.

37

Table 4. The Spanish Plural Morpheme.

 

 

-/s/ -/es/
Singular casa, auto, elefante, lapiz, ﬂor, pan,
menu, kiwi pared, reloj, arbol
Plural casas, autos, elefantes, lapices, ﬂores, panes,
menus, kiwis paredes, relojes, arboles

 

 

There are at least two types of phonological processes that affect the realization of
syllable ﬁnal /s/ in Spanish: (i) a process of voicing assimilation and (ii) a process of
lenition. The former process occurs in Mexican Spanish (Mexico City) and the latter in
Chilean Spanish. In the following sections, the phonological form of the Spanish plural
morpheme is ﬁrst presented for Mexican Spanish (Mexico City) and then for Chilean

Spanish (Punta Arenas).

2.2.1 Mexico City Spanish

In the Mexico City variety we ﬁnd the ﬁrst type of process. In such varieties, the
plural -/s/ is voiced, occurring as [2], when it precedes a voiced consonant and voiceless,
occurring as [s], when it precedes a voiceless consonant, a vowel or a pause (Barrutia and
Schwegler 1994). Importantly, [2] occurs only as an allophone of /s/ in Spanish. It is not

an independent phoneme, as in English. This process of assimilation is illustrated in (l 6).

38

(16) a. las bolitas [laz Bolitas#]
the.F.PL marblesrrr.
b. las manos [laz manos#]
the.F.PL hands.F.PL
c. las hojas [las oxas#]
the.F.PL sheets.F.PL
(1. las papas [las papas#]

the.F.PL potatoeS.F.PL

In Mexican Spanish (Mexico City) number is systematically marked by adult speakers.
Morgan (1998) cites Lipski (1994) as reporting that “throughout the interior regions of
Mexico, syllable-ﬁnal /s/ rarely deletes or even aspirates.” and Canﬁeld (1982) as noting
that “[a]mong other speakers of Spanish, a Mexican is recognized by his tendency to

lengthen the articulation of /s/. . .” (p. 82).

2.2.2 Chilean Spanish

The phonological form of the plural morpheme in Chilean Spanish is quite
different from that in Mexican Spanish. Chilean Spanish undergoes a process of lenition
where all syllable ﬁnal /s/ surfaces as either [5], [h], or is omitted Because the plural
morpheme occurs as /s/ in syllable ﬁnal position, this process of lenition affects the

pronunciation of the plural morpheme as well. While lenition is generally discussed in

39

terms of aspiration or omission in Chilean Spanish, our data indicate that /s/ can also

surface as a glottal stop or a lengthened vowel. For the purpose of this dissertation,

however, we will only examine the acquisition of aspiration [h] and the alveolar fricative

[5] forms, given that (1) previous research has mainly focused on these two variants, (2)

[h] appears to be the most common variant in Chilean Spanish and (3) [s] is the most

common form in Mexican Spanish. The Chilean pronunciation of syllable ﬁnal /s/ is

illustrated in (17) for morphological /s/ and (18) for non-morphological /s/.

(17)

9’

Possible Pronunciations

las bolitas [las],[1ah],[la]/[Bolitas], [Bolitah], [Bolita]

the.F.PL marbles.F.PL

las manos [las], [lah], [la]/ [manos], [manoh], [mano]

the.F.PL hands.F.PL

las hojas [las], [lah], [la]/ [oxas], [oxah], [oxa]

the.F.PL sheets.F.PL

las papas [las], [lah], [la]/ [papas], [papah], [papa]

the.F.PL potatoes.F.PL

40

(18) a. bus [bus], [buh], [bu]

bus.SG

b. lapiz [lapis], [lapih], [lapi]
pencil.SG

c. lapices [lapises], [lapiseh], [lapise]
pencils.PL

The phonological variant ([5], [h], or zero) that surfaces is dependent on both linguistic
and extra-linguistic factors. Cepeda (1995) collected 16,117 occurrences of syllable-ﬁnal
[s] lenition in Chilean Spanish by interviewing 34 Chilean adult speakers from Valdivia,
Chile in their homes for 30 minutes each. Valdivia, Chile is located in the southern part
of Chile, approximately 900 miles north of Punta Arenas, where the Chilean subjects in
this dissertation were tested, and 500 miles south of Santiago, the capital of Chile.
Cepeda found that the highest rate of syllable ﬁnal /s/ omission occurred with the plural
morpheme /-s/ on nouns and adjectives (but not on determiners), rather than with non-
morphological /s/. Table 5 shows the percentage of plural morpheme production ([5] or

[h]) vs. omission in Chilean Speakers.

41

Table 5. Plural Morpheme Lenition by Syntactic Category in Chilean Spanish.

 

 

 

 

[s] or [h] Omission Total Number
Premodiﬁers 79% 21% p 3028
Nouns 47% 54% 4580
*Adapted from Cepeda (1995).

In her analysis, Cepeda grouped all determiners as “noun premodiﬁers”, which in
addition to determiners included prenominal adjectives and numerals. For this reason, it
is difﬁcult to lmow for sure what percentage of determiners occurred with or without the
plural morpheme. In any case, Table 5 suggests that the plural morpheme is omitted
much more often on nouns than on determiners. In fact, Table 5 reveals that the plural
morpheme was omitted by adult Chilean speakers (ﬁ’om Valdivia, Chile) approximately
half of the time on nouns (54% of the time).

Cepeda (1995) also found that length of the word in terms of number of syllables
was a Isigniﬁcant predictor of plural morpheme omission on noun premodiﬁers
(determiners, adjectives, numerals), which is important because it provides insight on the
type of determiners that favored or disfavored plural morpheme omission. Table 6 shows
the percentage of plural morpheme omission in monosyllabic vs. polysyllabic noun

premodiﬁers.

42

Table 6. Plural Morpheme Lenition in Mono- and Polysyllabic Noun Premodiﬁers.

 

 

 

 

 

Morphological N on-morphological
Monosyllabic Polysyllabic Monosyllabic Polysyllabic
[s] 6% 6% 15% 14%
[h] 81% 60% 63% 37%
Omission 13% 34% 22% 49%
Total Number 1750 854 315 109
*Adapted ﬁ’om Cepeda (1995).

Table 6 reveals that omission of the plural morpheme in monosyllabic noun premodiﬁers
occurred only 13% of the time while omission of the plural morpheme in polysyllabic
noun premodiﬁers occurred 34% of the time. While Cepeda does not present her data in
terms of deﬁnite vs. indeﬁnite determiners, Table 6 suggests that the plural morpheme is
omitted more often on indeﬁnite determiners, which are polysyllabic in Spanish, (e.g.
unos, ‘some’ algunos ‘some’) than on deﬁnite determiners, which are monosyllabic in
Spanish, (e. g. las, ‘the.F.PL’, mis ‘my.PL’). Table 6 also shows there is a strong preference
for Chilean adults to use the plural variant [h] on determiners.

Phonological environment also inﬂuences syllable ﬁnal /s/ lenition in Chilean
Spanish. Cepeda found that the following sound affected whether syllable ﬁnal /s/ was

produced as [s], [h] or zero. This is illustrated in Table 7.

43

Table 7. The Effect of Phonological Environment on /s/ Lenition in Chilean Spanish.

 

 

 

 

 

_#C _#C _#C _#V _#V _#
l-cont] [+cont] [+nas] (U nstressed) (Stressed) (Pause)
[s] 1% 1% 1% 6% 25% 12%
[h] 83% 55% 71% 42% 38% 35%
Omission 16% 44% 28% 52% 36% 54%
Total 3594 3552 1406 2716 1018 3831
*Adapted from Cepeda (1995).

Table 7 shows that [h] is more common than [s] in Chilean adult speech. In addition,
syllable-ﬁnal /s/ is produced more often when followed by a non-continuant consonant or
a nasal consonant and omitted more often when followed by a continuant consonant, an
unstressed vowel, or a pause.

Finally, Cepeda reported that the level of syllable-ﬁnal /s/ omissiOn was linked to
social class, with omission being much more common in the language of working-class
speakers than in middle-class and upper class speakers. A summary of the distribution of

the various pronunciations of syllable-ﬁnal /s/ by social class is shown in Table 8.

Table 8. Social Distribution of Syllable Final /s/ in Chilean Spanish.

 

Social Class

High Mid Low

 

Retention [s/h] 68% 65% 50%
Omission 32% 35% 50%
Tokens 5338 4968 5791

 

 

Table 8 shows that working-class Chilean adults omit syllable-ﬁnal /s/ more often than
middle-class and upper class speakers, with working-class adults omitting syllable-fmal
/s/ on approximately half of all tokens.

In summary, Cepeda (1995) reported that morphological -/s/ is omitted more often
than non-morphological /s/ and that [h] was the most frequent variant in the data she
collected. Cepeda’s data also suggest that there are several constraints governing plural
morpheme omission: (1) the plural morpheme is omitted more often on nouns than on
determiners, (2) the plural morpheme is omitted more often on polysyllabic noun
premodiﬁers (e. g. algunos ‘some.PL’) than on monosyllabic determiners (e. g. los
‘the.PL’), (3) the phonological environment immediately following the plural morpheme
affects lenition, and (4) working-class speakers tend to omit the plural morpheme more
often than middle-class and upper-class speakers. Chapter 4 of this dissertation will
present additional data collected from Chilean adult speakers (from Punta Arenas, Chile)
with the intention of replicating Cepeda’s ﬁndings. Chapter 4 will also present data on

syllable-fmal /s/ lenition in Chilean children.

45

2.2.3 Syllable Final /s/ Lenition in Other Varieties of Spanish ,

Syllable ﬁnal /s/ lenition is found in several dialects of Spanish (Lipski 1985,
1986, 1999; Cepeda 1995). Terrell (1981) noted that aspiration and deletion of syllable
ﬁnal /s/ is found in all levels of society in the Caribbean (Cuba, Dominican Republic,
Puerto Rico) and in Southern Spain (Andalucia), the Canary Islands, and‘also in Chile,
Argentina, Uruguay and Paraguay. The only Spanish-speaking regions which do not
appear to aspirate or delete syllable ﬁnal /s/ are the interior highlands of Mexico
(including Mexico City), Guatemala, Costa Rica, Colombia, Ecuador, Peru, Bolivia and
Northern Spain. Syllable ﬁnal /s/ lenition, with a tendency for omission in working-class
speech, was ﬁrst reported for Chilean Spanish around the 19th Century (Cepeda 1995
cites the following authors: Lenz 1940, Alonso and Lida 1940, Oroz 1966). It has been
proposed that syllable ﬁnal /s/ lenition originated in the Canary Islands and reached the
Americas by way of slave ships that stopped at the Canary Islands before leaving for the
Americas (Barrutia and Schwegler 1994).

Terrell (1981) noted that syllable-fmal /s/ lenition appears to be at a more
advanced stage in Dominican Spanish than in other varieties of Spanish. Dominican
Spanish appears to have lost syllable ﬁnal /s/ altogether, and he notes that for the most
part any production of /s/ in ﬁnal position is often the result of hypercorrection. Morgan
(1998), for example, also notes that the stigrnatization of ﬁnal /s/ omission that
characterizes working-class speech in the Dominican Republic has given rise to
hypercorrection in Dominican Spanish-speaking adults, where working-class speakers
will often insert [s] and/or [h] in syllable ﬁnal position in more formal social contexts.

This phenomenon is often referred to by Dominican speakers as ‘Hablar F isno ’, which

46

comes from the phrase hablarﬁno (‘reﬁned speaking’), where an [s] or [b] has been

inserted in syllable-fmal position of the ﬁrst syllable in ﬁno (‘reﬁned’).

Morgan (1998) collected 16 minutes of ﬁee speech from a Dominican-speaking

working-class male and found that he hypercorrected in 115 words. Of those 115 words,

70 involved [s]-insertion while 45 involved [h]-insertion. Hypercorrection occurred in all

syntactic categories, including nouns, verbs, determiners, pronouns, adverbs and

prepositions, as shown in Table 9 below.

Table 9. Hypercorrection in Dominican Spanish.

Syntactic Category
Noun
Adverb
Numeral

Verb

Adjective

Determiner

Preposition

Pronoun

 

Spanish Word
foto (‘photo.SG’)
aqui (‘here’)
siete (‘seven’)
torturaba (‘he tortured’)
imitando (‘imitating’)
nuevo (‘new’)
la (‘the.SG’)
una (‘a/one.SG’)
de (‘of’)

se (himself)

Hypercorrected Form

[fotos]
[akiSl
[S’éteS]
[torturabas]
[imitandos]
[nwe'vos]
[1811]
[1111311]
[deb]

[ses]

 

*Adapted from Morgan (1998).

47

Table 9 shows that as a result of hypercorrection, even words that cannot carry plural
morphology, as adverbs and prepositions, have /s/ added to them. In other words, not
only is the plural morpheme sometimes omitted in plural noun phrases in Dominican
Spanish but it is also sometimes placed on semantically singular nouns, as shown in the
example taken from Morgan (1998) in (19) where a Dominican tourist guide is referring

to an important house on a tour in the Dominican Republic.

Singular: Standard Spanish
(19) a. Lg casa construid,a en piedra
The.SG house.SG built.SG in stone

‘The house built in stone...’

Plural: Standard Spanish
b. L_as cagas construidﬁ en piedra
The.PL house.PL built.PL in stone

‘The houses built in stone...’

Singular: Dominican Spanish
c. _L_as cas_a construid_a en piedra
The.SG house.SG built.SG in stone

‘The house built in stone...’ (taken from Morgan 1998)

48

Although the Dominican speaker is referring to a single house in (15c), the deﬁnite
determiner las (the.FL) is used. Morgan’s data suggests that the plural morpheme in
Dominican Spanish is no longer associated solely with the meaning of ‘more than one’
but rather seems to have taken on a different meaning, which is related to social factors.
The data reported in Morgan’s study is relevant for this dissertation as they lend
support to the idea of a gradual change in plural marking over time. The data suggests
that for many Dominican speakers [s]/[h] is no longer associated to an interpretation of
‘more than one’ as adult speakers mark [s]/[h] even on semantically singular nouns and
determiners and they also mark other syntactic categories (verbs, prepositions, adverbs)
with [s]/ [h]. Instead, it appears that [s]/ [h] have come to hold some social meaning for

adult speakers, but not necessarily number information.

2.3 The Underlying Syntactic Representation of the Spanish Plural

Chapter 1 introduced the theoretical proposal, put forth by Yang [(2001), that the
adult E-language does not always match the adult I-language and this may result in
language change as children acquire a grammar that is slightly different from their
parents. The present dissertation examines whether Spanish-speaking children acquire
plural morphology when the input, or the adult E-Language, is variable and whether
development of plural morphology differs between children exposed to and E-language
with systematic input vs. variable input. In this section I will present a syntactic analysis
of Spanish plural morphology with the intention of providing insight on the
representation of plural morphology in the Spanish-speaking adult I-language.

I will adopt Sauerland’s (2003) proposal for the syntactic representation of

plurality, which argues for a separate syntactic head for number, called ¢P, located above

49

the DP. While previous research on plurality based on English has assumed that the
plural feature on nouns is semantically interpretable while the plural feature on verbs is
uninterpretable, Sauerland notes that in languages like German, overt plural morphology
is also found on the determiner, in addition to the noun and that, for this reason, other
options are needed for determining exactly which nominal plural features have semantic
content. In other words, is it the plural morpheme on the noun or on the determiner that
has semantic content? In this sense, Spanish is like German in that both the determiner
and the noun have overt plural morphology.

In order to account for languages like German, Sauerland argues that number
features with semantic content are contained in the (t-head and that the plural feature on
both determiners and nouns are uninterpretable and are instead licensed by syntactic
agreement with the (b-head (the plural feature on the verb is licensed through syntactic
agreement with the ¢P). Syntactically the rp-head can combine with any DP as long as the
semantics of the DP match that of the q>-head. A syntactic representation is shown in

(20)?

(20) P

will )8

D NP
las (the.PL) vacas (cows.PL)7

 

6 This representation allows Sauerland to account for languages like German where the
ural morpheme can occur both on the noun and determiner.
It is assumed that indeﬁnite determiners and numerals occur in NumP, which occurs
between DP and NP.

50

Sauerland proposes that the number features of the tb-head are licensed solely by the
semantics and that the number features on the ¢-head license uninterpretable number
features on the determiner and noun through syntactic agreement, as illustrated in (20)8.
Sauerland proposes that the [SG] feature on the (p-head expresses a presupposition
of a single atom or a mass, while the [PL] feature has no inherent presupposition, rather it
gets a plural reading through implication. Sauerland argues that the distribution of [PL] is
constrained by the pragmatics, which requires that, of two alternative morphemes, the
one with the stronger presupposition must be used whenever its presupposition is
satisﬁed. In other words, the plural morpheme is only used if the singular is blocked.
While I will assume the representation in (20) for the Spanish plural, I will depart
from Sauerland with respect to his semantic representation of plurality as a contrast
between [+SG] vs. [-SG] and also his analysis that the plural interpretation arises through
implicature. Unlike Sauerland, I will assume that plurality is expressed in Spanish as [PL]
on the (ti-head and that the absence of [PL] is associated to an interpretation of ‘one’. This
will be discussed in more detail in Section 2.4. Like Sauerland, however, I will assume

that Spanish morphological number on the determiner and noun is interpreted indirectly

 

8 Within the Minimalist Program Chomsky (1995) discusses the notion of checking
where functional categories carry interpretable and uninterpretable features. Interpretable
features receive an interpretation at the level of derivation corresponding to Logical Form
while uninterpretable features are checked and eliminated before Logical Form (i.e. the
interface between the grammatical and the conceptual-intentional components). Hence,
given the sentence “The duck swims” the subject DP “the duck” has the interpretable
feature [3 person] and the verb “swims” has the uninterpretable feature [3 person]. These
two features agree, are checked and the uninterpretable features are eliminated before the
expression reaches the level of the derivation corresponding to Logical Form. The
uninterpretable features determine what elements can converge within the structure. On
the other hand, in the sentence “The ducks swim” the DP “the ducks” has the feature
[plural] which is an interpertable feature, and plays a role in determining the meaning of
the work ‘duck’.

51

through syntactic agreement. Any element within DP (e. g. noun, determiner, adjective),
which is marked with the plural morpheme, will be used by the speaker in his
interpretation of quantity because it is the reﬂex of agreement with a plural ¢-head.

While this proposal argues that the ¢P combines with DPs, I will assume that bare
singular count nouns are not DPs in Spanish and, hence, are not dominated by a ¢P. As a
result, bare singular count nouns, unlike bare plurals, have no number interpretation
associated to them; they are consistent with both an interpretation of ‘more than one’ and
‘one’. In the literature, bare Singulars have been treated as names of kinds as in
Carlsonian accounts (Carlson 1977, Chierchia 1998, Munn and Schmitt 2003) or as
incorporated nominals (Masullo 1992). For the purposes of this dissertation, I will follow
Masullo (1992) in claiming that Spanish bare singular count nouns are incorporated
nominals.

Masullo (1992) bases his analysis of bare singular count nouns on Grimshaw’s
(1991) notion of extended projection that distinguishes between verbal projections and
nominal projections. VP, IP and CP are projections of V and NP, DP, and PP are
projections of N. These syntactic categories can be distinguished in terms of the values
‘lexical’ and ‘functional’. With respect to nominal projections, N is ‘lexical’ while D and
P are ‘functional’. Masullo assumes that both PPs and DPs are complete maximal
projections, while NPs are defective projections. Masullo argues that bare singular count
nouns are defective nominal projections, in other words, they are nominal projections that

do not project to the maximal functional value (to DP).

52

There are two types of examples of bare singular count nouns that are relevant for
this dissertation, bare Singulars that incorporate into predicates, as in (21) and bare

Singulars that incorporate into light verbs as in (22).

(21) Pedro compro auto.
P. boughtPAsr.3.so car

‘Pedro bought a car.’

(22) Pedro tiene auto.
P. has.PRES.3. SG car

‘Pedro has a car.’

Unlike maximal nominal projections that can be licensed by structural case, Masullo
argues that Spanish bare singular count nouns must be licensed by incorporation. In (21)
the bare singular incorporates into a predicate that theta-marks it and assigns it inherent
case. Given that there is no overt manifestation of this incorporation, Masullo assumes
that incorporation occurs at LF. The examples in (23) and (24) show the steps of

incorporation.

(23) SS: Pedro [vp[v compré][Np auto]

LF: Pedro [vp[vcompro-autoi]ti]]

53

(24) SS: Pedro [vp[v tiene][Np auto]

LF: Pedro [vp[v tiene-autoi]ti]]

Because the bare singular count noun is not a DP, it is not dominated by a ¢P and, for that
reason, has no number information associated to it. It is consistent with either a ‘one’ or
‘more than one’ interpretation. Unlike bare Singulars, bare plurals can sometimes occur in

subject position and, for this reason, bare plurals appear to be full DPS.

(25) a. Llegaron soldados a Santiago.
Arrived.3.PL soldiers to Santiago

‘Soldiers arrived in Santiago.’

b. *Llego soldado a Santiago.
Arrived.3.SG soldier to Santiago

‘A soldier arrived in Santiago.’

Chapter 5 will present an experimental study on Chilean child interpretation of bare
singular count nouns vs. bare plural count nouns.

In summary, this syntactic account of the Spanish plural proposes that (1)
semantic number is represented on the (ti-head, which is the sister of DP, (2) nominal
number is interpreted indirectly through syntactic agreement between the rp-head and
elements within the DP, and (3) bare singular count nouns are not dominated by a ¢P and,

as a result, are consistent with both an interpretation of ‘more than one’ and ‘one’.

54

2.4 The Underlying Semantic Representation of the Spanish Plural

In this section I will provide a semantic representation for the feature [PL]. For
the purposes of this dissertation, I will assume Ojeda’s (1998) model-theoretic account of
the notions of singularity and plurality. While there are many ways of dealing with the
notions of singularity and plurality, Ojeda’s proposal has an advantage in that it allows us
to associate plurality with ‘more than one’.

Ojeda (1998) discusses the interpretations assigned to singular and plural nouns

through the use of mereologies. A mereology can be deﬁned as a theory of parthood

relations, of the relations of a part to a whole and the relations of parts to other parts within a
whole. Schwarzschild (1996), for example, provides a set-theoretical account of plurals that
links plurals to part structures in the sense that a set of individuals induces a larger domain
that not only contains the individuals but also the sets formed from those individuals.
Schwarzschild deﬁnes plurals as in (26).

(26) If or is a singular common noun and B is the plural of or, the lll3llM is the set

of all non-empty subsets of "all”.

This rule can be illustrated in example (27).

(27) The boys clapped.
According to rule (26), ‘boys’ denotes the set of all non-empty sets of boys. In other
words, ‘boys ’ denotes not only the individuals, or singleton sets, but also all the sets formed

from those individuals. We will see below that Ojeda departs from Schwarzschﬂd by not

allowing the singleton sets, or the individuals, to be included in the extension of the plural.

55

For the purposes of this dissertation, I will follow Ojeda in assuming that the plural does

not include singleton sets.
Ojeda (1998) examines the notions of singularity and plurality expressed by
nouns in Papago. In Papago nominal number is expressed through the rightward

reduplication of the ﬁrst syllable of the root as shown in (28)9.

(28) a. ban (root)

‘coyote/coyotes’
b. ban

‘coyote’
c. ba-ban

‘coyotes’

Ojeda notes that nominal roots in Papago are generally associated with both a singular and
plural reading.lo To account for the plural and singular interpretation associated with roots
he proposes that the nominal root can be represented by the mereology diagrammed in
Figure 4, where the denotation of the root contains both the monoatomic and polyatomic
elements of the mereology. The nominal root is associated to an interpretation of ‘one’ in

the sense that its denotation contains the set of atoms (or monoatomic elements) of the

 

9 Ojeda also discusses the notions of singularity and plurality expressed on verbs and the
notions of distributivity expressed on both nouns and verbs. This will not be discussed
here because it is not directly related to the experimental studies presented in this
dissertation.

1° Ojeda cites Wackemagel (1920) and (1883) as being among the ﬁrst to observe these
properties of nominal roots.

56

mereology and it is also associated to an interpretation of ‘more than one’ in the sense that

its denotation contains all of the polyatomic elements within the mereology.

Figure 4. Mereological Representation of a Nominal Root (adapted from Ojeda
1998)
abc
/I\

ab be ac

WT

To account for the reduplicated nouns, as in (28c), which are associated to an
interpretation of ‘more than one’ in Papago, Ojeda proposes that the denotation of
reduplicated nouns (or plural nouns) contains only a part or subset of the denotation of
the root. The denotation of plural nouns in Papago contains “the set of sums of
nonidentical atoms in the denotation of its root” (p. 253). In other words, the denotation
for plural nouns contains the set of polyatomic elements of the root. This is represented

by the mereology diagrammed in Figure 5.

Figure 5. Mereological Representation of a Plural Noun (adapted from Ojeda

 

1998).
abc
ab/blc\ ac

 

 

 

Finally, Ojeda notes that singular nouns in Papago denote the set of atoms in the
denotation; hence, they can be interpreted as the set-theoretical difference between the

denotation of their roots and the denotation of their plural counterparts. If the root is

57

represented as in Figure 4 and the plural as in Figure 5 then singular nouns in Papago can

be represented as in Figure 6.

Figure 6. Mereological Representation of a Singular Noun (adapted from Ojeda
1998).

 

Under Ojeda’s proposal, the singular arises as the complement of the plural and he notes
that his proposal captures the widely held intuition that plurality is about plural vs. non-
plural. Singular is what is left when plural is not present.

Ojeda’s proposal for singular vs. plural in Papago can be applied to Spanish. I will
ﬁrst start with Spanish bare singulars, proposing that their denotation is the same as for
Papago nominal roots, and then discuss Spanish plural and singular nouns. As noted in
Section 2.1 and Section 2.3 Spanish bare singular count nouns can occur as complements
of verbs and prepositions and are, for the purposes of this dissertation, analyzed as
incorporated nominals. Like Papago roots, Spanish bare singulars appear to be neutral

with respect to number distinctions, as illustrated by the translations of examples in (29).
(29) Tengo perro, tele, teléfono, hijo.

I have dog, tv, telephone, child

‘I have dog/dogs, tv/tvs, telephone/telephones, child/children. ’

58

Notice that (29) is true if the speaker possesses ‘one’ or ‘more than one’ of these items. In
(29) the bare singular indicates that the speaker is an owner of such objects (e. g. a dog-
owner, tv-owner, etc.) and, hence, having ‘at least one’ (but possibly ‘more than one’) is
sufﬁcient for (29) to be felicitous. Given the number neutral characteristics of bare
singulars, I would like to relate them to Ojeda’s proposal for Papago nominal roots.
Oj eda proposes that Papago roots are associated to a an interpretation of ‘one’ because
their denotation contains the set of atoms in the mereology and are also associated to an
interpretation of ‘more than one’ because their denotation contains a set of polyatomic
elements in the mereology. I will assume that Spanish bare singulars have the same
semantic representation as Papago root nominals. In other words, the denotation of
Spanish bare singulars is as shown in Figure 4.

Unlike Spanish bare singulars, Spanish bare plurals seem to be associated
primarily to an interpretation of ‘more than one’ but not to an interpretation of ‘one’ , as

illustrated by (30).

(30) Tengo perros, teles, teléfonos, hijos.
I have dogs.PL, tvs.PL, telephones.PL, kids.PL

‘I have dogs, tvs, telephones, kids.’
Notice that (30) is true if the speaker possesses ‘more than one’ of these items but would

be odd if the speaker possesses ‘one’ of the items. Hence, the examples in (30) suggest

that the plural morpheme is associated to an interpretation of ‘more than one’ in Spanish.

59

Following Ojeda, we can represent the semantics of Spanish plural nouns as in Figure 5.
Oj eda proposes that the denotation of the Papago plural contains the set of polyatomic
elements. I will assume that Spanish plurals have the same semantic representation.

Turning now to the Spanish singular we ﬁnd that the form of the noun is identical
in phonological form to the Spanish bare singular but its semantic representation is
different. According to Ojeda, “singulars can be interpreted as the set-theoretical
difference between the denotation of their roots and the denotation of their plural
counterparts” (p. 255). Hence, if Spanish bare singulars are represented as in Figure 4 and
Spanish plurals is as in Figure 5, then the Spanish singular can be represented by Figure
6, which is the set of all on monoatomic elements. Given the representation in Figure 6,
singularity is represented by all of the monoatomic elements, which are located in the
lowest tier and the interpretation of the singular arises as the complement of the plural, as
a contrast of plural vs. non-plural.

Based on Ojeda (1998), this section proposes the following about the semantic
representation of Spanish plurality: (1) the Spanish bare singular is semantically
represented as all the polyatomic and monoatomic elements in the mereology and is
consistent with both an interpretation of ‘one’ and of ‘more than one’ (2) the Spanish
plural is semantically represented as all of the polyatomic elements in the mereology and
is to be associated to an interpretation of ‘more than one’ (3) the Spanish singular is
semantically represented as the complement of the plural, in other words, as the set-
theoretical difference between the denotation of the plural and the denotation of the root

and is associated to an interpretation of ‘one’.

60

2.5 Implications for Interpretation by Spanish Speakers

The discussion above assumes that Chilean and Mexican Spanish are equivalent
with respect to the underlying syntactic and semantic representation for plural
morphology but differ with respect to which phonological form of the plural morpheme
agrees with or links to the iii-head. In Mexican Spanish both [5] and [2] agree with the
[+PL] ¢P and hence are to be associated to an interpretation of ‘more than one’ while the
absence of [s] and [2] (zero) shows agreement with an unmarked ¢P and are to be

associated to an interpretation of ‘one’. This is shown in (31) - (33).

(31) P
4) /\ P
[+Pll A

D NP
las (the.PL) vacas (cows.PL)

(32) <l>P

la (theﬂ) vaca (cowﬂ)

(33) ¢P

,/\ D,
“P” m

vacas (cow.PL)

61

Bare singular count nouns are extremely restricted in Mexican Spanish and are almost
absent.

In Chilean Spanish when the rp-head is [+PL] the determiner and the noun must
agree with it and this agreement is spelled out as the phonological forms [8], [h], or zero.
The absence of the plural morpheme, or the [zero] form, is also linked to an unmarked r1)-
head and, as a result, is also associated to an interpretation of ‘one’. As a result, there
appears to be an overlap in Chilean Spanish where the [zero] form may be linked to both
a [+PL] ctr-head and an unmarked tb-head and, as a result, be associated to an

interpretation of ‘more than one’ or ‘one’, respectively. This idea is illustrated in (34) —

(37).
(34) P
«p P
W” A
D NP
las (the.PL) vacas (cow.PL)
(35) P
c]; P
[+P1] A
D NP

las (the.PL) vacah (cow.PL)

62

(36) P
¢/\ P
[+P1] A

D NP
las (the.PL) vacaQ (cow.PL)

07> /”’\
‘i’ )P\
D NP
la (the.SG) vacaQ (cow.SG)

The examples in (34) - (36) suggest that [s], [h], and [zero] on the noun can be linked to a
[+PL] d-head. (37) shows that the [zero] form can also be linked to an unmarked (ti-head.
However, based on the adult Chilean data collected for this dissertation, we ﬁnd that
there is generally some element within the plural DP that indicates plurality. This element
may be plural morphology itself, a numeral, a quantiﬁer or the epenthetic [e] that surfaces
when the noun ends with a consonant (e. g. peces pronounced as [pese] instead of
[peses]). In rare cases, the context may be used to distinguish plural from singular. Table

10 provides some examples of Chilean adult production data.

63

Table 10. Chilean Adult Speech Samples.

 

Plural (Isl/[h])

Zero Plural Morpheme

 

Deﬁnites

Indeﬁnites

Quantiﬁers

Nouns w/

Numerals

Nouns w/

Determiners

 

mi-[h] niﬂita-[o]
my.PL daughter.PL

‘my daughters’

uno- [h] aﬁo-[h]
some.PL years.PL

‘some years’

de toda-[h] manera-[o]
in all.PL cases.PL

‘in any case’

treinta grado-[s]
thirty degrees.PL

‘thirty degrees’

una-[h] murIeca—[s]
some.PL dolls.PL

‘some dolls’

todo-[o] lo-[o] aﬁo—[e]
all.PL the.PL year.PL

‘every year’

uno-[o] mese-[e]
some.PL months.PL

‘some months’

todo-[e] lo-[h] materiale-[h]
all.PL the.PL materials.PL

‘all of the materials’

uno se va do-[h] mese-[e]
one goes two months.PL

‘one goes for two months’

lo-[h] dia-[o]
the.PL days.PL

‘the days’

 

That data in Table 10 suggest that there is always something in the semantically plural

DP that indicates that a plural morpheme is present and hence that there is agreement

between this element and the [+PL] (ti-head. We do not ﬁnd instances in Chilean adult
speech, for example, of semantically plural DPs that completely lack any plural number
information whatsoever. For this reason, we must conclude that Chilean adult speakers
associate the phonological variants [s] and [h], and sometimes zero, to an interpretation
of ‘more than one’ as long as there is some information in the DP that shows that the 4)-
head is [+PL]. In addition, Chilean adult speakers can link inherently plural lexical items
such as plural quantiﬁers and numerals to a plural morpheme that has undergone syllable-
ﬁnal /s/ lenition and, hence, to a [+PL] cp-head and assign that plural DP a ‘more than
one’ interpretation.

Given the representation above, the interpretation that adult Mexican and adult
Chilean speakers will assign to plural and singular noun phrases is discussed below. Let

us ﬁrst consider Mexican Spanish, as illustrated in the examples (3 8) — (41) below.

(38) Pon unas bolitas en la caja.
Put some.F.PL marbles.F.PL in the box

‘Put some marbles in the box.’

(39) Pon una bolita en la caja.

Put a.F.SG marble.F.SG in the box

‘Put a/one marble in the box.’

65

(40) Dame las hormigas.
Give.me the.F.PL ants.F.PL

‘Give me the ants.’

(41) Dame la hormiga.
Give.me the.F.SG antrso

‘Give me the ant.’

Adult Mexican Spanish speakers assign a plural interpretation to (3 8) and (40) because
the plural morpheme /s/ indicates agreement with [+PL] ¢P and is associated to
interpretation of ‘more than one’. On the other hand, Mexican Spanish speakers assign an
interpretation of ‘one’ to (39) and (41) because the absence of the plural morpheme
indicates agreement with an unmarked (bP.

The interpretation that Chilean adults would assign to plural and singular DPs is a
little more complex. Given that number marking on the Spanish determiner, noun and
adjective is the result of agreement with a [+PL] ¢-head, it seems reasonable to assume
that as long as one of the elements within the DP is marked with the plural morpheme [s]
or [b], the adult would link that element with a [+PL] (ii-head. As a result, the entire DP
would be associated to an interpretation of ‘more than one’. Consider the following

sentences.

66

(42)

(43)

(44)

(45)

Adult Chilean speakers would assign a ‘more than one’ interpretation to (42) and (44)
because there is at least one element within the noun that agrees with the [+PL] o-head.
However, they would assign an interpretation of ‘one’ to (43) and (45) because the

absence of the plural morpheme on all elements within the DP agrees with an unmarked

cp-head.

While the availability of the plural morpheme on at least one element within DP

may be sufﬁcient evidence for Chilean adult speakers in their interpretation of number, it

Pon unas bolitaQ en la caja.
Put some.F.PL marbles.F.PL in the box

‘Put some marbles in the box.’

Pon una bolita en la caja.
Put a.F.SG marble.F.SG in the box

‘Put a/one marble in the box.’

Dame lah hormigaZ.
Give.me the.F.PL ants.F.PL

‘Give me the ants.’

Dame la hormiga.

Give.me the.F.SG ant.F.SG

‘Give me the ant.’

67

may not be sufficient for Chilean children in their early development of plural
morphology. The task of acquiring plural morphology requires Chilean children to
acquire the forms associated to plural morphology, acquire the sociolinguistic variation
associated to the different forms, and use plural morphology to interpret the contrast
between ‘one’ vs. ‘more than one’. For Mexican Spanish this seems straightforward since
both [3] and [z] consistently map to a [+PL] (ti-head in the input (adult E-language) and
the alternation between [s] and [z] is categorical, hence there is not no sociolinguistic
variation to acquire. Furthermore, there is no ambiguity. However, for Chilean children
the task seems a bit more complex. Chilean children must link the phonological forms [3]
and [h] to [+PL] ¢~head in order to use plural morphology to interpret the contrast
between ‘one’ vs. ‘more than one’. They must also realize that [zero] can link to a [+PL]
¢-head, as long as there is something else in the DP that marks plurality (e. g. a plural
morpheme, quantiﬁer, numeral). Because [zero] also agrees with an unmarked e-head
and is associated to an interpretation of ‘one’ there appears to be some ambiguity in the
input that Chilean children receive. In Chapter 4 and Chapter 5 I will present a series of
experiments on Mexican and Chilean Spanish-speaking children that investigate how
these different types of input affect their production and comprehension of plural

morphology.

68

CHAPTER 3

PREVIOUS LANGUAGE ACQUISITION RESEARCH

3.0 Introduction

This thesis is working from the assumption that language acquisition involves two
main interacting components: (1) an innate language acquisition device (LAD) and (2)
linguistic experience (input), I further assume that, while the LAD is generally invariable
across typically-developing human populations (all typically-developing humans have the
ability to acquire language), the linguistic experience or input varies across and within
human populations, to the extent that even the linguistic experience of two siblings
differs, just not to the extent of creating unintelligible languages. This dissertation is
primarily concerned with the second component, the linguistic input and the underlying
research question is not simply how many tokens does a child need to be exposed to in
order to acquire a grammar that matches the adult grammar but rather how many variable
tokens (including zero tokens) can children support before they construct a grammar that
does not match the adult grammar?

Previous studies examining the effect of input on language development have
been primarily focused on the frequency of a particular morpheme in language. This type
of research is often carried out by cross-linguistic comparisons (Slobin 1985, Lillo-
Martin and Snyder 2002, Kupisch 2006) but it has also been done by comparing the
development of different grammatical morphemes (that occur at different levels of
frequency) within one language. For example, Brown (197 3) compared the order of
grammatical morpheme acquisition in children to the frequency of grammatical

morphemes in adult speech and found that the grammatical morphemes that are more

69

frequent in the adult speech are acquired more quickly by children Brown found that the
plural morpheme was more frequent than the possessive morpheme in adult speech and
English-speaking children likewise acquired the plural morpheme /s/ before the
possessive morpheme /s/. Frequency studies of this sort, while very informative, appear
not to take into consideration neither the different complexities of the grammatical
elements being compared nor the reliability of the input with respect to the grammatical
morphemes under consideration. For example, while the plural morpheme /s/ is almost
always present when plurality is expressed in Standard English, there are cases when it is
not (e. g. children, teeth, sheep) and there are cases where it is present and does not mean
‘more than one’ (e. g. scissors, pants). It is interesting to note that these exceptions appear
to be tolerated by English-speaking children and do not appear to pose a great problem.
However, we could imagine other situations where the input is much more
variable, as is the case when we encounter sociolinguistic variation. For example, if we
were to examine the development of another property of English that seems to have a
much more variable behavior, as appears to be the case with the production of auxiliaries
by adult English speakers, we would need to consider not only the frequency of auxiliary
production but also the reliability of that production. Examples (1) — (7) in Table 11 show

that the production of auxiliaries in English is variable.

70

 

Table 11. Variation in Auxiliary Production.
Standard Nonstandard

Production of Auxiliary Omission of Auxiliary

(1) Did she eat? a. Did she eat? b. She eat?

(2) Does she eat carrots. a. Does she eat carrots? b. She eat carrots?

9?

(3) Do you like carrots? . Do you like carrots? b. You like carrots?

Standard Nonstandard
Agreement No Agreement
(4) Doesn’t she know? 3. Doesn’t she know? b. Don’t she know?

(5) She doesn’t like him. a. She doesn’t like him. b. She don’t like him.
(6) She does like him. 3. She does like him. b.*She do like him.

(7) Does she like carrots? a. Does she like carrots? b. *Do she like carrots?

 

The paradigm above shows that auxiliaries can be omitted and may or may not be
inﬂected for agreement in some non-standard varieties of English. The omission and lack
of agreement is both linguistically (e. g. questions, negation, declaratives) and extra-
linguistically constrained (e. g. social class, speech style). The only unacceptable

examples in adult speech are those starred examples in (6b) and (7b). Given two children,

71

Child A and Child B, with Child A being exposed to a variety of English which prefers
omission of the auxiliary and no agreement between the auxiliary and the subject while
Child B being exposed to a variety of English which prefers production of the auxiliary
and Subj ect-Aux agreement, we might predict that the underlying systems that are
initially constructed by Child A and Child B may be different, even though Child A and
Child B are exposed to all of the forms above, just at different ﬁ'equencies. Would it be
possible to detect this underlying grammatical difference between Child A and Child B in
their speech? Given that the input to Child A would contain less agreement and more
omissions, there would be very little evidence for the auxiliary ‘does’ and hence this
might predict that Child A would produce sentences like (6b) and (7b) while Child B
would not. Already, research on middle-class English-speaking children has shown that
sentences like (6b) and (7b) are unattested in child speech (Guasti and Rizzi 2002,
Schutze 2006). However, if sentences like (6b) and (7b) are found in the speech of
children whose parents sometimes omit the auxiliary, it would suggest that both the
properties of the input and the ﬁequency of such properties are relevant for determining
the course of development of any grammatical feature.

There have been a few studies that have taken both the properties and frequency
of the input into consideration. However, studies of this sort are very rare. Those that
exist have been carried out mainly in the ﬁeld of sociolinguistics and SLI research and
can be classiﬁed as one of two types: (1) as comparing the development of Form A to
Form B in a particular context when the input provides evidence for both at varying
degrees (Park 1978, Gathercole 1986, Kopcke 1998) and (2) examining the development

of Form A when the input provides both evidence for (overt realization of the form) and

72

evidence against (e. g. omission of the form) this form (Ramer and Rees 1973, Moore
1979, Kovac 1981, Johnson 2005, Hudson Kam and Newport 2005). Research that
examines both the frequency and reliability of the input on language development then
poses a different research question than has previously been proposed in frequency
studies. As noted above, the research question is not simply how many tokens does a
child need to be exposed to in order to acquire a grammar that matches the adult grammar
but rather how many variable tokens (including omitted tokens) can children tolerate
before they construct a grammar that does not match the adult grammar?

This chapter is set up as follows: First, I will present previous research on the
production of plural morphology in English-speaking and Spanish-speaking children who
are exposed to an input with systematic plural marking. I will discuss both free speech
and elicitation tasks. It is important to compare both languages given that Spanish marks
plural on all elements within the determiner phrase while English generally only marks
plural on the noun. The purpose of this ﬁrst section is to see whether acquisition of the
plural morpheme will be the same in both languages as long as the input is systematic.
Secondly, I will discuss research that has examined where in the determiner phrase
Spanish-speaking children initially place the plural morpheme. Third, I will present
previous research on the comprehension of plural morphology in English-speaking
children who are exposed to an input with systematic plural marking. Only research with
English-speaking children will be discussed because, as far as I know, there are no
studies on the comprehension of plural morphology by Spanish-speaking children who
are exposed to systematic plural marking in the input. Finally, I will present previous

research on the acquisition of variable input and also research on the acquisition of

73

inconsistent input and discuss how exposure to variable vs. inconsistent input makes

different predictions for language acquisition.
3.1 Acquisition of Plural Morphology

3.1.1 Production of Plural Morphology

What do we know about the production of the plural morpheme by children in
languages like English, where the plural morpheme is only marked on the noun, vs.
standard varieties of Spanish, where the plural morpheme is marked on all elements
within the determiner phrase? This question is important because it may be the case that
the more redundant the plural marking is (i.e. marked on various elements within the
determiner phrase), the quicker children will acquire plural morphology in their own
language. One could argue that the plural morpheme is more frequent in Spanish than in
English because in Spanish the plural morpheme occurs on all elements within the
determiner phrase. This is especially important for this dissertation given that Mexican
Spanish will have more redundancy in plural marking than Chilean Spanish because
Chilean Spanish-speaking adults omit the plural morpheme on various elements within
the determiner phrase. If redundancy facilitates plural morpheme development, then we
would expect both Chilean and Mexican Spanish-speaking children to acquire plural
morphology earlier than English-speaking children and, likewise, we would expect
Mexican children to acquire plural morphology before both Chilean children and
English-speaking children. Even if Chilean Spanish-speaking adults omit the plural
morpheme some of the time, they will still produce it on more elements within the

determiner phrase than English-speaking adults. This section will examine the production

74

of plural morphology in Spanish and English varieties where the plural morpheme is
systematically produced in the adult speech.

It has been reported for English that the plural morpheme is the ﬁrst bound
morpheme that children produce in the noun phrase11 (Brown 1973, de Villiers and de
Villiers 1973). In free speech, English-speaking children start producing the plural
morpheme systematically on all semantically plural nouns by approximately 2 years of
age (Cazden 1968, Mervis and Johnson 1991). In addition, Mervis and Johnson found
that the only errors in free speech that their 1:7 year old English-speaking subject
produced with respect to the plural morpheme were over-regularization errors where the
child added -s to an irregular plural noun (“sheeps”, “mens”) or omitted the ﬁnal /s/ from
a semantically singular noun (“scissor” instead of “scissors”). In other words, by 1:7
years of age this child produced the plural morpheme on semantically plural nouns more
than 90% of the time.

In addition to free speech, elicitation tasks have been carried out to test child
production of the plural. Berko (195 8) tested child production of the three allomorphs of
the English plural morpheme by eliciting plural nouns ﬁ'om chilch’en between the ages of
4 — 7. Children were shown a picture of one novel object (Figure 7) followed by a
sentence like (8a). Next, they were shown two of the objects and asked to complete the

sentence in (8b).

 

’1 Brown (1973) and Cazden (1973) considered that a child had acquired the morpheme
when it was produced at least 90% of the time in obligatory contexts.

75

Figure 7. The Wug Test (Berko 1958).

 

 

 

 

 

(8) a. This is a wug.

b. Now there are two of them. There are two

This technique allowed Berko to elicit the plural morpheme in several phonetic
environments in order to see whether children produced the correct allomorphs of the
plural morpheme ([s], [2], [oz]). Importantly, the plural form of both real words and novel
words was elicited from children in order to ensure that children’s use of the allomorphs
was productive. The ﬁndings of Berko’s experiment revealed that children had little
difﬁculty providing the plural allomorphs [s] and [z] with both real and novel words.
However, while they were able to produce the plural allomorph [oz] with real words, they
were not able to do so with novel words. While this study shows that children use the
plural morpheme (and at least the two plural allomorphs [s] and [2]) productively, it does

not show whether children use the plural morpheme in contexts associated to an

76

interpretation of ‘more than one’. Given the nature of the task, the fact that children had
to produce a plural noun after a numeral, it is possible that children are simply making
the noun agree with the numeral, not really comprehending that the plural morpheme can
be used to indicate ‘more than one’.

In a different type of elicitation task, Ferenz and Prasada (2002) elicited the plural
morpheme on deﬁnite noun phrases and noun phrases preceded by quantiﬁers and
numerals with 48 Mainstream English-speaking children ages 1:9 - 5:6. For the deﬁnite
noun phrase condition, there were two stories: (1) Singular Story: Big Bird told a joke to
one cat and (2) Plural Story: Big Bird read a story to several frogs. The children were
then asked to say what happened by completing sentences: (1) Big Bird told a joke to the
caL (singular noun) and (2) Big Bird read a story to the _mgL (plural noun). Since the
deﬁnite determiner in English provides no number information, children who produce the
plural morpheme in the plural condition, but not in the singular condition, demonstrate
that they are associating the plural morpheme to an interpretation of ‘more than one’.
Their ﬁndings showed that 47 out of 48 children produced the plural morpheme in the
plural condition and only 1 out of 48 produced the plural morpheme in the singular
condition. This experiment indicates that English-speaking children produce the plural
morpheme to express the meaning of ‘more than one’ by 2:0 years of age. Additionally,
these same children showed that they systematically produced plural nouns when the
noun was preceded by quantiﬁers and plural numerals.

Next, we turn to the production of plural morphology by Spanish-speaking
children. First, we will discuss the production of plural morphology in ﬁ'ee speech and

then in elicitation tasks. For comparative purposes, we want to see whether Spanish-

77

speaking children with redundant plural marking begin to produce plural morphology at
the same time or before English-speaking children, who are exposed to non-redundant
plural marking. In other words, we are interested in whether Spanish-speaking children
start producing the plural morpheme by at least 2 years of age.

Kvaal et al. (1988) collected spontaneous speech samples from 15 Mexican-
American Spanish-speaking monolingual working-class children between the ages of 2:0
— 4:8 who lived in or around San Diego, California They showed that children began
producing the plural morpheme by 2:0 years of age. Their ﬁndings are consistent with a
similar study by Marrero and Aguirre (2003) who found that two Peninsular Spanish-
speaking children living in Madrid started producing the plural morpheme by 1:9 years of
age (see Section 3.1.4 for details associated to variation). Although the authors did not
mention whether these Spanish-speaking children were acquiring a variety of Spanish
with syllable ﬁnal /s/ lenition, it does appear that Madrilei’lo Spanish and several varieties
of Mexican Spanish do not have syllable ﬁnal /s/ lenition (Lipski 1994, Barrutia et al.
1994). On the other hand, Vivas (1979) found that four Spanish-speaking immigrant
children living in Colorado (USA) did not begin producing the plural morpheme on
plural nouns until around 2:4 years of age and only did so on semantically plural nouns
between 50% (7/14) — 65% (13/20) of the time. Even by 3:5 years of age, one Spanish-
speaking child produced the plural morpheme on plural nouns only 83% of the time.
Their ﬁndings show a gradual increase in usage of the plural morpheme and at the same
time show that Spanish-speaking children as old as 3:5 still omit the plural morpheme
approximately 20% of the time. They suggested that the omission of the plural morpheme

(approximately 20% of the time) may possibly be due to the aspiration of the plural

78

morpheme in some dialects of Latin American Spanish: however, they provided no
information about the extent to which the plural morpheme was aspirated in the linguistic
input these children were exposed to. In any case, these studies examining the free
speech of Spanish-speaking children suggest that when the plural morpheme is
systematically produced in the adult speech, Spanish-speaking children, like English-
speaking children, begin using the plural morpheme by at least 2:0 years of age. From
these studies we have no evidence that the plural morpheme shows up earlier in the free
speech of Spanish-speaking children and hence have no evidence that redundancy of
plural marking provides a beneﬁt for acquiring plural morphology.

Subsequent to the publication of Berko (195 8), several researchers began to carry
out similar studies in a variety of languages using the Berko elicitation technique
described above. I will outline here those experiments that were carried out with Spanish-
speaking children. Kernan and Blount (1966) tested 92 5-12 year old working-class
Mexican Spanish-speaking children from the city of Guzman in the state of Jalisco,
Mexico on a Berko style task that required children to provide the plural forms of
singular novel nouns. Subjects were shown novel objects and told “Este es unfetor.
Ahora hay otro. Hay dos de ellos. Hay dos . " (“This is a fetor. Now there is
another one. There are two of them. There are two .”). While 18 adults (from
the same community) provided the correct plural forms 100% of the time on novel words
for both the [s] and [es] forms of the plural, children produced the correct plural 93% -
100% of the time on novel words requiring the [5] form but only between 38% (for the
younger children) to 53% (for the older children) of the time for novel words requiring

the [es] form. When the [es] form was required, children generally just omitted it,

79

although a few children incorrectly produced the plural form of the word with [s] (e.g.
fetors).

In an almost identical task", Perez-Pereira (1989) ‘showed that for real and novel
words requiring the [8] form, 109 3 — 6 year old Spanish-speaking children from La
Corur‘la, Spain (in Galicia) produced the plural morpheme more than 92% of the time.
However, for words requiring the [es] form (e. g. real words: ﬂor ‘ﬂower’, arbol ‘tree’,
autobus ‘bus’: novel words: sibil, tipon, patus, astor), children provided the plural
morpheme for real words between 76% (for 3 year olds) to 100% (for 4, 5, 6 year olds) of
the time but only provided the correct form for novel words between 36% (for 3 year
olds) to 55% (for 6 year olds) of the time. The most common mistakes in both studies
consisted of using the form [3] instead of [es] or simply repeating the word in its singular
form with no plural morpheme (especially when the novel word already ended in [s]).
These ﬁndings are consistent with what Berko (1958) found for English-speaking
children.

The experiments reported in Keman and Blount (1966) and Perez-Pereira (1989)
show that by at least 3 years of age, Spanish-speaking children produce the plural
morpheme on nouns that follow quantiﬁers; however, their studies do not show that
Spanish-speaking children use the plural morpheme to express the meaning of ‘more than
one’. Instead, it may simply be the case that children are making the noun agree with the

plural numeral or quantiﬁer.

 

’2 Researcher showed child real and novel objects and said, “Esta es un globo. Ahora

hemos puesto otro mds. Ahora hay dos . (“This is a balloon. Now we have put
one more here. Now there are two .”).

80

One study that provides more information about the interpretation that Spanish-
speaking children assign to the plural morpheme was carried out by Bedore and Leonard
(2001), who examined the production of plural morphology in SLI (Speciﬁc Language
Impairment) and typically developing Mexican Spanish-speaking working-class children
living in the metropolitan area of San Diego, California. Children were acquiring Spanish
as their ﬁrst language and had minimal contact with English. Although they do not
mention whether these children were acquiring of dialect of Spanish subject to syllable
ﬁnal /s/ lenition, we do know that several varieties of Mexican Spanish do not allow
syllable ﬁnal /s/ lenition (Lipski 1994). 45 children participated in the study: 15 SLI
children, 15 typically developing age matched controls (4:0 — 5:6), and 15 typically
developing MLU matched controls (2:4 —— 3:10). There were several tasks that required
children to produce the plural forms [es] and [s] with real words. I will discuss only the
two that are relevant for this dissertation. In the ﬁrst task the researcher showed children
pictures of plural and singular objects and asked the child, “4' Qué hay aqui? ” (“What
is/are here”). Importantly for this task, the Spanish existential ‘hay’ (‘there is/are’) does
not have any number information associated with it: it is used with both plural and

singular nouns, as shown in (9) and (10) below.
(9) Hay un elefante en el patio.

There-is an/one.SG elephant.SG in the backyard

‘There is an elephant in the back yard ’

81

(10) Hay elefantes en el patio.
There-are elephants.PL in the backyard

‘There are three elephants in the backyard."

This means that children who produce plural nouns may not be doing so for reasons of
agreement (i.e. agreement between the verbal existential predicate and the NP) but rather
because they associate the plural morpheme with an interpretation of ‘more than one’.
The results of their study showed that 4:0 — 5:6 year old typically developing children
produced the plural morpheme on real word nouns when presented with pictures of plural
objects 96% of the time, the 2:4 — 3:10 year old typically developing children did so 75%
of the time but the SL1 children only 56% of the time. The differences between the three
groups were signiﬁcant. Child errors always consisted of children omitting the plural
morpheme on the noun when presented with a plural picture. These results indicate that
by at least 4:0 years of age, most Mexican Spanish-speaking children produce the plural
when describing plural sets.

One problem with Bedore and Leonard’s (2001) study is that it is unclear from the
way the results are presented whether children were producing bare plurals, indeﬁnite
plurals or ‘quantiﬁer + plural noun’ constructions to describe pictures of plural sets of
objects. For this reason, there is no way of telling whether children are using the plural
morpheme to express ‘more than one’ or if they are using a lexical determiner to express
‘more than one’ and the plural morpheme surfaces through agreement. For example, if
children produced ‘una vaca’ (a/one.SG cow.SG) for a picture of a single cow and ‘unas

vacas’ (some.PL cows.PL) for a picture of several cows, the only difference between the

82

two responses being the plural morpheme —s on the noun and the determiner, then it
would appear that children are aware that the plural morpheme is associated to an
interpretation of ‘more than one’. However, if a child produces ‘una vaca ’ (alone. 50
cow.SG) for the singular picture and ‘vacas’ (cows.PL) for the plural picture, we cannot
conclude that the plural morpheme is being used to distinguish plural from singular given
that the plural morpheme is not the only difference between the two responses given by
this child. It is important to note that, in a separate task, Bedore and Leonard attempted to
elicit just indeﬁnite plurals (e. g. unas vacas ‘some.PL cows.PL’) but children did not
produce just indeﬁnite plurals, rather they also produced bare plurals, which indicates
that children most likely produced both indeﬁnite plurals and bare plmals in the task just
described above, but it is unclear to what extent. In any case, it can be concluded ﬁ'om
Bedore and Leonard’s results that by 4:0 years of age most typically developing Mexican
Spanish-speaking children use the plural morpheme in plural noun phrases.

Canti'r-Sanchez and Grinstead (2004) carried out an experiment with 10 SLI
(speciﬁc language impairment) Spanish-speaking children and 20 typically-developing 4-
year old (10 age-matched, Mean Age: 4:7 and 10 MLU matched, Mean Age: 4.0)
Spanish-speaking children from Mexico City. The experiment was similar to that of
Bedore and Leonard (2001) in that it elicited singular and plural nouns without providing
any number information in the experimental question. The researcher presented a picture
of a single object to children followed by a picture of two objects and said, “Aqui tengo
una mariposa. g Y, aqui? ” (“Here I have a/one butterﬂy. And here?”). The results showed
that the typically-developing MLU-matched children (Mean Age: 4:0) produced the

plural forms [5] and [es] on real word nouns on average about 97% of the time and the

83

typically-developing Age-matched children (Mean Age: 4:7) produced the plural forms
[5] and [es] on real words approximately 93% of the time. The SLI children only
produced the plural morpheme on average about 82% of the time. Cantt't-Sénchez and
Grinstead’s (2004) results are in line with those of Bedore and Leonard (2001), showing
by at least 4 years of age, Mexican Spanish-speaking typically-developing children
produce the plural morpheme systematically when describing plural sets.

The studies presented here show no difference between Spanish-speaking and
English-speaking children with respect to when they begin to produce the plural
morpheme. Although the age groups in each study differ, there is evidence that both
English—speaking and Spanish-speaking children start producing the plural morpheme by
at least 2:0 years of age and that by 4:0 years of age they are producing plural
morphology systematically. This indicates that redundancy in plural marking does not
facilitate (nor hinder) the acquisition of the production of plural morphology. Instead, it
appears that as long as the evidence for plural morphology is systematic in the adult
speech, English-speaking and Spanish-speaking children ( 1) start producing the plural
morpheme by at least 2:0 years of age, (2) use the plural morpheme productively on real
words by at least 4:0 years of age (most likely earlier) at least after quantiﬁers and
numerals and (3) at least English-speaking children (there are no studies for Spanish-
speaking children) use the plural morpheme to express the meaning of “more than one”

by 1:9 years of age.

3.1.2 Placement of the Plural Morpheme in Production
Given that previous studies suggest that Spanish-speaking children start to

produce the plural morpheme by two years of age, another important question is whether

84

Spanish-speaking children start out placing the plural morpheme on all elements within
the determiner phrase (i.e. the determiner, noun, and adjectives) or whether they start out
placing the plural morpheme on only one element within the determiner phrase.
Furthermore, if they place the plural morpheme on only one element within the
determiner phrase, where do they place it and is this placement systematic in the speech
of children?

Although there are very few studies examining where Spanish-speaking children
initially place the plural morpheme, there is some evidence that l:6-2:0 year old Spanish-
speaking children initially start out placing the plural morpheme only on the noun and
then by 2:1-2:6 they begin to extend the plural morpheme to the determiner, marking
both the determiner and the noun with the plural morpheme (Marrero and Aguirre 2003).

Similar ﬁndings were reported for children acquiring a dialect of Brazilian
Portuguese (BrP) in which in the researcher reports that in the adult speech the plural
morpheme is always placed on the determiner and sometimes omitted on the following
noun and adjectives (Vasconcellos Lopes 2006). Vasconcellos Lopes reports that two
Brazilian children at about 2:5 years of age produced the plural morpheme on the noun
and not the determiner in as many as 35% of all DPs. This ﬁnding is interesting given
that Vasconcellos Lopes also reports that in the adult speech the plural morpheme is
systematically placed on the determiner and optionally placed on the noun. Samples of

child speech taken from Vasconcellos Lopes is shown in (l 1).

85

(11) a. a hienas (Child Age: 2:4)
the.SG hyenas.PL

‘the hyenas’

b. meu chinelinhos (Child Age: 2:6)
my.SG slippers.PL

‘my slippers’

c. o patinhos (Child Age: 2:6)
the.SG duckies.PL

“the duckies”

According to Vasconcellos Lopes, the child utterances shown in (l l) are not attested in
the dialect of BrP adult speech these children were exposed to.

Cantﬁ-Sz’mchez and Grinstead’s (2004) had a second experiment that looked at
placement of the plural morpheme by 10 SLI and 20 typically-developing 4 year old (10
age-matched, Mean Age: 4:7 and 10 MLU matched, Mean Age: 4.0) Spanish-speaking
children from Mexico City. Children were shown a picture of a single object (e. g. a red
ﬂower) followed by a picture of multiple objects (e. g. several yellow ﬂowers). The
researcher would say, “Aqui hay unaflor roja. g Yaqui? ” (“Here there is a/one red
ﬂower. And here?”). The intention of the experiment was to elicit plural ‘N + Adj.’
constructions (e.g. ﬂares amarillas ‘yellow.PL ﬂowers.PL’). The results showed that

typically-developing Spanish-speaking Age-matched children and MLU-matched

86

children produced the plural morpheme on both the noun and adjective on average 98%
of the time. The SLI Spanish-speaking children produced the plural morpheme on both
the noun and adjective approximately 90% of the time.

It is difﬁcult to compare the ﬁndings of Cantr'r-Sénchez and Grinstead (2004) with
the ﬁndings of Vasconcellos Lopes (2006) and Marrero and Aguirre (2003) because of
the differences in age. It is most likely the case that the children in Cantﬁ-Sénchez and
Grinstead (2004) produced more plural marking because they were older. However, the
ﬁndings of these three studies suggest that (1) whether plural marking is systematic in
adult speech (as in Madrileﬁo adult Spanish) or whether it is variable in the adult speech,
with number marking always occurring on D and optionally on N (as in the variety of
BrP studied by Vasconcellos Lopes 2006), Spanish-speaking children may initially start
out placing the plural morpheme mainly on nouns and (2) by 4:0 years of age, Spanish-
speaking children acquiring a variety of Spanish with systematic plural marking (as in the
Spanish of Mexico City) have learned to produce the plural morpheme on all elements (or

at least more than one element) within the DP.

3.1.3 Comprehension of Plural Morphology

The following studies will present research on the comprehension of plural
morphology by English-speaking children where the plural morpheme is systematically
produced in the adult speech. As far as we know, there is only one experimental study of
the comprehension of plural morphology in Spanish-speaking children. However, we will
discuss this study later in the chapter because it was carried out in a variety of Spanish

where the plural morpheme is generally omitted in the adult speech.

87

In order to determine whether young children were sensitive to the syntactic
constraints of plural morphology, Gouvea et a1. (2005) examined whether 1:6 year old
English-speaking children were sensitive to ungrammaticalities involving the plural
morpheme. In a Headtum Preference Procedure children were tested on their preference
for passages containing the plural morpheme in grammatical noun phrases involving
deﬁnite determiners (e. g. the dogs, the babies) vs. passages containing the plural
morpheme in ungrammatical noun phrases involving indeﬁnite determiners (e. g. a dogs,
a babies). Their ﬁndings showed that children looked longer at grammatical passages
than at ungrammatical passages. Although this study does not tell us whether 1:6 year old
children associate meaning to the plural morpheme, it does indicate that by 1:6 years of
age, children are sensitive to the syntactic restrictions of the plural morpheme. In any
case, Gouvea et al. note that their ﬁndings must be interpreted with caution given the
experimental set up. In their study the ungrammatical passages always involved indeﬁnite
noun phrases while the grammatical passages always involved deﬁnite noun phrases. For
this reason, it is unclear whether children simply had a preference for sentences involving
deﬁnite noun phrases over indeﬁnite noun phrases.

Kouider et al. (2006) investigated the comprehension of the plural morpheme on
novel words in 1:6 — 3:0 year old English-speaking children using a Headtum Preference
Procedure where children listened to sentences involving plural and singular nouns and
were measured on how long they looked at pictures involving plural sets vs. singular sets.
The purpose of this experiment was to determine whether English-speaking children

associate the plural morpheme to the interpretation of ‘more than one’. Importantly, the

88

ages of the children in this experiment parallel the age at which children begin to produce
plural morphology in free speech.

In Kouider et al.’s ﬁrst experiment children were tested on sentences like Look,
there Qre some blickets vs. Look, there is a blicket, where the noun, determiner and verb
all provide number information. Children were presented with two screens, one
displaying a singular novel object and one displaying a plural set of novel objects, while
they listened to sentences. Their ﬁndings showed that by 2:0 years of age, but not 1:6
years of age, children looked at the screen that matched the target sentence for a longer
period of time. Because the determiner and verb also had number information, it is not
clear from this experiment whether 2:0 year old children associate the interpretation of
‘more than one’ with only the nominal plural morpheme. For this reason, in a second

experiment, children were tested on sentences like Look at the blickets vs. Look at th_e

 

blicket, where the only difference between the singular sentence and the plural sentence is
the presence of the nominal plural morpheme. If children treat these sentences differently,
associating the plural sentence with the plural picture, this would indicate that they
comprehend the plural morpheme as associated to the interpretation of ‘more than one’.
Their results showed that by 3:0 years of age, but not 2:0 years of age, children looked
longer at the screen that matched the target sentence.

While these two experimental studies do suggest that at least by 3:0 years of age,
English-speaking children associate the plural morpheme to the interpretation of ‘more
than one’, because of the difﬁculty of the task, they do not necessarily show that younger
children (children under 3:0 years of age) do not associate the plural morpheme to the

interpretation of ‘more than one’. The experimental task involved novel nouns and novel

89

pictures. It seems reasonable that children would look at either set of pictures given the
plural or singular noun. In fact, it seems that looking at the plural picture even in the
singular noun condition is a correct response since the child could be looking at only one
of the many novel objects presented in the plural picture. For this reason, it is very
interesting that 3:0 year old children did so well and suggests very strongly that at least
by 3:0 years of age, and possibly earlier, that English-speaking children associate the
plural morpheme to the interpretation of ‘more than one’.

It is important to note that the Kouider et al. (2006) study used an indirect testing
method in order to detect implicit understanding of the plural morpheme in children.
Children listened to sentences with plural and singular noun phrases and experimenters
measured where and how long they looked However, children were not given any
explicit instructions about the task. While this study shows that 3:0 years have an implicit
knowledge of the plural morpheme, it does not show whether they have explicit
knowledge of the plural morpheme. In other words, can children use the presence or
absence of the plural morpheme when making conscious decisions about number? The
only study that we are aware that tests explicit knowledge of the plural morpheme in
comprehension was carried out by Munn et al. (2006). In an Act-out task, 15 3:0-5:5
(Mean Age: 4:1) year old English-speaking children were tested on their comprehension
of the plural morpheme in deﬁnite noun phrases in sentences like Give me Mg next to

the barn vs. Give me the tags next to the barn in a context as in Figure 8.

Figure 8. Experimental Display in Munn et al. (2006).
FrogA FrogB FrogC F rog3 Frog2 Frogl E
I"Adapted from Munn et al. (2006).

 

 

 

 

90

Given the context in Figure 8, if children hear the sentence Give me the ﬂag next to the
barn with the singular noun phrase, they should choose F rogl. If they hear the sentence
Give me the frogs next to the barn with the plural noun phrase, they should choose at
least two or more of the frogs labeled with numerals, if we do not consider the role of the
deﬁnite. The second set of frogs (labeled with letters) is included in order to make the
sentence with the plural noun phrase felicitous. In other words, modiﬁcation of the plural
set is only necessary if there is an alternative set of ﬂags included in the display. Munn et
al. found that English-speaking children performed correctly on both the plural and
singular sentences about 80% of the time indicating that when plural morphology is
systematically marked in the adult speech, English-speaking children have explicit
knowledge of the plural morpheme as associated to an interpretation of ‘more than one’
by at least 4:0 years of age.

In addition to being associated to the interpretation of ‘more than one’, the plural
morpheme is also associated with count nouns but not mass nouns. From the studies
presented thus far it is unclear whether English-speaking children associate the plural
morpheme with the count-mass distinction. To test this, Bamer and Snedeker (2005)
examined 4:1 — 4:6 year old English-speaking children on their comprehension of
sentences like Who has more sting vs. Who has more m, in a context where one
character always has a single large object (e. g. a long string) while the other character has
three small objects (e. g. three short strings) of the same kind. The three objects had a
smaller combined volume and surface area than the large object. The only difference
between the mass vs. count nouns was the presence or absence of the plural morpheme.

The ﬁndings showed that children treated the plural count nouns differently from the

91

mass nouns, indicating that in addition to the ‘more than one’ interpretation, by 4:0 years
of age English-speaking children also associate the plural morpheme only to count nouns.
In summary, the ﬁndings of the above studies examining the comprehension of
plural morphology in English-speaking children indicate that children associate the plural
morpheme to the interpretation of ‘more than one’ by at least 3:0 years of age (implicitly)
and that, furthermore, they associate the plural morpheme to count nouns, and not mass

nouns, at least by 4:0 years of age.

3.2 Acquisition of Variable Input

Up to this point, the research presented in this chapter indicates that when plural
morphology is systematically produced in the adult speech, both Spanish and English-
speaking children begin producing the plural morpheme by at least 2:0 years of age and
English-speaking children (for Spanish-speaking children see Chapter 5) appear to
associate the plural morpheme to the interpretation of ‘more than one’ as early as 3 years
of age. In this section, I will ﬁrst discuss acquisition of the plural morpheme when plural
morphology is variable in the adult speech. Next, I will discuss research that has
examined the development of other grammatical and phonological features when the
adult input is variable. Finally, I will present research that has examined the effect of
inconsistent input on language acquisition and discuss the differences between ﬁrst
language acquisition of inconsistent input vs. ﬁrst language acquisition of variable input.

In addition to examining plural morpheme production in two Madrileﬂo children,
as noted above, Marrero and Aguirre (2003) also examined the production of plural
morphology in one Canary Island Spanish-speaking child and found that unlike the

Madrileﬁo Spanish-speaking children, the Canary Island child did not start producing

92

plural morphology until 3:0 years of age and, at that time, still omitted the plural
morpheme about 20% of the time. They suggested that this delay and subsequent
omission of plural morphology might be due to the aspiration of the plural morpheme in
dialects of Canary Island Spanish: however, they provided no details about the linguistic
input this particular child was exposed to. This ﬁnding, although based on only one child,
is comparable to those ﬁndings reported in Vivas (1979) above, where a 2:5 year old
child still omitted the plural morpheme approximately 50% of the time and a 3:5 year old
child omitted the plural morpheme approximately 20% of the time. It is interesting that,
while neither study examined the input these children were exposed to, both studies
suggested that the later development of plural morphology in these children might have
been related to syllable-ﬁnal /s/ lenition in the dialects the children were exposed to.

There is one study that examines the comprehension of plural morphology in
Spanish-speaking children ﬁ'om the Dominican Republic, where syllable-ﬁnal /s/ is often
omitted (Terrell 1981, Morgan 1998, Bradley 2006). Pérez-Leroux (2005) used a Picture
Matching Task to test 23 3:0-6:0 year old Dominican Spanish-speaking children on their
comprehension of verbal and nominal morphology. Children were shown sets of pictures
(as in Figure 9) and read sentences like (12) and (13). Children heard sentences read in
the natural Dominican Speech, which included aspiration of the plural morpheme (this
was conﬁrmed through personal correspondence with the author), and were asked to

choose the corresponding picture.

93

Figure 9. Experimental Display in Pérez-Leroux (2005).

 

 

 

 

 

 

 

 

 

 

 

 

 

*Adapted from Pérez-Leroux (2005).

(12) El gato duerme en la carna.
The.SG cat.SG sleeps.3.SG on the bed

‘The cat is sleeping on the bed.’

(13) Los gatos duermen en la carna.
The.PL cats.PL sleep.3.PL on the bed

‘The cats are sleeping on the bed.’

Importantly for the purposes of this dissertation, the ﬁndings of Pérez-Leroux showed
that Dominican Spanish-speaking children between the ages of 3:2—4:5 only assigned a
plural interpretation to plural defmites 45% of the time and that they assigned an
incorrect plural interpretation to the singular defrnites 48% of the time. The older children
between 4:8-6:6 years of age assigned a plural reading to the deﬁnite plural 79% of the
time but they also incorrectly assigned a plural reading to the singular deﬁnite 33% of the
time. These ﬁndings are interesting if we compare them to those presented for English-

speaking children who receive an input with systematic plural marking in Munn et al.

94

2006. As noted above, Munn et al. found that English-speaking children associated plural
defmites to an interpretation of ‘more than one’ 81% of the time and only incorrectly
associated it to a interpretation of ‘one’ 17% of the time. While the English-speaking
children clearly outperformed the younger Dominican Spanish-speaking child group, they
perform only slightly better than the older Dominican Spanish—speaking child group, even
though the older Dominican children were slightly older than the English-speaking
children.

The differences in results between the Spanish-speaking and English-speaking
children is even more interesting when we take into consideration the fact that the tasks
differed for the Dominican Spanish-speaking children and the English-speaking children
in that the English-speaking children only had the plural morpheme to rely on for their
interpretation (e. g. the cow vs. the cows) while the Dominican Spanish-speaking children
had the plural morpheme, the form of the deﬁnite determiner, which was different in the
plural vs. singular condition (e.g. lgs gatos ‘the.PL cats.PL’ vs. _el gato ‘the.SG cat.SG’) and
subject-verb agreement. For this reason, it is quite astonishing that Dominican Spanish-
speaking children performed so poorly and it suggests that the variability of plural
morphology production by adult speakers may have an effect on Dominican children’s
comprehension of the plural morpheme. Note, however, that this comparison must be
interpreted with caution as the experimental tasks were different and the social class and
educational experiences of the children were not controlled for across the two languages.

Turning to varieties of American English, it has been reported that the plural
morpheme is sometimes omitted on plural nouns in African American Vernacular

English (AAVE) adult speech (Wolfram 1969, Fasold and Wolﬁam 1970, Moore 1979).

95

In an acquisition study, Rarner and Rees (1973) compared the production of plural
morphology by AAVE-speaking children vs. Mainstream English-speaking children
(MAE) and found that AAVE-speaking children and MAE-speaking children differed in
their production of plural morphology after quantiﬁers, with AAVE—speaking children
showing variable behavior. While this study shows differences in production, it does not
show whether the variable input that AAVE-speaking children are exposed to affects
comprehension of plural morphology.

Importantly, we want to know whether variable input also affects comprehension.
Moore (1979) examined the comprehension and production of plural morphology by
AAVE-speaking children by testing 100 4:6 — 14:6-year-old children from inner city
Detroit, Michigan who came from working-class backgrounds. Moore notes that Detroit
speakers of AAVE omit the plural morpheme when the noun is preceded by plural
quantiﬁers and that the frequency of plural morpheme omission varied across social
classes with the working-class speakers showing the highest percentage of plural
morpheme omission (see Wolfram 1969, Fasold and Wolfram, 1970). In a sentence
repetition task involving novel words, Moore found that 4:6 -— 7:4 year old AAVE
speaking children omitted the plural morpheme between 50 — 55% of the time. In an
Berko style elicitation task testing whether children produce plural nouns headed by
quantiﬁers, Moore found that 4:6 — 7:4 year old AAVE-speaking children only produced
plural nouns approximately 67% of the time for real words and 35% of the time for novel
words (see also Rarner and Rees 1973 for similar results). On a comprehension picture-

matching task, Moore found that 4:6 —— 7:4 year old AAVE-speaking children only

96

comprehended the plural morpheme approximately 55% of the time. That percentage
jumped to 80% by about 8:0 years of age.

Another feature of AAVE that appears to have a variable behavior is the third
person singular morpheme —s on English verbs. It has been reported that the third person
singular morpheme is frequently omitted in the speech of AAVE-speaking adults, being
omitted as much as 85% (see Wolﬁ'am and Schilling-Estes 1998) while it is not omitted
in the speech of MAE-speaking adults. Likewise for child speakers, Washington and
Craig (1994) showed that the 3rd person singular morpheme was omitted approximately
70% of the time in the speech of 5:0 year old AAVE-speaking children.

Johnson (2005) examined comprehension of the 3rd person singular morpheme in
MAE-speaking children vs. AAVE-speaking children by testing whether they could
indirectly assign number to the subject based on the presence or absence of the 3rd person
singular morpheme on the verb. Thirty working class 4 — 6 year old AAVE-speaking
children ﬁom an urban area of Connecticut and sixty-two middle-class 3—6 year old
Mainstream English-speaking children from a small town in western Massachusetts
participated in the study. A Picture Matching Task was used. Children were shown two
pictures (Picture 1: one cat sleeping on a bed vs. Picture 2: two cats sleeping on the bed)
and were asked to show the researcher the picture where (l) the cat sleeps on the bed or
(2) the cats sleep on the bed. The target sentences mask the plural morpheme on the noun
because the following verb begins with an alveolar ﬁicative (the catsleep/catsleeps on the
bed). For this reason, only the presence or absence of the third person singular /s/
provides number information with respect to the subject. The ﬁndings of these studies

showed that while 5-6 year old MAE-speaking children (but not 3-4 year old MAE-

97

speaking children) performed signiﬁcantly different from chance in their interpretation of
the third person singular morpheme, AAVE-speaking children of the same age did not.
These ﬁndings from Moore (1979) and Johnson (2005) suggest that variable input seems
to cause a delay in the development of this grammatical morphology in comprehension.

Kovac and Adamson (1981) examined ﬁnite ‘be’ usage in 3 — 7 year old MAE-
and AAVE-speaking children from Washington, DC. Finite ‘be’ is sometimes produced
and sometimes omitted in AAVE: ‘she’s/is jumping’ vs. ‘she jumping’ but not in the
speech of MAE-speaking adults (always produced: ‘she’s/is jumping’). Children
provided free speech samples through picture description and role-play activities and
through researcher-child interviews. Their results showed differences in development of
ﬁnite ‘be’ usage between AAVE- and MAE- speaking children. While 3 — 5 year old
working-class AAVE-speaking children showed steady increases in ﬁnite ‘be’ omission,
3 — 5 year old working-class MAE-speaking children showed steady decreases in ﬁnite
‘be’ deletion with steady increases in contraction. Even more interesting, however, is that
while omission of ﬁnite ‘be’ is syntactically and phonologically constrained in the speech
of adult AAVE-speakers, AAVE-speaking children who omitted ﬁnite ‘be’ in their own
speech, had not yet acquired all of the various grammatical constraints on ﬁnite ‘be’
omission, even at 7 years of age. In other words, AAVE-speaking children were variable
in their omission of ﬁnite ‘be’ but that variability was not governed by the same
constraints governing the adult variability. Speciﬁcally, seven year old working-class and
middle-class AAVE-speaking children differed ﬁ'om adults in that children omitted ﬁnite
‘be’ more often before predicative adjectives, as in (14), than when it occurred as an

auxiliary, as in (15) and (16).

98

(14) a. She’s/ is pretty.
b. She pretty.

(15) a. She’s/ is eating.
b. She eating.

(16) a. She’s/ is gonna eat.
b. She gonna eat.

Kovac and Adamson conclude that “children who have a non-standard model and who
may be expected to have variable features in their adult speech may have the feature by
age 5, even though the various constraints on the behavior of that feature may not be in
order until considerably later” (p. 409).

Finally, Gathercole (1986) examined the production of the present perfect in the
free speech of 4 — 6 year old Scottish English-speaking and American English-speaking
children and adults. While Scottish and American adults both produce present perfect
constructions, the degree to which they use the present perfect varies. Their results
showed that although the distribution of the present perfect was similar between the two
adult groups (used in a variety of semantic contexts), Scottish adults used the present
perfect approximately 5 times as much as American adults, with American adults often
preferring the simple past tense in the same contexts in which Scottish adults prefer the

present perfect. Differences were also found between children. While Scottish children

99

produced the present perfect (excluding ‘have got’) 75 times in 6 hours of recorded
speech, American children (one child) only produced the present perfect once. In other
words, American English-speaking children almost never produced the present perfect
construction. Gathercole concluded that frequency is a major'contributor to the
differences in present perfect production found between Scottish and American children.
However, she does not address directly the issue of variability in the input presented to
American children, an input where the present perfect and simple past are used in the
same contexts to varying degrees. Although American adults produce both forms,
American children appear to have not yet constructed a grammar with the present perfect
construction. Further evidence for this is provided by the fact that in Gathercole’s study
American children often “incorrectly” used the simple past where the present perfect
would have been more appropriate in the adult grammar (Adult: “How long have you
been dancing? ” Child: “F or 50 years I did it. ”, Adult: “You ’ve been dancing for 12

years? ” Child: “I really danced for 300 years. ") and American children often treat ‘got’

 

of the construction ‘have got’ as a verb by placing verbal morphology on it (Child: “She
8.0L? 0 new baby ” for She has got a new baby). It appears that the variability in the input
to American children (between the use of the present perfect and simple past forms in the
same contexts) causes American children to initially rely on one form, the simple past
(which is the more frequent form). I believe that the American English-speaking children
avoided the present perfect in their own speech, not only because the simple past is more
frequent than the present perfect in the input, but more importantly because both
constructions overlap in the contexts in which they are uttered in the adult speech. While

this study does not examine the acquisition of a form that is omitted, rather the

100

acquisition of two structures that may be in competition, it does show that American
English-speaking children did not pattern with American English-speaking adults, rather
the children consistently chose the simple past even when it was ungrammatical. This is
interesting because it suggests that the American English-speaking children were
acquiring a grammar that was different from the adults.

Most of the studies reported above on the acquisition of variable input suggest
that children exposed to variable input are variable in their own production although they
may not necessarily have acquired all of the constraints that govern that variability.
However, note that Gathercole’s (1986) study suggests that children regularize variable
input, initially only acquiring one of two forms that are in competition. In addition, these
studies suggest that variable input involving omission of a particular form, appears to
result in a delay in comprehension of the form. In general, these ﬁndings on the
acquisition of variable input are consistent with the Variability Delay Hypothesis, which
was presented in Chapter 1.

In addition to the research discussed above, there have been other investigations
on the acquisition of variable input, where the variable input does not affect grammatical
features in the language. Smith et a1. (2006) examines the acquisition of phonological and
morphological variation in Scottish children living in Buckie, Scotland. Eleven children
(2:10-3:6) and their parents were recorded and the data was examined for the
phonological variation between the use of the standard diphthong N‘ '/ vs. the local (non-
standard) monophthong /u:/ and also for morphological variation with respect to the non-

standard local use of the 3rd person singular verb morpheme with third person plural

101

subjects (‘My trousers is falling down’ (non-standard) vs. ‘My geysers are falling down’

 

(standard)).

The ﬁndings of this study revealed different degrees of variation in adult speech
depending on whether the adult was speaking with other adults vs. when they were
speaking with their own children. In addition, Smith et al. found that patterns in child
acquisition of phonological variation differed from patterns in child acquisition of
morphological variation. For acquisition of phonological variation, their ﬁndings showed
that while adult-to-adult speech revealed near categorical use of the local nonstandard
form, the adult-to-child speech only contained 43% use of the local nonstandard form and
children’s speech matched closely to the adult speech (child speech contained 37% use of
the local form). In other words, caregiver speech was variable between the local and
standard forms and children were also producing both forms. However, for the
morphological variation involving the 3rd person singular form of the verb, no such
difference was found between adult-to-adult vs. adult-to-child speech. In'fact, in this
case, adults used more nonstandard forms with their own children than with other adults
and, as with the phonological variation, children showed variable use of the third person
singular morpheme as well. One important ﬁnding, with respect to the linguistic and
extra-linguistics constraints that govern the variability of each variant, is that, while
children seemed to understand the social implications of the phonological variants, they
did not appear to understand the social constraints governing the morphological variants.
On the other hand, they did seem to understand a number of grammatical constraints with

respect to the production of the 3rd person singular morpheme.

102

Roberts (1994, 1997) examined (t/d) deletion and the production of (ing) ([irj] vs.
[m]) in English-speaking children in order to determine whether children have mastered
both the linguistic and extra-linguistic constraints related to the deletion of these word-
ﬁnal sounds. (-t/d) deletion comes about as a result of the reduction of consonant clusters
where words like ‘cold’ are produced as [kol], where [d] has been omitted. The
progressive marker —ing is also reduced in English adult speech to [m], where words like
‘kicking’ are pronounced as [klkln]. Both phenomena are fairly widespread in English
adult speech and have a variable behavior, with reduction occurring more in certain
grammatical and phonological contexts and being more prevalent in adult working-class
speech and adult male speech. Roberts recorded 17 3 — 4 year old English-speaking
children from South Philadelphia working-class to lower middle-class children in order to
determine whether reduction in child language followed the linguistic (phonological,
grammatical) and extra-linguistic (social) constraints as those in the adult language.
Similar to Smith et al. (2006), Roberts found that children acquired the linguistic
constraints governing the variability before they acquired the extra-linguistic constraints.
Taken together, the results from Roberts and Smith et al. suggest that (1) children acquire
linguistic constraints for variable rules before extra-linguistic constraints, (2) that
phonological variable rules may be acquired before morpho-syntactic variable rules and
(3) adult speech varies in the use of certain variable forms depending on whether they are
talking to adults or to their own children.

We will next discuss the acquisition of inconsistent input by children. Inconsistent
input is different from variable input in that variable input is constrained by linguistic and

extra-linguistic factors, and therefore is predictable, while inconsistent input is not

103

constrained linguistically nor extra-linguistically and is, therefore, not predictable.
Inconsistent input is found in the speech of non-native speakers of a language to their
children. Singleton and Newport (2004) hypothesized that children exposed to
inconsistent input would, nevertheless, be consistent in their own production. The basis
of this hypothesis arises from research showing that children exposed to Creole languages
spoken as a second language by parents who provide inconsistent input, nevertheless
come to build a grammar that does not have any inconsistencies in grammatical marking.
Singleton and Newport (2004) investigated how an ASL-signing child (Simon) acquired
verbal morphology in ASL when the input of his parents was inconsistent with respect to
these morphemes because they were late learners of ASL (they learned ASL in their
teens). While they found that the non-native ASL-signing parents differed from native
ASL-signing adults, their child Simon did not differ from native signers in his production
of these morphemes, and in most cases Simon outperformed his parents, which indicates
that Simon was regularizing the inconsistencies in the ASL he was acquiring.

Hudson Kam and Newport (2005) investigated language acquisition in the
presence of consistent vs. inconsistent input by presenting English-speaking adults and
children Mean Age: 6:4) with an artiﬁcial language (matched to real world objects) that
contained either consistent or inconsistent input with respect to the determiner system.
The language contained 17 words: 4 verbs, l2 nouns and 1 determiner. Children were
ﬁrst explicitly taught all of the words of the artiﬁcial language, except for the
determiners, through translation. In other words, the experimenter said, “if you want to
say ‘hit’ in Sillyspeak (the artiﬁcial language), you say /ﬂIm/”. Next, children were

presented with scenes followed by sentences in the artiﬁcial language describing the

104

scene. All subjects were exposed to the same basic input sentences. There were two
experimental conditions and the input sentences differed across conditions only in the use
of determiners. In the consistent condition, subjects heard the determiner produced with
nouns 100% of the time, while, in the inconsistent condition, the determiner was present
with nouns only 60% of the time. In the inconsistent condition there were no constraints
governing the presence or absence of the determiner so there were no patterns for
children or adults to learn.

After being taught the vocabulary of the artiﬁcial language, subjects were played
the artiﬁcial language in 7 sessions over the period of 9 days and then tested on their
acquisition of the determiner system through an elicitation task and a grammaticality
judgment task. I will only discuss the elicitation task here. The elicitation task consisted
in a translation task in which subjects were asked to translate sentences in English into
the artiﬁcial language. For example, child and adult subjects were told, “I want you to tell
me how to say ‘the bear moves’ in Sillyspeak” (p. 177). Based on the results of the
elicitation task, subjects were grouped into three categories: Systematic Users, Systematic
Nonusers, and Variable Users.l3 Systematic Users refers to child and adult subjects who
systematically produced determiners in the elicitation task. Systematic Nonusers were
subjects who systematically omitted the determiner and Variable Users were subjects
who sometimes used and sometimes omitted the determiner in the elicitation task. The

results of the elicitation task are shown in Table 12.

 

’3 There was one additional category, “Systematic Other”, but only one child fell into this
group and so it will not be discussed here.

105

Table 12. Percentage of Systematic Responses in Hudson Kam and Newport (2005).

 

Input Type Systematic User Systematic Nonuser Variable User

 

Children

100% 50% 25% 12.5%
(4/8) (2/8) (1/8)

60% 14.3% 57% 28.6%
(1/7) (4/7) (2/7)

Adults

100% 100% O 0
(4/4)

60% O 50% 50%

(2/4) (2/4)

 

*Adapted ﬁ'om Hudson Karn and Newport (2005).

The results of their study showed that of the 8 children exposed to consistent input
(determiner produced 100% of the time), 4 of them systematically produced determiners
in the elicitation task, 2 of them systematically omitted the determiner and l was variable
in his own production of determiners (sometimes omitting and sometimes producing the
determiner). Of the 7 children exposed to inconsistent input (determiner produced 60% of
the time), 1 of them systematically produced the determiner in their own speech, 4
systematically omitted it, and 2 were variable in their production of the determiner.

The results suggest that children exposed to inconsistent input are generally

systematic in their own production. In other words, the production of the children did not

106

match the input to which they were exposed. However, note that there were 2 children
who were variable users. In other words, the production of these two children did match
the input to some extent. If we are to generalize these results to the larger population, the
results would suggest that, in general, inconsistent input results in systematic production,
as approximately 70% (5 of 7 children) of the children were systematic in their
production. This ﬁnding is consistent with Singleton and Newport (2004). While this
generalization would not contradict previous ﬁndings that children exposed to variable
input are variable, given that variable input is different from inconsistent input, we must
still be cautious in making this generalization for a number of reasons.

First of all, both the language and the setting in which the language was learned
and used was artiﬁcial. Secondly, it is not clear what we can make of the results for the
children exposed to inconsistent input when we take into consideration the results for the
children exposed to consistent input, where the determiner was always present in the
input. In such a context, we would expect that 100% of children would be systematic
users. However, the results show that only 50% (4 of 8 children) of them were systematic
users, while 25% (2 of 8 children) of them always omitted the determiner and 12.5% (1
of 8 children) were variable users. There are at least two possible reasons why children
exposed to an input with 100% determiner usage would not produce the determiner 100%
of the time in their own production, either (1) they had difﬁculties with the experimental
task or (2) there are differences in language learning strategies between children, if we
assume that the task models natural language learning. In any case, once we take into
consideration the ﬁnding that even children exposed to consistent input (100%) are

unable to acquire the determiner in their own production, the response patterns of

107

children exposed to inconsistent input becomes less clear. Are systematic non-users in the
inconsistent condition non-users because (1) of the input they were exposed to, (2)
because they had difﬁculty with the task, or (3) because of differences in learning
strategies compared to the other children in the inconsistent condition? Nevertheless,
taken together with the ﬁndings from Singleton and Newport (2004), it appears to me that
the results of Hudson Kam and Newport (2005) reveal that inconsistent input does result
in systematic use or non-uses by children and it may just be the case that some children
had difﬁculty with the experimental task. The Hudson Kam and Newport (2005) study is
unique in that it provides a very nice way to manipulate the input (making it consistent or
inconsistent) in order to see how children deal with such input. This is important for the
present study as we ﬁrst examine the type of input Chilean and Mexican children are
exposed to and then determine how that input affects their comprehension of plural

morphology.

3.3 Summary of Previous Research

Acquisition of Plural Morphology:

1. When plural morphology is systematically produced in the input, Spanish-
speaking and English-speaking children start producing the plural morpheme as early as
2:0 years of age.

2. When plural morphology is systematically produced in the input, English-
speaking children associate the plural morpheme to an interpretation of ‘more than one’
as early as 3:0 years of age in experimental studies investigating implicit knowledge of
the plural morpheme and at least by 4:0 years of age in experimental studies investigating

explicit knowledge of the plural morpheme.

108

3. Spanish-speaking and English-speaking children between 4:0-5:0 years of age
do not use the Spanish allomorph [es] or the English allomorph [os] productively (with
novel words).

4. It appears that Spanish-speaking children initially place the plural morpheme
on nouns and then extend marking to the determiner and adj ectives: however, further
research is needed here.

Acquisition of Variable Input:

1. It appears that Spanish-speaking children exposed to an input with syllable-
ﬁnal /s/ lenition may start producing the plural morpheme later than those exposed to
varieties of Spanish with no syllable ﬁnal /s/ lenition. However, the evidence is rather
weak given that this late production was observed in only 3 children.

2. It appears that variable input that involves omission of a grammatical
morpheme delays acquisition of that morpheme.

3. Children exposed to variable input appear to be variable in their own
production although they may have not yet acquired all of the various linguistic and
extra-linguistic constraints governing that variability by 6 years of age.

4. Children exposed to inconsistent input appear to pattern differently from
children exposed to. variable input. Inconsistent input appears to result in systematic use
or non-use by children (and also possiblty in variable production by children), while

variable input only seems to result in variable production by children.

109

CHAPTER 4

PRODUCTION OF PLURAL MORPHOLOGY

4.0 Introduction

The purpose of this chapter is to present production data that was collected from
Chilean and Mexican children and adults in order to compare the linguistic input to
Chilean vs. Mexican children and to determine how the linguistic input may be related to
how children perform on the comprehension tasks presented in Chapter 5. Chapter 4 has
three main goals. The ﬁrst goal is to conﬁrm previous ﬁndings on the variability of plural
morpheme production by Chilean Spanish-speaking adults reported in Cepeda (1995) and
presented in Chapter 2 of this dissertation and, furthermore, to conﬁrm reports that
Mexican Spanish-speaking adults (of Mexico City) consistently produce the plural
morpheme. The second goal is to compare the production of the plural morpheme
between children vs. their parents. In other words, will Chilean and Mexican children
pattern with adults in their production of the plural morpheme or will there be
developmental differences between child vs. adult groups? Marrero and Aguirre (2003)
found that 3-year-old Spanish-speaking children omitted the plural morpheme on some
elements within the noun phrase. It may be the case that Mexican and Chilean children at
4 years of age still omit the plural morpheme some of the time and that patterns of
omission between the two child groups are similar to each other, yet different from the
adult groups. On the other hand, there may be no differences between 4-year-old children
and adults in their production of plural morphology: instead, we may ﬁnd that by 4 years
of age child production of plural morphology is adult-like. The third goal is to compare

the production of plural morphology between Mexican children and adults vs. Chilean

110

children and adults. Here, there are at least two possible alternatives for the production of
plural morphology in the child groups. The ﬁrst alternative is that if 4-year-old children
pattern with adults, then, while Mexican children would be systematic in their production
of plural morphology, Chilean children would be variable and their variability would be
similar to that found in the Chilean adult speakers. This is consistent with studies
showing that children exposed to variable input are variable in their own production
(Kovac and Adamson 1981, Labov 1989, Roberts 1994, 1997, Smith et al. 2006).
However, if Chilean Spanish adult speakers sometimes omit and sometimes produce the
plural morpheme, creating somewhat of an ambiguous input with respect to plural
marking, it may be that Chilean children are unable to acquire such variability. Hence, a
second alternative is that Chilean children, like Mexican children, would also be
systematic in their production of plural morphology, either always omitting the plural
morpheme or always producing it. This is consistent with the ﬁndings of Hudson Kam
and Newport (2005) of children exposed to inconsistent input produced by adult non-
native speakers to their children.

Based on previous research on the production of variable input, which was
presented in Chapter 3, this dissertation predicts that while Mexican children will always
produce the plural morpheme, Chilean children will be variable in their production of the
plural morpheme, sometimes producing it and sometimes omitting it. In addition, Chilean
children will differ from Chilean adults in the constraints governing their variable
production. However, it is important to note that although Chilean children may produce
the plural morpheme some of the time, it does not necessarily mean that they have

associated the plural morpheme to an interpretation of ‘more than one’. We know from

111

previous research that children produce certain grammatical morphemes before they are
able to associate those morphemes to an interpretation (Johnson et al. 2005, Pe’rez-Leroux
2005). To ensure that production of the plural morpheme indicates that the child
associates the plural morpheme to an interpretation of ‘more than one’, the child would
have to provide plural-singular minimal pairs when naming plural vs. singular sets. Such

a study will be presented in Chapter 5.

4.1 Method and Design

A Free Speech task, a Repetition Task and a Naming Task were used to collect
production data ﬁ'om children and adults. The Free Speech Task consisted of collecting
approximately 5-10 minutes of ﬁ'ee speech data from child and adult subjects while they
talked about topics of interest (e. g. children: cartoons, play-time activities: adults: family,
work, and their children). In the Repetition Task subjects were presented with pictures
and were asked to repeat sentences that the researcher read about each picture. Subjects
only listened to the sentences. They did not see the written version. The "research
assistants produced the plural morpheme as [s] on all plural lexical items for both
Mexican and Chilean subjects. All sentences involved plural or singular complex
indeﬁnite noun phrases and bare noun phrases, as in (1) and (2) below. The full set of

sentences is shown in Appendix A.
(1) Unos bomberos estan comiendo manzanas.

Some.PL ﬁremen.PL are.3.PL eating apples.PL

‘Some ﬁremen are eating apples.’

112

(2) Un bombero esta comiendo una manzana.
A/One.SG ﬁreman.SG is.3.SG eating an/one.SG apple.SG

‘A/One ﬁreman is eating an/one apple.’

The phonological environment of syllable ﬁnal /s/ was controlled for in the Repetition
Task. Cepeda (1995) found that ﬁnal /s/ lenition was affected by the initial sound of the
following word. Table 13 shows the distribution of syllable ﬁnal /s/ lenition reported in
Cepeda (1995). For this reason, the phonological environment in the Repetition Task was
controlled for in such a way that only environments where syllable ﬁnal /s/ omission and
production were frequent were included. In this way, we provided an environment where
Chilean subjects could omit the plural morpheme in order to see whether they would. In
the Repetition Task, the plural morpheme was always preceded by words whose initial
sounds were a bilabial ﬁicative( [B], 4 tokens), a bilabial nasal ([m], 4 tokens), an
unstressed vowel ([e], 8 tokens), or a sentence ﬁnal pause (6 tokens). These phonological

environments are shaded in Table 13.

Table 13. /s/ Lenition by Phonological Environment in Chilean Spanish.
Plural _#C _#C _#C _#V _#V _#

 

Variant [-cont] [+cont] [+nas] (Unstressed) (Stressed) (Pause)

 

[s] 1% 1% 1% 6% 25% 12%
[h] 83% 55%. 71% 42% 38% 35%
Omission l 6% 44% 28% 52% 36% 54%
Total 3594 3552 1406 2716 1018 3831

 

 

*Adapted from Cepeda (1995).

113

The type of noun phrase in the Repetition Task was also controlled for. With
respect to the determiner in the noun phrase, Cepeda (1995) found that the plural
morpheme was omitted about 34% of the time on polysyllabic determiners but only 13%
of the time on monosyllabic determiners. In addition Cepeda-also found that the plural
morpheme was omitted more often on plural nouns headed by a determiner than on bare
plural nouns. In order to provide an environment where plural morpheme omission was
possible on the determiner, the Repetition Task included only indeﬁnite determiners
(unos ‘some.PL’ and algunos ‘some.PL’). The experimental sentences contained four
tokens of the plural unos (2 masculine and 2 feminine) and four tokens of the plural
algunos (2 masculine and 2 feminine). The singular forms of these indeﬁnites (un/una
alguna/alguna) were included as controls. In addition, ﬁve bare plural noun phrases
were included to allow for comparison with nouns headed by determiners. The indeﬁnite
noun phrases always occurred in subject position where verbal agreement was also an
indicator of nominal number while the bare plural noun phrases always occurred in object
position because Spanish bare nouns are restricted to object position. For this reason, the
plural morpheme on the bare plural nouns in the Repetition Task was the only indicator
of number”.

Finally, in the Naming Task subjects were shown sets of miniature toys and asked
to name the toys. The goal was to elicit bare plurals. The experimental question was
always gQué son? (“What are.PL they?”), which includes the copular verb son

(‘are.3.PL’) that is grammatically plural in order to ensure that plural nouns would be

 

‘4 The fact that the subject noun phrases agree with the verb in number while the object
noun phrases do not may cause subjects to omit more often on the subject than on the
object, because there is additional number information for the subject.

114

elicited from subjects. The miniature toys included: vacas [ﬂakas] (‘cows’), perros
[pel‘os] (‘dogs’), muﬁecas [muflekas] (‘dolls’), la'pices [lapises] (‘pencils’), copas [kopas]
(‘cups’), autos [autos] (‘cars’), arar’ias [araﬁas] (‘spiders’), barcos [Barkos] (‘ships’),
peces [peses] (‘ﬁsh’), bolitas [Bolitas] (‘marbles’) and monos [monos] (‘monkeys’).
Given that Chilean subjects always produced bare nouns in this task, the initial sound of

the noun would have no effect on the omission of the plural morpheme.

4.2 Subjects

16 (4:5-5:l 1: Mean 5:2) ChWC children, 8 (4:8-5:11: Mean 5:3) ChMC children,
3 ChWC mothers, 4 ChMC mothers, 8 (4:8-5:9: Mean 5:4) MexWC children, 2 MexWC
parents and 2 MexMC adults (MexMC adults were parents but not parents of the
Mexican children who participated in the study) participated in the study. There are more
ChWC children than ChMC children or MexWC children because not all ChWC children

were asked to participate in the Free Speech task.

4.3 Procedure

Subjects were tested in the school or in their own homes by a native Spanish-
speaking research assistant who was ﬁom the same local area as the subjects. During the
Free Speech task, the native-speaking research assistant used colloquial forms of the
plural morpheme in their own speech. In other words, the Chilean research assistants
used mainly [s], [h] and 6 while the Mexican research assistants used [3] and [2].
However, during the Repetition Task, both the Chilean and Mexican research assistants
consistently pronounced the plural morpheme as an alveolar ﬁicative. Given that in
Chilean Spanish [h] and [zero] are the results of syllable ﬁnal /s/ lenition, having the

Chilean researchers pronounce the plural as [s] is providing subjects with the underlying

115

form. The Free Speech Task was administered ﬁrst, followed by the Naming Task and
then the Repetition Task. The Repetition Task was carried out last so as not to prime the
use of the plural or any of its variant forms in subsequent tasks.

All data was recorded by a Sony Minidisc Digital Recorder (MZ-R70) with a
standard Sony clip-on microphone or a Marantz PMD 222 recorder with a Shure
miniature clip-on cardioid condenser microphone. Data was transcribed by research
assistants who were native-speakers of Chilean or Mexican Spanish. After transcriptions
were ﬁnished, all plural lexical items (e. g. nouns, determiners, adjectives) were coded for
the pronunciation of the plural morpheme as [s], [h] or [zero]. All Chilean data was coded
for pronunciation of the plural morpheme by two different Chilean Spanish-speaking
research assistants. Issues related to reliability in coding between research assistants is
discussed in Section 4.5. The Mexican data showed no variation in the pronunciation of
the plural morpheme; it was pronounced as an alveolar fricative by both children and

adults. For this reason, only one Mexican research assistant coded the Mexican data.

4.4 Analysis

The following analyses were carried out First, the Mexican and Chilean data
were compared to each other. Second, because the Chilean data showed variable behavior
in the production of the plural morpheme, a variable rule analysis was carried out using
GoldVarb 2.0 (Rand and Sankoff 2001). GoldVarb 2.0 is a multivariate analysis program
which uses a binary logistic regression as a statistical model for variation analysis. An
analysis by logistic regression is a way of using quantitative data to make predictions
about possible relationships among different independent variables or factor groups

(linguistic and extra-linguistic) that the researcher is investigating. Goldvarb 2.0 uses a

116

stepwise regression which considers incrementally complex models by adding factor
groups and comparing the new model to previous ones until it ﬁnds the model that best
describes the data (Paolillo 2002). Because the Mexican data showed no variability in
plural morpheme pronunciation, no further analyses of the distribution of Mexican
pronunciation of the plural morpheme could be carried out. The distribution for Mexican
adult and child subjects was categorical.

The dependent variable was the omission of the plural morpheme. If Chilean
subjects produced [5] or [b], they were scored as producing the plural morpheme. If
subjects did not produce [5] or [h], they were scored as omitting the plural morpheme.
Our coders noted that adult speakers sometimes produced an unreleased glottal stop as
the plural morpheme but it was very difﬁcult to distinguish these unreleased glottal stops
from [h]. For that reason, any coding of the plural morpheme as an unreleased glottal stop
was counted as [h] in the analysis. Cepeda (1995), the only study that I am aware of on
Chilean syllable ﬁnal /s/ lenition, did not document the use of the unreleased glottal stop.
In future studies, a more detailed investigation will be needed to determine the frequency
in which unreleased glottal stops are used by Chilean speakers.

Plural morphemes followed by words beginning with [s], [h] and/or [x] were not
included in the data analysis since the presence of a following [3], [h] or [x] makes it
impossible to determine whether or not the Chilean plural variant [5] or [b] was produced,

as shown in (3) below.

117

(3) a. todos juntos
all.PL together.PL

[todos__xuntos], [todoh_xuntos], [todeuntos]

b. 105 sabados
the.PL Saturdays.PL

[los_saf3ados], [lohgaﬁaeos], [lgaﬁaoos]

In cases as noted above, the plural token todos in (3a) and the plural token [OS in (3b)
would not be included in the analysis. Plural morphemes followed by words beginning
with [s], [h] or [x] only accounted for 1% of the total Chilean production data

The logistic regression contained six independent variables or factor groups and

each factor group had up to ﬁve factors each. The six factor groups are shown in Table

14.

118

Table 14. Factor Groups in Analysis of Chilean Production Data.

 

 

Independent Variables Factors
Age of Speaker Children
Adults

Social Class

Style

Independent Number

Information

Syntactic Category

Order

 

Working-class

Middle-class

Free Speech
Repetition Task

Naming Task

Yes: DP with numerals or quantiﬁers

No: DP without numerals or quantiﬁers

Deﬁnites (mis, nuestros, sus, los, ellas; clitics: les, los)
Indeﬁnite Determiner (unos)

Quantiﬁers (algunos, muchos, pocos, todos, hartos, etc.)
Adjectives

Nouns

First
Second

Third or more

 

119

The factor groups included in the analysis were Age, Social Class, Style, Independent
Number Information, Syntactic Category and Order. Age, Social Class and Style are
extra-linguistic factors and Syntactic Category, Number Information and Order are
linguistic factors. For the factor group ‘Age of Speaker’ the tokens were coded as being
produced either by adult speakers or child speakers.

For the factor group ‘Social Class’, the tokens were coded as being produced
either by working-class speakers or middle class speakers. Social class was determined
primarily by the profession of the children’s parents and the tuition of the school children
attended. The occupations of the working-class Chilean and Mexican parents included
secretary, ﬁsherman, butcher, low-ranking military, taxi driver, homemaker, janitor, and
inventory worker. The occupations of the middle-class Chilean and Mexican parents
included business men/women, high-ranking military officials, doctor, university
professors, and lawyer. While the working-class Mexican and Chilean children attended
free daycare/preschool and kindergartens, the middle-class Chilean children (middle-
class Mexican children were not tested) attended private schools which were considered
prestigious and had the highest tuition rates in the town in which they were located (in
Punta Arenas, Chile). Social class was included in the analysis because Cepeda (1995)
found that plural morpheme omission is more ﬁ'equent in the speech of Chilean working-
class speakers than in Chilean middle-class and upper-class speakers. Working-class
Mexican children were only included in the study to ensure that any differences between

working-class and middle-class Chilean children were not due entirely to social class

120

(e. g. educational experiences) but rather to the differences in plural morpheme omission
in the adult speech (i.e. the linguistic input to which children were exposed).

For the factor group ‘Style’ tokens were coded as being produced during the Free
Speech Task, the Repetition Task, or the Naming Task. The Free Speech Task was the
most informal production task and the Naming Task and Repetition Task were the most
formal. It was predicted that we would ﬁnd more omissions in the Free Speech Task and
less omissions in the Naming Task and Repetition Task.

The factor group ‘Independent Number Information’ refers to whether or not the
determiner phrase contained independent number information in the form of plural
numerals or plural quantiﬁers (e. g. muchos (‘many.PL’), pocos (‘few.PL’), todos (‘all.PL’),
hartos (‘several.PL’), varios (‘various.PL’). This was included because it was observed in
pilot data that Chilean subjects often omitted the plural morpheme on nouns when they
were preceded by numerals and quantiﬁers, as in phrases like Tiene 5 ar'ioﬂ (‘He is 5
years.PL old’) or todoﬂ loﬂar‘ioﬂ (‘all.PL the.PL years.PL’).

For the factor group ‘Syntactic Category’ tokens were coded for the syntactic
category to which the plural lexical item belonged. There were ﬁve syntactic categories,
as shown in Table 14. The category ‘Deﬁnite’ included deﬁnite determiners, possessive
determiners, demonstratives and clitic pronouns (les, los, las). The category indeﬁnite
determiner only included the indeﬁnite forms unos/unas (‘some.PL’). Indeﬁnite
determiners were coded separately from other determiners because of the similarity
between the word for ‘one’ (un/una) and the plural indeﬁnite (see comprehension
experiments in Chapter 5). I was interested to see whether speakers would retain the

plural morpheme more often on the indeﬁnite determiner because without it, the plural

121

indeﬁnite would be identical in form to the numeral ‘one’ in Spanish. Quantiﬁers like
algunos (‘some.PL’), muchos (‘many.PL’), pocos (‘few.PL’), hartos (‘many.PL’), and
todos (‘all.PL’) were coded as Quantiﬁer. Algunos (‘some.PL’) was coded as a quantiﬁer
in the factor group Syntactic Category because it is used more often in the plural form
than in the singular and hence its presence generally indicates quantity. In the singular it
appears to have very special restrictions. The syntactic category Noun included common
nouns only.

Finally, the factor group “Order” was included because Cepeda (1995) found that
plural elements which occur ﬁrst in the determiner phrase are generally marked for plural
while plural elements which occur later in the determiner phrase are not. For this factor
group, each plural token was coded for its linear position within the determiner phrase
with respect to other plural tokens. The sentences in (4) and Table 15 provide an example
of how lexical items were ordered. The method for ordering the items in this way was
based on Cepeda (1995) in order to allow for comparison with previous research on

Chilean Spanish. The examples in (4) would be coded for order according to Table 15.

(4) a. unos gatos

some.PL cats.PL

b. los gatos y los perros

the.PL cats.PL and the.PL dogs.PL

122

cuatro gatos

four cats.PL

tengo gatos

I have cats.PL

los otros gatos

the.PL other.PL cats.PL

los otros aﬁos anteriores

the.PL other.PL years.PL before.PL

Table 15. Ordering of Lexical Items within the DP.

 

 

l 2 3+ 3+

a. unos gatos
b. los gatos

los perros
c. (cuatro) gatos
d. gatos
e. los otros gatos
f. 105 otros aflos anteriores

 

123

As shown in Table 15, ordering consisted of ﬁrst position, second position and any
position beyond second position (3+). Only plural tokens and numerals were counted in
the ordering. Cepeda was interested in whether redundancy of plural marking caused
subsequent elements within the noun phrase to occur without a plural morpheme.

While there is very little research about the amount of recorded speech needed to
carry out an individual variable rule analysis on children using binary logistic regression,
Roberts (1994) found that a total of 30 tokens per factor in each factor group was
sufficient for such an analysis (in line with Guy 1980). Since the factor groups in the
present analysis have up to ﬁve factors each, my goal was to collect approximately 150
tokens of plural morpheme production for each subject group: ChWC children, ChMC
children, ChWC adults and ChMC adults. While an optimal goal would have been to
collect 150 tokens per subject (instead of per subject group) in order to carry out an
individual variable rule analysis of each subject, this would have required approximately
1.5 hours of free speech per child (we found that children produced approximately 35
plural tokens in a 15-minute period). Such an extensive study of production is beyond the
scope of this dissertation given the numerous comprehension tasks that were also being
carried out. In any case, I believe that the production data presented in this chapter, along
with previous research on Mexican and Chilean Spanish, will nonetheless provide an
informative overview of the input to which Mexican and Chilean children are exposed

and the production of plural morphology in Mexican and Chilean children.

4.5 Results
A total of 1739 plural tokens were collected from Chilean subjects. Of these 1739

tokens, 646 were from ChMC adults, 336 from ChMC children, 258 from ChWC adults

124

and 499 from ChWC children. A total of 1308 plural tokens were collected ﬁom Mexican
subjects. Of these 1308 plural tokens, 433 were from MexWC adults, 340 from MexMC
adults, and 535 from MexWC children.

It is ﬁrst important to note that with respect to the Chilean data, inter-rater
reliability between the two Chilean native speaking coders reached approximately 83.5%
of all plural tokens. In other words, the coders disagreed on 285 of the 1739 tokens
(16.4%) and disagreement was between [h] vs. Q or the unreleased glottal stop vs. Q,
where one native speaker coded as g and the other has [h] or the glottal stop. In such
cases of disagreement between coders (approximately 16.4% of the tokens), the token
was coded as Z under the reasoning that if native speaking coders could not decide
between zero vs. plural ([h] or the unreleased glottal stop), native speaking children
would also have difﬁculty distinguishing between presence vs. absence of the plural
morpheme.

The logistic regression revealed that only ﬁve of the six factor groups contributed
signiﬁcantly to the model describing the distribution of plural morpheme omission. These
factor groups are: ‘Age of Speaker’, ‘Social Class’, ‘Style’, ‘Independent Number
Information’ and ‘Syntactic Category’. The factor group ‘Order’ did not inﬂuence the
omission of the plural morpheme in the model generated by the logistic regression. Table
16 shows the ﬁve factor groups followed by the factor weights. The factor weights are
reported as probabilities between zero and one. The further the factor weight is from 0.5,
the greater its effect on the resulting probability. A factor weight above 0.5 can be
considered as favoring the omission of the plural morpheme and a factor weight below

0.5 as disfavoring omission of the plural morpheme (Paolillo 2002, Smith et al. 2006).

125

Table 16. Factor Groups and Factor Weights for Chilean Production Data.

 

 

Factor Groups Factor Groups
Age of Speaker Adult Child
.468 .535
Social Class MC WC
.359 .679
Style F. Speech Repetition Naming
.609 .407 .349
Independent Yes No
Number .579 .484
Information
Syntactic Adj. Noun Quantiﬁer Indeﬁnite Determiner
Category .71 .563 .478 .428 .257

 

 

Log likelihood = -1038.802

 

The results of the logistic regression showed that the 5 factors presented in Table 16 had
a signiﬁcant effect on the omission of the plural morpheme (p<.05). The overall
probability that the plural morpheme would be omitted, given the model presented in
Table 16, is .63. Table 16 basically conﬁrms what was reported in Cepeda (1995) with

respect to Chilean adult Spanish-speakers. First of all, the factor ‘Working-class’ (.678)

126

favors plural morpheme omission while the factor ‘Middle-class’ (.360) disfavors plural
morpheme omission. Moreover, like Cepeda, we ﬁnd that the factor ‘Determiner’
(deﬁnite, possessive and clitics) (.251) disfavors omission. while the factors ‘Adjective’
(.702) and ‘Noun’ (.563) favor omission of the plural morpheme. The data in Table 16
also reveal that Style affects the omission of the plural morpheme, with the Free Speech
Task (.609) favoring omission and the Repetition Task (.407) and the Naming Task
(.349) disfavoring omission. With respect to independent number information (i.e. the
plural morpheme occurring in noun phrases with quantiﬁers or numerals), the data show
that the presence of a quantiﬁer or numeral favors omission slightly more than the
absence of quantiﬁers and numerals. With respect to the age of the subject, the data
reveal that Chilean children are slightly more likely to omit the plural morpheme than
Chilean adult speakers.

Of the 1739 plural tokens collected from Chilean subjects, 671 (38.5%) were 2”
and 1068 (61.4%) were marked for plural ([h] or [s]). Of the 1308 plural tokens collected
from Mexican subjects, 11 (0.8%) were O and 1297 (99.1%) were marked for plural.
Table 17 shows the amount of plural morpheme production and omission broken down
per child and adult groups. The amount of data collected from each subject was not
uniform for a variety of reasons, including the fact that some children were shy and more
reluctant to talk than others or because of interference in recording (e. g. child ﬁdgeting

during the recording session, etc.).

 

'5 The reader is reminded here that disagreement between coders on [h] vs. Q or the
unreleased glottal stop and Q was counted in the analysis as Q

127

Table 17. Percentage of Plural Responses for Production Data.

 

 

Subject Subject Morpheme Omission Total Plural
SES Age ([s] or [h]) (E) Tokens
ChMC Adults 67% (433) 33% (213) 100% (646)

Children 83% (278) 17% (58) 100% (336)
ChWC Adults 56% (145) 44% (113) 100% (258)
Children 42% (212) 58% (287) 100% (499)
MexMC Adults 99% (335) 1% (5) 100% (340)
MexWC Adults 99% (430) 1% (3) 100% (433)
Children 99% (532) 1% (3) 100% Q35)

 

Table 17 shows that the production of plural morphology differs between Mexican and

Chilean subjects. While Mexican adult and child subjects produced the plural morpheme

on almost 100% of all plural lexical items, Chilean adult and child subjects omitted the

plural morpheme on 17% to 58% of all plural lexical items. Hence, the data conﬁrm that

the linguistic input to Chilean vs. Mexican children differs with respect to the plural

morpheme. Given the difference in linguistic input, the data further reveal that Chilean

children and Mexican children do not pattern alike in their production of plural

morphology. Rather Chilean children, like Chilean adults, appear to show variability in

their production of plural morphology and Mexican children, like Mexican adults, appear

to consistently mark plural lexical items with a plural morpheme. This suggests that

128

children exposed to variable input will be variable in their own production. Furthermore,
it suggests that when plural morphology is systematically produced in the input, children,
in this case Mexican children, will have acquired that adult-like systematicity in their
own production by approximately 4 years of age.

Secondly, Table 17 shows that differences in plural morpheme production based
on SES only occur in Chilean speakers but not Mexican speakers. The data show that
while there are no differences in plural morpheme production between MexWC adults
and MexMC adults and that MexWC children are at ceiling in their production of the
plural morpheme (suggesting very strongly that MexMC children would also perform at
ceiling and hence, pattern with MexWC children), there are several differences between
the Chilean working-class and Chilean middle-class groups. First of all, ChWC adults
omit the plural morpheme at a higher rate than ChMC adults, which is consistent with
Cepeda (1995). Likewise, the data indicate that ChWC children omit the plural
morpheme at a higher rate than ChMC children. Hence, the data show an overall trend of
Mexican subjects producing plural morphology all of the time followed by ChMC
subjects producing plural morphology less than Mexican subjects but more than ChWC
subjects. ChWC subjects omit the plural morpheme more than any of the other dialect
groups. This is consistent with previous research indicating that Mexican Spanish always
marks the plural and with research indicating that ChWC subjects omit the plural
morpheme more often than ChMC subjects (Cepeda 1995, Lipski 1994, Canﬁeld 1982).

One unexpected ﬁnding was that while ChWC children omitted the plural
morpheme more often than ChWC adults, ChMC children produced the plural morpheme

more often than ChMC adults. ChMC adults omitted the plural morpheme on

129

approximately 33% of all plural tokens while ChMC children only omitted the plural
morpheme on 17% of all tokens. This suggests that ChMC adults may omit the plural
morpheme less frequently when addressing their children than when addressing other
adults. This is consistent with Smith et al. (2006) who found that adult speakers differ in
their use of variable features depending on whether they are addressing their own
children or other adults. Production data of adult ChMC speakers talking with their own
children will be needed in future studies to determine whether adults indeed omit less
when addressing their own children: however, the data here suggests that this may be the
case. Alternatively, it is possible that ChMC children are regularizing the variable input
to some extent, using the plural morpheme less variably than adults. If children have not
acquired all of the constraints governing the variable behavior of plural morpheme
omission, the data may be inconsistent for them. Hence, we might expect them to
regularize to some extent. This is consistent with Hudson Kam and Newport 2005. Future
research will need to address this question.

i The data show that Chilean subjects are variable in their production of plural
morphology, sometimes omitting the morpheme and sometimes producing it, while
Mexican subjects are not. Looking more closely at the variants of the plural morpheme
produced by subjects we ﬁnd that, while Chilean adults and children produced both [5]
and [h] as the plural morpheme to varying degrees, Mexican adults and children almost
always produced the plural morpheme as an alveolar fricative. Table 18 shows the
percentage of the variants [s] and [h] of the plural morpheme and also of the omission of

the plural morpheme.

130

Table 18. Percentage of Plural Variants in Production Data.

 

 

Subject Subject [s] [h] Z
SES Age

ChMC Adults 15% (95) 52% (338) 33% (213)
Children 44% (149) 38% (129) 17% (58)

ChWC Adults 13% (34) 43% (110) 44% (113)
Children 22% (108) 21% (104) 58% (287)

MexMC Adults 98% (332) 0.9% (3) 1.5% (5)

MexWC Adults 98% (425) 1% (5) 0.7% (3)
Children 99% (531) 0.2% (1) 0.5% (3)

 

In Table 18 we see that ChWC and ChMC adult groups seemed to pattern together by
using [h] as the plural morpheme more often than [5]. Unlike the Chilean adult groups,
ChWC and ChMC children produced the plural variant [5] just as often as [h]. In fact,
both ChWC and ChMC children produced the plural as [s] more often than Chilean
adults. This suggests that Chilean children are exposed to both the [s] and [h] variants of
the plural morpheme in the input they are exposed to and that Chilean adult speakers may
be using the [s] variant more often with their children than with other adult speakers.
However, we must be cautious in assuming that this ﬁnding suggests that Chilean adults
may use [5] more often with their children given that we have not yet broken the data

down by task type. For example, there are more Chilean children than Chilean adults who

131

participated in the Naming Task and Repetition Task, which might elicit more [s] variants
than the Free Speech Task. The next section (Section 4.5.1) will look at this in more
detail.

A very important question at this point is whether there were Chilean children
who were systematic in their production or omission of the plural morpheme, in other
words, whether there were individual children who always omitted or always produced
the plural morpheme. While we found that all Mexican children were systematic in their
production of plural morphology, always producing it, the above data does not clarify this
for the Chilean children. A closer look at the data show that for the most part Chilean
children were not systematic, but rather they were variable, in their production of the
plural morpheme. The number of plural tokens produced with and without the plural

morpheme for each child and adult subject is shown in Table 19.

132

Table 19. Number of Plural Tokens by Subject.

 

 

 

 

ChWC Children

Subject Z [s]/[h]
l 3 4
2 12 , 22
3 7 O
4 l7 l4
5 27 9
6 29 11
7 34 7
8 22 14
9 8 34
10 10 7
ll 8 O
12 18 13
13 27 23
14 22 8
15 17 29
16 18 3

ChMC Children

Subject Q [s]/[h]
1 9 24
2 6 24
3 19 50
4 4 39
5 3 3O
6 7 39
7 7 31

ChWC Adults

Subject 9 [s]/[h]
l 23 19
2 38 76
3 50 49

ChMC Adults

Subject Q [s]/[h]
1 52 125
2 60 121
3 34 72
4 67 115

 

133

Table 19. Number of Plural Tokens by Subject cont’d.

 

 

 

MexWC Children
Subject 9 Isl/[h]
l 0 92
2 0 l9
3 l 76
4 0 60
5 l 43
6 O l 7
MexMC Adults
Subject 9 [SI/[h]
1 2 248
2 3 87
MexWC Adults
Subject Q [s]/[h]
1 3 l 54
2 0 276

 

There were only 2 ChWC children who never produced a plural token with a plural
morpheme (ChWC 3 and ChWCl 1); however, these are only two children and also each
child produced few tokens. For that reason, we must be cautious in generalizing this
particular ﬁnding to the larger population. The data in Table 19 show that each individual
Chilean subject is variable in their production of the plural morpheme, while the Mexican

subjects are systematic.

4.5.1 The Effect of Style on Plural Morpheme Production

Research has shown that different types of speech tasks (free speech task vs.
picture description task) affect the variable forms used by 4 — 6 year old children
(Washington et al. 1998). Furthermore, research has indicated that by four years of age

children have acquired some but not all of the constraints governing the variable rules

134

and that extra-linguistic constraints appear to be acquired later than linguistic constraints
(Roberts 1994, 1997, Smith et al. 2006). The production data collected for this
dissertation was analyzed for Speech Style as it was predicted that in more careful
speech, speakers would produce the plural morpheme more often. In other words, it was
predicted that there would be more omissions in the Free Speech Task and less omissions
in the Repetition and Naming Tasks, under the assumption that the Free Speech task
would elicit more casual speech and the Repetition and Naming Tasks more careful
speech. Table 20 shows the percentage of plural morpheme omissions according to

speech task.

135

Table 20. The Effect of Style on Plural Morpheme Production.

Free Speech Repetition Task Naming Task

 

ChMC Adults
[5]
[h]
a
ChWC Adults
[5]

[h]
g

ChMC Children
[8]

[h]
g

ChWC Children
[8]
[h]

 

46% (43/93)

4% (19/505)
57% (287/505) 45% (42/93)
39% (199/505) 9% (8/93)

3% (6/194) 24% (10/42)

46% (89/194) 48% (20/42)
50.5% (98/194) 29% (12/42)

32% (38/117) 45% (80/179)

42% (49/117) 43% (77/179)

26% (30/117) 12% (22/179)

16% (1 1/67)
31% (21/67) 21.5% (56/261)
52% (35/67) 62% (162/261)

69% (33/48)
19% (9/48)

12.5% (6/48)

86% (19/22)
4.5% (1/22)

9% (2/22)

85% (34/40)
0%

15% (6/40)

16.5% (43/261) 31.5% (54/171)

16% (27/171)

53% (90/171)

 

Table 20 shows the percentage of plural tokens marked with the plural variants [s] or [b]

and also the percentage of plural morpheme omission for Chilean Subjects. Table 20 does

not display results for Mexican subjects because there was no variability. First, note that

136

there is less Free Speech Task data for children than for adults but more Repetition Task
and Naming Task data for children than for adults. This is due to the fact that less free
speech data was collected ﬁ‘om Chilean children than ﬁ'om Chilean adults and some
children produced very few plural tokens in free speech. In any case, it is revealing to
observe that even in the Repetition Task” and the Naming Task, ChWC children omitted
the plural morpheme more often than any of the other three groups in any of the other
tasks, including the Free Speech Task. In fact, ChWC children omitted the plural
morpheme as much in the Repetition Task and the Naming Task as they do in the Free
Speech Task, which suggests that the small number of plural tokens collected for ChWC
children in the Free Speech Task does not directly affect the overall ﬁndings of the
production data presented here.

With respect to the variant forms of the plural morpheme, it is ﬁrst important to
note that both the [s] and [h] variant forms and also omissions were found in the speech
of all Chilean groups: ChMC children, ChMC adults, ChWC children, ChWC adults. The
data show that frequency of the variant forms changed depending on the speech task.
Table 20 shows that in the Free Speech Task all Chilean subjects used [h] more
frequently than [8]; however, note that in the Free Speech Task both ChWC adults and
ChWC children omitted the plural morpheme more often than they used [h]. As the
experimental task increases in formality, the percentage of omissions decreases while the
percentage of the variant [s] increases for both ChWC adults and ChMC adults. The same
trend was found for ChMC children but not for ChWC children, who maintained

approximately the same percentage of omissions across speech tasks. However, ChWC

 

'6 Remember that the Chilean researcher was pronouncing the plural morpheme as [s] in
the Repetition Task.

137

children, like the other three Chilean groups, did drastically increase their usage of [s] in
the Naming Task. A second Naming Task (Study 4) is presented in Chapter 5 where I
will discuss in more detail the increase of the plm'al variant [5].

It is interesting to note at this point the fact that the [s] variant was more frequent
in the Free Speech and Naming Task of ChMC children than in the speech of the ChMC
adults. In the Repetition Task the ﬁ'equency of [s] was similar between the two middle-
class groups. This data further suggests that ChMC adults may be using more [s] when
addressing their own children than when addressing other adult speakers. In a sense,
ChMC children tend to overuse the [s] variant in Free Speech compared to adult
speakers. On the same token, notice that ChWC children produce more omissions than
the ChWC adults in the Repetition Task and the Naming Task, but not in the Free Speech
Task, where the frequency of omissions is similar between ChWC adults and ChWC
children. This suggests that ChWC adults may produce more omissions in their speech to
their children yet they understand the social constraints governing plural morpheme

omission resulting in a decrease of omissions in the more formal speech tasks.

4.5.2 The Effect of Syntactic Category on Plural Morpheme Production

Cepeda (1995) found that the plural morpheme was produced more often in
certain syntactic categories than in others. For example, she found that the plural
morpheme was produced more often on determiners than on nouns or adjectives. It is
difﬁcult to conclude from her data whether the plural morpheme was produced equally
on quantiﬁers, indefmites and adjectives because Cepeda combined these together in her
analysis. Table 21 shows the data collected in the present dissertation broken down by

syntactic category.

138

Table 21. The Effect of Syntactic Category on Plural Morpheme Production.

 

ChMC Adults Indefinite Definite Quantifier Noun Adjective

 

Plural [s]/[h] 81% 86% 71% 61% 39%
(22/27) (136/158) (46/65) (206/337) (23/59)

[s] 22% 1% 12% 23% 3%
(6/27) (2/158) (8/65) (77/337) (2/59)

[h] 59% 85% 58% 38% 36%
(16/27) (134/158) (38/65) (129/337) (21/59)

a 19% 14% 29% 39% 61%

(5/27) (22/158) (19/65) (131/337) (36/59)

ChWC Adults Indefinite Definite Quantiﬁer Noun Adjective

 

Plural [s]/[h] 100% 66% 63% 48% 50%
(12/12) (40/61) (15/24) (73/151) (5/10)

[s] 17% 3% 8% 19% 10%
(2/12) (2/61) (2/24) (28/151) (1/10)

[h] 83% 62% 54% 30% 40%
(10/12) (38/61) (13/24) (45/151) (4/10)

a 0% 34% 38% 52% 50%

(0) (21/61) (9/24) (78/151) (5/10)

ChMC Children Indeﬁnite Deﬁnite Quantifier Noun Adjective

 

Plural [s]/[h] 94% 79% 86% 81% 67%
(30/32) (22/28) (51/59) (173/214) (2/3)

[s] 19% 11% 17% 60% 67%
(6/32) (3/28) (10/59) (128/214) (2/3)

[h] 75% 68% 69.5% 21% 0%
(24/32) (19/28) (41/59) (45/214) (0)

a 6% 21% 13.5% 19% 33%

(2/32) (6/28) (8/59) (41/214) (1/3)

ChWC Children Indefinite Definite Quantiﬁer Noun Adjective

 

Plural [s]/[h] 41% 65% 36% 43% 36%
(18/44) (15/23) (24/67) (151/354) (4/11)

[s] 9% 4% 6% 28% 9%
(4/44) (1/23) (4/67) (98/354) (1/11)

[h] 32% 61% 30% 15% 27%
(14/44) (14/23) (20/67) (53/354) (3/11)

a 59% 35% 64% 57% 63%

 

(26/44) (8/23) (43/67) (203/354) (7/11)

 

139

The Chilean data show that both ChMC and ChWC adults produce the plural morpheme
more often on determiners (including indef'mites and quantiﬁers) and omit the plural
morpheme more often on nouns and adjectives. Like ChWC adults, ChWC children
appear to produce the plural morpheme more often on determiners than on nouns.
However, ChMC children do not appear to be inﬂuenced by syntactic category in their
production of the plural morpheme. While it may appear that syntactic category does
affect their production (94% on Indeﬁnites appears different ﬁ'om 67% on adjectives), we
must take into consideration that there are only 3 tokens total for the syntactic category
Adjectives, which is too small to draw any conclusions from.

However, the data do show that all Chilean subjects, including child subjects,
appear to be constrained by syntactic category to some extent in their production of the
two plural variants [s] and [h]. All four subjects groups produced the [h] variant more
often on determiners and although the [s] variant was not very frequent in the data,
Chilean adult subjects produced it more often on nouns and on indeﬁnite determiners
while Chilean child subjects produced it more frequently on nouns. This is especially true
for ChMC children who produced the variant [s] 60% of the time on nouns but only 11%

of the time on Determiners.

4.5.3 The Effect of Order on Plural Morpheme Production

In addition to syntactic category, Cepeda (1995) found that Chilean adult speakers
produced the plural morpheme in different frequencies depending on the linear position
of the pllnal token within the determiner phrase. Cepeda found that it was the ﬁrst NP

constituent that carried the functional load of the plural morpheme and that subsequent

140

elements within the determiner phrase often occurred without the plural morpheme.

Cepeda’s data is shown in Table 22.

Table 22. The Effect of Linear Order on Plural Morpheme Production.

 

 

Premodifier17 Noun
1 2 N1 N2 N3
[s]/[h] 79% 71% 55% 42% 39%
Z 21% 24% 45% 58% 61 %

 

*Adapted from Cepeda (1995).

Cepeda’ s results showed that the plural morpheme was omitted more often on nouns than
on nominal premodiﬁers (i.e. determiners), regardless of the linear order. Note that in
column N1 , which indicates that the noun occurred bare, without a determiner, the plural
morpheme is omitted approximately 45%. However, in the Premodiﬁer columns the
plural morpheme is only omitted approximately 21% of the time. Both the Premodiﬁer 1
column and the N1 column are the ﬁrst linear position in the DP. Cepeda’s data suggest
that the ﬁrst linear position retains the plural morpheme more often than the second and
third linear positions.

While the factor group ‘Order’ did not contribute signiﬁcantly to the overall
model of plural morpheme omission in the present dissertation, I will nevertheless
discuss it in order to highlight any trends. Table 23 shows the percentage of plural

morpheme production and omission on determiners and on nouns when the noun

 

'7 Cepeda (1995) used the term “Premodiﬁer” to describe determiners, quantiﬁers,
adjectives and numerals. Unfortunately, she did not break the data down by type.

141

occupies the ﬁrst linear position (bare nouns) and the second linear position (nouns

headed by a determiner or numeral).

Table 23. The Effect of Order on Plural Morpheme Production.

 

 

 

Determiner
D1 N1 N2
(Bare Nouns) (D/Q + N)
ChMC Plural Isl/[h] 87% 75% 55%
Adults (122/ 141) (82/110) (109/198)
Q 13% 25% 45%
(19/141) (28/110) (89/198)
ChWC Plural [SI/[h] 68% 68% 37%
Adults (37/57) (36/53) (32/86)
@ 32% 32% 63%
(20/57) (17/53) (54/86)
ChMC Plural Isl/[h] 77% 81% 83%
Children (20/26) (82/101) (90/109)
Q 23% 19% 17%
(6/26) (19/101) (19/109)
ChWC Plural Isl/[h] 64% 44% 40%
Children (14/22) (100/227) (SO/125)
Z 36% 56% 60%
(8/22) (127/227) (75/125)

 

Table 23 is consistent with data reported in Cepeda (1995) in that Chilean adult subjects

produce the plural morpheme more often on determiners and on NI (bare nouns) than on

N2 (nouns followed by a determiner or numeral). ChMC adult speakers produced the

142

plural morpheme on Determiner 87% of the time and on N1 75% of the time while they
produced the plural morpheme on N2 only 55% of the time. Likewise, ChWC adult
speakers produced the plural morpheme on Determiner 68% of the time and on N1 68%
of the time while they produced the plural morpheme on-N2 only 37% of the time. While
the production of the plural morpheme appears to be affected by whether or not a
previous element within the determiner phrase was already marked for plural in Chilean
adult speakers, this does not seem to affect the production of the plural morpheme in
Chilean children. ChMC children produced the plural morpheme on Determiner 77% of
the time, on N1 81% of the time, and on N2 83% of the time. Likewise, ChWC children
produced the plural morpheme on Determiner 64% of the time, on N1 44% of the time
and on N2 40% of the time. In this sense, while the existence of a previous element
marked for the plural within the determiner phrase affects plural morpheme production
on subsequent elements in Chilean adult speech, it does not appear to affect the

production of the plural morpheme in Chilean child speech.

4.5.4 The Effect of Number Information on Plural Morpheme Production

Our pilot data suggested that Chilean subjects may omit the plural morpheme
more often on nouns when the determiner phrase contained independent number
information in the form of plural quantiﬁers or numerals and this seems to be true given
that the logistic regression analysis showed that the factor group ‘Independent Number
Information’ added signiﬁcantly to the model predicting plural morpheme omission.
Table 25 shows the percentage of plural morpheme omission and production on nouns
when the noun was headed by a plural quantiﬁer or numeral (coded as ‘Yes’ in the table)

and when the noun was not headed by a plural quantiﬁer or numeral (coded as ‘No’ in the

143

table). An example of noun phrases coded as ‘Yes’ and ‘No’ are shown in Table 24

below.

Table 24. Coding of Number Information in Production Data.

 

 

 

YES NO
todos los dias los dias
all.PL the.PL days.PL the.PL dayS.PL
‘every day’ ‘the days’
muchos nifios unos niﬁos

many.PL children.PL

‘many children’

some.PL children.PL

‘some children’

 

hartas personas

several.PL persons.PL

algunos niﬁos

some.PL children.PL

 

‘many people’ ‘some children’
2 manzanas niﬂos
2 apples.PL children.PL
‘2 apples’ ‘children’

 

 

Table 25 shows the effect of number information on plural morpheme production by

Chilean subjects.

Table 25. The Effect of Number Information on Plural Morpheme Production.

 

 

Yes No
ChMC Adults

[s]/[h] 56% 63%
(42/75) (165/263)

Q 44% 37%
(33/75) (98/263)

ChWC Adults

Isl/[h] 31% 54%
(13/42) (61/113)

Q 69% 46%
(29/42) (52/1 13)

ChMC Children

[s]/[h] 65% 82%
(ll/l7) (162/197)

Q 35% 18%
(6/17) (3 5/ l 97)

ChWC Children

Isl/[h] 50% 42%
(6/ 12) (145/342)

Q 50% 58%
(6/ 12) (197/342)

 

145

The data presented in Table 25 reveal an increase in plural morpheme omission for
ChWC adults and for ChMC children and a slight increase in omissions for ChMC adults
when the noun is headed by a plural quantiﬁer or numeral. On the other hand, no such
increase in omissions is found for ChWC children. This data suggest that Chilean adults
omit the plural morpheme more often when the noun is headed by a plural quantiﬁer or
numeral, in other words, when there is independent number information within the
determiner phrase. In addition, the data suggest that only ChMC children, but not ChWC

children, have acquired this constraint.

4.6 Summary of Results

The following is a summary of the results for the production data presented in this
chapter.

1. Chilean adult and Chilean child speakers have variable production of the plural
morpheme, sometimes producing it as [s] and [h] and sometimes omitting it.

2. Mexican adult and Mexican children speakers have systematic production of
the plural morpheme, always producing it as an alveolar fricative ([s] and [z]).

3. Chilean working-class children and adult speakers omit the plural morpheme
more often than Chilean middle-class child and adult speakers.

4. Chilean working-class children omit the plural morpheme more often than
adult speakers and also more often than Chilean middle-class children.

5. Chilean middle-class children produce the plural variant [s] more often than
adult speakers and also more often than Chilean working-class children. Chilean middle-

class children appear to overuse the [s] variant in comparison with the other three subject

groups.

146

6. Chilean middle-class and Chilean working-class adult speakers appear to be
constrained by Style, Independent Number Information, Syntactic Category, and Order in
their omission of the plural morpheme.

7. Chilean working-class children do not appear to be constrained by Style,
Order, nor Independent Number Information in their omission of the plural morpheme
but they do appear to be constrained by Syntactic Category, omitting the plural
morpheme more often on nouns than on determiners.

8. Chilean middle-class children do not appear to be constrained by Syntactic
Category (i.e. nouns vs. determiners) nor Order in their omission of the plural morpheme
but they do appear to be constrained by Style and Independent Number Information in
their omission of the plural morpheme.

9. All Chilean adult and child groups appear to be constrained in their production
of the plural variants [s] vs. [h] by both Syntactic Category (e. g. nouns vs. determiners)
and Style (e. g. Free Speech vs. Repetition and Naming Tasks), using [s] more often on

nouns than on determiners and also more often in careful speech (The Naming Task).

4.7 Discussion

This chapter had three main goals. First, it aimed to conﬁrm previous reports on
the patterns of plural morpheme production in Mexican and Chilean adult speakers by
showing that Mexican adults are systematic in their production of the plural morpheme,
always producing it as an alveolar ﬁ'icative, while Chilean adults are variable in their
production of the plural morpheme, producing it as both [8] and [h] and sometimes
omitting it. The second goal was to compare the production of plural morphology by

Chilean and Mexican children with that of their parents to determine whether there are

147

developmental differences in the acquisition of plural morphology in general. Finally, the
third goal of this chapter was to compare the production of plural morphology between
Chilean adults and children vs. Mexican adults and children to determine any differences
between dialects. Ultimately, the overall purpose of these three goals is to determine how
differences in the linguistic input to Mexican and Chilean children may affect production
and comprehension of the plural morpheme in Chilean and Mexican children.

The data presented in this chapter conﬁrms previous research on Mexican and
Chilean adult speech by showing that Chilean speakers have variable behavior in their
production of the plural morpheme while Mexican speakers systematically produce the
plural morpheme as an alveolar fricative. In addition, the results showed that ChWC
adults omitted the plural morpheme more often than ChMC adults. The data presented in
Table 26 provides some examples of plural morpheme omission and production in

Chilean adult speech.

148

Table 26. Free Speech Samples of Plural Morphology Produced by Chilean Adults.

 

Plural ([s]/[h])

Omission of Plural

 

Defmites

Indefinites

Quantifiers

Nouns w/ Numerals

Nouns w/

Determiners

Adjectives

Bare Nouns

 

mi-[h] niﬁita—[o]
my.PL daughter.PL
‘my daughters’

uno—[h] aflo-[h]
some.PL years.PL
‘some years’

de toda- [h] manera-[o]
in all.PL cases.PL
‘in any case’

treinta grado-[s]
thirty degrees.PL
‘thirty degrees’

una-[h] muﬁeca-[s]
some.PL dolls.PL
‘some dolls’

tengo bueno- [h] recuerdo- [a]
I.have goodPL memory.PL
‘I have good memories’

no tengo
I don’t have siblings.PL
‘I don’t have siblings’

hermano-[h]

Todo-[o] lo-[o] aﬁo—[o]
all.PL the.PL year.PL
‘Every year’

uno-[o] mese-[o]
some.PL months.PL
‘some months’

todo-[o] lo-[h] materiale-[h]
all.PL the.PL materials.PL
‘all of the materials’

uno se va do—[h] mese-[o]
one goes two month.PL
‘one goes for two months’

lo-[h] dia-[o]
the.PL days.PL
‘the days’

estamo- [1211'8 acostumbrado—[o]
we are accustomed.PL
‘we are used to it’

tienen actividade-[o] . .
they have activity.PL.
‘they have activities...’

However, the reader should not rely solely on this chapter to be convinced of the

differences in plural morpheme production in these two varieties of Spanish, especially

given the small amount of data presented here. Instead, the reader should also consider

previous reports on Chilean and Mexican Spanish (Cepeda 1995, Lipski 1994, Canﬁeld

1982).

 

'8 /s/ omission on verbs, like estamos (‘are. 1.PL’), were not included in the analysis of
this dissertation. The adjective acostumbrados (‘accustomed.PL’) was counted as not
having a plural morpheme.

149

The data also show that no differences appear to exist between adult and child
speakers with respect to the production of the plural morpheme, as Mexican children like
Mexican adults were systematic in their own production and Chilean children like
Chilean adults were variable in their own production. However, it does appear that for
Chilean subjects there are differences between adult and child speakers with respect to
the linguistic and extra-linguistic constraints that govern the variability in the production
of the plural morpheme. The variability found in Chilean children is not identical to the
variability found in Chilean adults, as noted above in the Chapter summary.

Finally, it is clear from the data presented here and in previous studies that the
linguistic input to Chilean children differs drastically from the linguistic input to Mexican
children with respect to plural morphology. Likewise, the data show that the linguistic
input to ChWC children is also quite different ﬁ'om the linguistic input to ChMC
children. More speciﬁcally, the data show that there are no omissions in the input to
MexWC while plural morpheme omissions range from 33% (ChMC adults) to 44%
(ChWC adults) in the input to Chilean children, providing Chilean children with a fair
amount of ambiguous input. Given these diﬂ‘erences in input and the fact that all three
child groups patterned differently in their own production, the next important question is
whether MexWC children, ChMC children and ChWC children will pattern differently in
their comprehension of the plural morpheme. Or, is the fact that the plural is marked
between 50% - 65% of the time enough for Chilean child subjects to associate the plural
morpheme to an interpretation of ‘more than one’. This question will be addressed in a

series of experimental studies on comprehension presented in the following chapter.

150

CHAPTER 5

COMPREHENSION OF PLURAL MORPHOLOGY

5.0 Introduction

The production data presented in Chapter 4 showed that Chilean children were
variable in their production of the plural morpheme while Mexican children were
systematic, always producing the plural morpheme as an alveolar fricative. Within the
Chilean child groups, the data showed that ChWC children omitted the plural morpheme
much more often than ChMC children. This data alone suggests very strongly that
ChWC, ChMC and Mexican children are receiving very different linguistic input with
respect to the plural morpheme. The production data collected from Mexican and Chilean
adults conﬁrm that differences in the input between the three groups exist. However, the
data also show that Chilean children, although variable in their production of the plural
morpheme, have not yet acquired all of the various linguistic and extra-linguistic
constraints governing that variability. These ﬁndings are consistent With previous
research on child acquisition of linguistic variability, which shows that variable input
results in variable production in children, yet even by 6 years of age children have not
acquired all of the constraints governing the variability (Kovac and Adamson 1981 ,
Roberts 1994, 1997, Smith et al. 2006).

What about comprehension of grammatical forms that are variable in the input?
Chilean and Mexican children differ in their production of the plural morpheme, will they
also differ in their comprehension of the plural morpheme? The production data
presented in Chapter 4 showed that both Chilean and Mexican subjects produced the

plural morpheme some of the time, with the difference being that the Chilean subjects

151

produced the plural morpheme as [s] and also as [h] and they also omitted the plural
morpheme some of the time. The production data show that Chilean children receive an
input with [s] as the plural morpheme some of the time. Will they associate [s] to an
interpretation of ‘more than one’? The production data also show that Chilean children
receive an input with [h] as the plural morpheme some of the time. Will they also
associate [h] to an interpretation of ‘more than one"? Or, will the ambiguous nature of the
input due to the omission of the plural morpheme, delay Chilean child comprehension of
plural morphology?

The Variability Delay Hypothesis, which was presented in Chapter 1 of this
dissertation and is restated below in (1) predicts that Chilean children will differ from
Mexican children in their comprehension of the plural morpheme. More speciﬁcally, it
predicts that Chilean children will have a delay in their comprehension of plural

morphology.

(1) Variability Delay Hypothesis (based on Yang 2000, 2002): Variability in
the input will delay child comprehension of grammatical morphemes when the variability
causes an ambiguity in the input (involves a zero form) and is constrained not only by

linguistic (phonological, grammatical) but also extra-linguistic (SES, age, sex) factors.

The Variability Delay Hypothesis is adapted from Yang’s (2000, 2002) Variation Model
of language acquisition, which proposes an approach to language acquisition that is based
on a competition between a finite set of grammars. The set of grammars in competition is

determined by the interaction between the biological constraints on human grammar

152

(UG) and the properties of linguistic data in the environment during the course of
language acquisition. His competition-based model proposes that each grammar is
associated with a weight, which denotes the probability with which the child acquiring
the language accesses that particular grammar. When the child is presented with an
utterance, he selects a grammar in order to analyze the utterance. The probability that a
particular grammar is selected is based on its weight. If the child can parse the utterance
with this grammar, the grammar is rewarded and all of the other grammars are indirectly
punished. If the grammar is unable to parse the utterance, it is punished and all the other
grammars are indirectly rewarded By examining the relative ﬁequency of structures
within a sample of input data, it is possible theoretically to determine which grammars
will not be able to parse the structures in the input and hence end up being grammars that
are punished and not selected later on. Ambiguous input would cause a delay in
acquisition, as the child would be rewarding more than one grammar.

With respect to plural morphology, I do not assume that Chilean children have
two competing grammars, one with a projection above DP where number is interpreted
and one that does not have a projection above DP. Instead, the problem that Chilean
children might face is to learn whether the underlying representation associated to ‘more
than one’ is overtly marked or not. In other words, the issue is matching different
phonological forms to an underlying representation of number and that the competition
occurs between the two or more phonological forms that will ultimately be matched to
the underlying representation. Phonological forms associated with number that are
systematically produced in the adult speech (e. g. lexical quantiﬁers, numerals) will be

matched to the underlying form more quickly than ambiguous phonological forms (i.e.

153

those that are sometimes produced and sometimes omitted in the input of adult speakers
to children, as is the case for the Chilean plural morpheme). Ultimately, the phonological
form that is initially matched to the underlying representation will affect the grammar the
child is constructing. Either the child initially constructs a grammar that marks number
morphologically, as in English, or the child initially constructs a grammar that marks
number lexically, as in Chinese. As a result, the hypothesis predicts a delay in the
acquisition of grammatical forms (i.e. mapping of those forms to an underlying
representation) that are ambiguous.

At this point, it is important to note that any results we ﬁnd for Chilean children
will be inconclusive unless we also test the comprehension of plural morphology by
Mexican children, who are exposed to systematic input. Until we know what the general
patterns for the comprehension of Spanish plural morphology are when the input is
systematic, we cannot interpret the results for Chilean children. For this reason, both
Chilean and Mexican comprehension data is needed in order to answer the questions

posed above.
5.1 Study 1. Act-out Task: Indeﬁnite Noun Phrase

5.1.1 Background

Study 1 investigates Chilean and Mexican children’s interpretation of plural and
singular indeﬁnite noun phrases like those in sentences (2) and (3) by examining
children’s ability to give plural responses based on the type of indeﬁnite determiner and

the presence or absence of the plural morpheme.

154

(2) a. Pon una bolita en la caja.
Put a/one.SG marble.SG in the box

‘Put a/one marble in the box.’

b. Pon unas bolitas en la caja.
Put some.PL marbles.FL in the box

‘Put some marbles in the box’

c. Pon alggas bolitas en la caja.
Put some.PL marbles.FL in the box

‘Put some marbles in the box.’

(3) a. Pon unit de las bolitas en la caja.
Put a/one.SG of the.PL marbles.FL in the box -

‘Put alone of the marbles in the box.’

b. Pon algu_nas de l_as bolita_s en la caja.
Put some.PL of the.PL marble.PL in the box

‘Put some of the marbles in the box.’

Sentences (2a) and (3a) are associated to an interpretation of ‘one’ while sentences (2b),

(2c), and (3b) are associated to an interpretation of ‘more than one’. Both masculine and

feminine noun phrases were examined. For feminine nouns only the plural morpheme

155

provides number information my.SG bolita.F.SG vs. MPH bolitas.F.PL). For
masculine noun phrases the form of the indeﬁnite determiner is different in the plural vs.
the singular condition (_u_n_.M.SG auto.M.SG vs. mMJ’L autos.M.PL). The singular
indeﬁnite noun phrase in (23) also has a number interpretation: it can mean ‘a marble’ or
‘one marble’. Given the context of Study 1, both plural indeﬁnites unos and algunos are
felicitous and require a ‘more than one’ interpretation while the singular indeﬁnite un
requires an interpretation of ‘one’. The experimental sentences in (3) test overt partitives.
While the nouns inside the prepositional phrase in the partitive constructions in (3a) and
(3b) are both marked with a plural morpheme, (3a) is associated to an interpretation of
‘one’ and (3b) is associated to an interpretation of ‘more than one’. It should be noted
that while the Spanish adult grammar allows the singular form of (3b) (alguno de [as ‘one
of the’), it does not allow a plural form of (3a) (*unos de [as ‘some of the’).

As noted above, the input with respect to plural morphology is different for
Chilean working-class children (ChWC) vs. Chilean middle-class children (ChMC) vs.
Mexican working-class children (MexWC); MexWC children receive systematic input
with respect to plural morphology and Chilean children receive variable input. Table 27
from Chapter 4 is placed again here to remind the reader of the overall differences in

plural morpheme production between the three varieties of Spanish.

156

Table 27. Percentage of Plural Responses for Production Data.

 

Morpheme Omission Total Plural

([s] or [h]) (ﬂ) Tokens
ChMC Adults 67% (433) 33%‘(213) 100% (646)
Children 83% (278) 17% (58) 100% (336)
ChWC Adults 56% (145) 44% (113) 100% (258)
Children 42% (212) 58% (287) 100% (499)
MexMC Adults 99% (335) 1% (5) 100% (340)
MexWC Adults 99% (430) 1% (3) 100% (433)
Children 99% (532) 1% (3) 100% (535)

 

 

Study 1 asks the following questions: (1) Given the differences in the input, will Chilean

children differ ﬁom Mexican children in their comprehension of indeﬁnite plural noun

phrases? (2) Will Chilean children with variable input associate the plural morpheme to

an interpretation of ‘more than one’? (3) Will Chilean children treat both [s] and [h] as

associated to an interpretation of ‘more than one’. (4) Given that the plural masculine

noun phrase has a different form than the singular masculine noun phrase (e. g. un auto

vs. unos autos), will children perform better on masculine noun phrases than on feminine

noun phrases?

5.1.2 Method and Design

An Act-out Task was used. Subjects were presented with sets of miniature toys (6

toys per set) followed by sentences as in (2) — (3) above. There were three plural targets

(2b, 2c, 3b) and two singular targets (2a, 3a). The noun phrases under investigation were

157

always in object position, rather than in subject position where they would trigger
subject-verb agreement. Placing the noun phrase in object position, guarantees that only
the plural morpheme on the noun provides number information. Furthermore, placing the
object in sentence ﬁnal position, guarantees very little memory burden for the child. The
task procedure was very simple and matched very closely to children’s everyday
language experiences. Ferenz and Prasada (2002) showed that children as young as 1;9
years of age can carry out this type of task with very little difﬁculty.

Because Cepeda (1995) found that syllable ﬁnal /s/ was omitted more often in
certain phonological environments, the initial sound of the nouns was controlled for in
order to include environments where the determiner preceded sounds that favored both

omission and production of the plural morpheme. Cepeda’s ﬁndings are shown again in

Table 28.

Table 28. Lenition of Syllable Final /s/ by Phonological Environment.

 

 

 

 

_#C _#C _#C _#V _#V _#

[-cont] [+cont] [+nas] (U nstressed) (Stressed) (Pause)
[s] 1% 1% 1% 6% 25% 12%
[h] 83% 55% 71% 42% 38% 35%
Omission 16% 44% 28% 52% 36% 54%
Total 3594 3552 1406 2716 1018 3831
*Adapted from Cepeda (1995).

The nouns used in this study with Chilean children were: araﬁas ‘spiders’ (unstressed

vowel), autos ‘cars’ (stressed vowel), bolitas ‘marbles’ (+ continuant consonant), monitos

158

‘monkeys’ (nasal consonant) and for Mexican children: araiias ‘spiders’, anillos ‘rings’,
vacas ‘cows’, perros ‘dogs’. Because the names for these items differ across the two
dialects, the same items could not always be used (e. g. in Chile monos but in Mexico
changos means ‘monkey’). -

In addition, gender was controlled for. Two nouns were feminine and the other
two were masculine for each indeﬁnite condition. For the feminine indeﬁnites, only the
plural morpheme provides number information (4a — 4b). For the masculine indeﬁnites

the form of the determiner is also different in the singular vs. plural conditions (5a — 5b).

(4) a. una bolita
a/one.F.SG marble.F.SG

‘a/one marble’

b. unas bolitas
some.F.PL marbles.F.PL

‘some marbles’
(5) a. un mono

a/one.M.SG monkey.M.SG

‘a/one monkey’

159

b. unos monos
some.M.PL monkeys.M.PL

‘some monkeys’

Chilean children were tested either with the plural morpheme pronounced as [s] or as [h]
while Mexican children were only tested on the alveolar fricative.

Controls involved the quantiﬁers muchos (many.PL), pocos (few.PL), and todos
(all.PL). While todos (all.PL) requires an interpretation of ‘more than one’ where children
put the maximal set of items in the box, muchos (many.PL) requires a ‘more than one’
interpretation yet the maximal set of items is not required, but optional. Finally, although
pocos (few.PL) has a plural morpheme, in this context it is possible to respond with a
‘more than one’ or ‘one’ interpretation (Pon pocas bolitas en la caja “Put few marbles in
the box”). Subjects could perform perfectly on the controls even if they ignore the plural
morpheme. In addition, all children were tested on the numbers dos (two) and tres (three)

and their ability to count to seven.

5.1.3 Subjects

89 children participated in this experiment. For 19 MexWC (4:7-6:5), 36 ChWC
(4;7-7;3), 34 ChMC (4;8-8;2) children and 10 Chilean adults (5 ChWC and 5 ChMC) and
6 Mexican adults (3 MexWC and 3 MexMC) the plural morpheme was always
pronounced by the researcher as [s] in the experiment. In addition, for 11 ChWC (4:9-
6;4) and 9 ChMC (5;1-6; 1) children the researcher pronounced the plural morpheme as

[h] in the experiment. The plural morpheme was pronounced as [h] only for Chilean

160

children, and not Mexican children, given that Chilean adults, but not MexiCan adults,
sometimes pronounce the plural morpheme as [h].

The Chilean children were recruited from preschools and daycares in Punta
Arenas, Chile and the Mexican children from a daycare in Mexico City. All ChWC,
ChMC and MexWC adult subjects were the parents of children who participated in this
study. Three additional MexMC adult subjects were interviewed to ensure that MexWC
and MexMC adult comprehension of the plural morpheme did not differ. Children were
divided into two age groups in the version of the experiment where the researcher
pronounced the plural morpheme as [s], an older and younger age group. In the version of
the experiment where the researcher pronounced the plural morpheme as [h], children
were not divided by age group. This decision was based primarily on the fact that there
were not enough subjects to divide for age. Table 29 shows the distribution of children

who participated in Study 1.

161

Table 29. Study 1: Distribution of Children.

 

Pronunciation Group Number Age Range Mean Age
MexWC [s] Younger 14 4;7-5; 10 5;2
[s] Older 5 6;0-6;5 6:3
ChMC [s] Younger 15 4;8—5;1 l 5;2
[s] Older 10 6;0-7;3 6:7
[h] ---- 9 5;1-6;1 5;7
ChWC [s] Younger 15 4;7-5;l l 5;3
[s] Older 10 6;7-8;2 7:4
[h] ----- l 1 4;9-6;4 5;6

 

 

The distribution and mean age of the younger-[s] children (younger children who

heard the plural pronounced as [s] in the experiment) is similar across all three child

groups. All of these children were in preschool or kindergarten and had not yet received

any formal reading education. However, this is not true for the older children, where the

distribution and mean age differs across the three groups, with the older MexWC children

being approximately 4 — 13 months younger than the older ChMC and ChWC children

(based on mean age). In addition, although the older MexWC children were in

kindergarten and had not yet received any reading instruction, the Chilean children were

in ﬁrst and second grade and had received reading instruction. For this reason, any

162

comparisons made between the older child groups, must take into consideration the

different levels of education and age.

5.1.4 Procedure

All subjects were tested by native Spanish speakers who lived in the same city as
the subjects. The author of this dissertation was present during the testing of all children
to ensure that procedures were identical for all subjects. The experimenter started the
testing session by placing 7 miniature puppies in a row in front of the child and asking
the child to count the puppies. Children were then asked to place two puppies (dos
perritos) and then three puppies (tres perritos) in a small box. Of the 91 children invited
to participate in this study, there were only two children who did not pass this warm-up
task, leaving a total of 89 child subjects. Next, the experimenter presented each child with
two sets of objects (autos ‘cars’ and bolitas ‘marbles’) and told each child: “Te voy a
decir cudntas bolitas y cudntos autos tienes que poner en la caja y tu panes la cantidad
que yo te digo, te parece? ” (“I’m going to tell you how many marbles and how many
cars you have to put in the box and you put in the quantity that I tell you, sound good?”)
The experimenter then began nmning the main section of the experiment, which involved
target and control conditions. The test was administered the same way for adults;
however, they were not tested on their ability to count, number terms nor controls.

A within subjects design was used. There were four trials of each target sentence
and four trials of each control sentence, making a total of 32 experimental sentences. All
experimental sentences were presented in pseudorandom order and sentences involving

the same determiner never immediately followed each other.

163

In addition to this task, there was a follow-up highlighting task that was
administered only to children who consistently responded incorrectly in the plural
conditions unos, algunos, or algunos de los by associating the plural indeﬁnite to an
interpretation of ‘one’ (by only putting one item in the box). Children who provided a
singular response in at least 3 out of 4 trials in any one of the three plural conditions were
presented with the highlighting task illustrated in Table 30. The goal was to ensure that
children were focused on the phonological difference between the singular and plural
nouns. Note that this Highlighting Task only highlighted the plural morpheme on the

plural indeﬁnite unos, but not on algunos nor algunos de 103.

Table 30. Study 1: Highlightinglask.
Researcher: “Ahora escucha bien. SSSSSSS BOLITASSS tiene ‘s’ al ﬁnal.

 

BOLITA y BOLITAS. LTe das cuenta? A ver, BOLITA. Repitelo.
(child repeats) y BOLITAS (child repeats). Ahora, Pon UNAS
BOLITAS en la caja (child responds). Y ahora pon UNA BOLITA en

la caja (child responds)”

Researcher: “Now listen. SSSSSS MARBLESSS has an ‘s’ at the end. MARBLE
and MARBLESSSS, Do you hear that? Let’s see, MARBLE. Repeat it
(child repeats) and MARBLESSS (child repeats). Now, Put SOME
MARBLES in the box (child responds). And now put A/ONE

MARBLE in the box (child responds)”

 

164

5.1.5 Results

There were no differences found between Chilean and Mexican adults. Both adult
groups always associated the plural indefmites to an interpretation of ‘more than one’ and
the singular indeﬁnites to an interpretation of ‘one’. Although all three child groups
performed the same on the controls, they did not all perform the same in the target
conditions. The dependent variable was the number of plural responses children gave.
Putting more than one item into the box was considered a plural response. Putting only
one item in the box was considered a singular response. The mistakes that children made
in the plural conditions were always the same, they only put one item in the box. Table

31 shows the overall percentage of plural responses for control conditions.

165

Table 31. Study 1: Plural Responses in Control Conditions.

 

Plural Morpheme Pronounced as [s]

YOUNGER todos muchos pocos
‘all.PL’ ‘many.PL’ ‘few.PL’
MexWC 100% 100% 75%
(56/56) (56/56) (42/56)
ChMC 100% 100% 80%
(60/60) (60/60) (48/60)
ChWC 100% 100% 73%
(60/60) (60/60) (44/60)
OLDER todos muchos pocos
‘all.PL’ ‘many.PL’ ‘few.PL’
MexWC 100% 100% 70%
(20/20) (20/20) (14/20)
ChMC 100% 100% 75%
(40/40) (40/40) (30/40)
ChWC 100% 100% 70%
(40/40) (40/40) (28/40)

Plural Morpheme Pronounced as [h]

todos muchos pocos
‘all.PL’ ‘many.PL’ ‘few.PL’

ChWC 100% 100% 82%
(44/44) (44/44) (36/44)

ChMC 100% 100% 61%
(36/36) (36/36) (22/36)

 

166

All three child groups in both the [s] and [h] version of the study assigned plural readings
to the controls todos (‘all.PL’) and muchos (‘many.PL’) 100% of the time. Although there
was a tendency to assign plural readings to pocos (‘few.pl), this was not required by the
context and a one-way AN OVA showed no signiﬁcant differences between the three
younger-[s] child groups (F(2,42)=. 147, p = .864), nor the three older-[s] child groups
(F(2,24)=.052, p = .949), nor between the ChWC and ChMC children tested on [b]
(F(1,19)=1.084, p = .312). While children performed similarly in the Control conditions,
they did not perform the same in the Target conditions. Table 32 shows the percentage of

plural responses in the target conditions.l9

 

'9 Phonological environment of the nouns used in the experiment did not affect child
comprehenesion of the plural morpheme.

167

Table 32. Study 1: Plural Responses in Target Conditions.
alg de 103 uno de las
unos an algunos some.PL one.SG of
some.PL alone.SG some.PL of the.PL the.PL

 

 

Adults 100% 0% 100%' 100% 0%
(64/64) (0/64) (64/64) (0/64) (0/64)

Plural Morpheme Pronounced as [s]
Younger Children

MexWC 77% 0% 77% 52% 14%
(43/56) (0/56) (43/56) (29/56) (8/56)

65% 0% 88% 87% 38%
ChMC (39/60) (0/60) (53/60) (52/60) (23/60)

32% 0% 38% 35% 10%
ChWC (19/60) (0/60) (23/60) (21/60) (6/60)

Plural Morpheme Pronounced as [8]
Older Children

95% 0% 100% 90% 10%
MexWC ( 19/20) (0/20) (20/20) (1 8/20) (2/20)

90% 0% 90% 88%] 0%
ChMC (36/40) (0/40) (36/40) (3 5/40) (0/40)

75% 0% 78% 70% 8%
ChWC (30/40) (0/40) (3 1/40) (2 8/40) (3/40)

Plural Morpheme Pronounced as [h]

31% 0% 53% 25% 8%
ChMC (11/36) (0/36) (19/36) (9/36) (3/36)

23% 7% 36% 39% 28%
ChWC (10/44) (3/44) (16/44) (17/44) (10/44)

 

Within every group (both Mexican and Chilean, older and younger), there were children

who associated the plural indeﬁnites unos (‘some.PL’), algunos (‘some.PL’), and algunos

168

de los (‘some.PL of the.PL’) to an interpretation of ‘more than one’; however, the groups
differed signiﬁcantly in how often they associated the plural indeﬁnite to an
interpretation of ‘more than one’.

Unos. The number of plural responses for each child was examined for the three
plural conditions: unos, algunos, and algunos de los. In the unos condition a one-way
ANOVA showed signiﬁcant differences between younger-[s] ChWC, younger-[s] ChMC,
younger-[s] MexWC and Adult subjects (F(3,59)=10.385,p<.001). Post hoc Bonferroni
tests showed that the only younger-[s] groups that behaved signiﬁcantly different from
adults were the younger-[s] ChWC children (p<.001) and younger-[s] ChMC children
(p<.05). Mexican children did not behave signiﬁcantly different from adults (p=.434).
Post hoc Bonferroni tests also showed for the younger-[s] child groups that only younger-
[s] MexWC and younger-[s] ChWC children differed signiﬁcantly from each other
(p<.01). Younger-[s] ChMC children did not behave signiﬁcantly different from
younger-[s] MexWC nor younger-[s] ChWC children (ChWC vs. ChMC: p=.065,
MexWC vs. ChMC: p=l .0). Turning now to the older-[s] children groups a one-way
ANOVA showed no signiﬁcant differences between the older-[s] ChWC, older-[s]
ChMC, older-[s] MexWC and Adult groups (F (3,40)=1 .890,p=. 148). Finally, if we look
at the ChWC and ChMC tested on [h], we ﬁnd no signiﬁcant differences between these
two groups (t(1,18)=-.433,p=.633). Child responses were not compared to adult responses
given that we did not test adults on [h].

Given that we ﬁnd no differences between the ChWC vs. ChMC children when
the plural is pronounced as [s] or when it is pronounced as [h], the next step is to

determine whether there are signiﬁcant differences in the younger ChWC and the

169

younger ChMC child behavior when the plural is pronounced as [s] as compared to [h].
For the ChMC children, there is a marginally significant decrease in plural responses on
unos (‘some.PL’) when the plural is pronounced as [h] (t(1,22)=2.027,p=.055). For the
ChWC children, although there was a decrease in the amount of plural responses
provided when the plural was pronounced as [h], this difference was not signiﬁcant
(t(1,24)=.528,p=.602). In other words, neither ChWC nor ChMC children improved on
unos (‘some.PL’) when the plural morpheme was pronounced as [h].

To determine whether there was an age effect, the younger-[s] children were
compared to the older-[s] children. The results show an age effect for the ChWC children
(t(1,23)=-2.418,p<.05) but no age effect for ChMC children (t(1,23)=-1.629,p=.117) and
no age effect for MexWC children (t(1,17)=-l.137,p=.271).

Finally, in order to determine whether the same children always treated the plural
indeﬁnite unos as plural or whether behavior was variable for each child, we examined
the systematicity in response patterns on unos for each child, as shown in Table 33. We
are deﬁning systematic behavior as having the same response (plural or singular) in at
least 3 out of 4 trials. Variable comprehension would be having a singular response for 2

of the 4 trials and a plural response for the other 2 of 4 trials.

170

Table 33. Study 1: Unos: Systematic Responders.

 

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response
(3-4 plural) J34 singular) Total (2 pl/ 2 sgL
Plural Pronounced as [s]
Younger
MexWC 86% 14% 100% 0%
ChMC 53% 33% 86% 13%
ChWC 27% 67% 94% 6%
Older
MexWC 100% 0% 100% 0%
ChMC 90% 10% 100% 0%
ChWC 70% 20% 80% 10%
Plural Pronounced as [h]
ChMC 22% 56% 78% 22%
ChWC 27% 73% 100% 0%

 

Table 33 shows that children in general were consistent in their responses. Children either

always selected a plural set of objects or a single object in at least 3 out of 4 trials. In

other words, they always treated the plural unos (‘some.pl’) as plural or always as

singular. Very few children treated it as plural half the time (in 2 trials) and as singular

half the time (in 2 trials).

171

Algunos. In the algunos condition a one-way AN OVA showed signiﬁcant
differences between younger-[s] ChWC, younger-[s] ChMC, younger-[s] MexWC and
Adult subjects (F(3,59)=13. 162,p<.001). Post hoc Bonferroni tests showed that only the
younger-[s] ChWC children behaved signiﬁcantly different from adults (p<.001).
Younger-[s] ChMC (p=l .0) and younger-[s] MexWC children (p=.191) did not behave
signiﬁcantly different from adults. Post hoc Bonferroni tests also showed that while the
younger-[s] ChWC children behaved signiﬁcantly different from both younger-[s] ChMC
(p<.001) and MexWC children (p<.01), the ChMC and MexWC children did not behave
differently from each other (p=l .0). Turning now to the older-[s] children groups a one-
way ANOVA showed no signiﬁcant differences between the older-[s] ChWC, older-[s]
ChMC, older-[s] MexWC and Adult groups (F (3,40)=1 .751,p=. 174). Finally, if we look
at the ChWC and ChMC tested on [b], we ﬁnd no signiﬁcant differences between these
two groups (t(1,18)=-.649,p=.524). Child responses were not compared to adult responses
given that we did not test adults on [h].

Next we will determine whether there are signiﬁcant differences in the younger
ChWC and younger ChMC child behavior when the plural is pronounced as [s] as
compared to [h]. For the ChMC children, there is a signiﬁcant decrease in plural
responses on algunos (‘some.PL’) when the plural is pronounced as [h]
(t(l,22)=2.423,p<.05). For the ChWC children, although there was a decrease in the
amount of plural responses provided when the plural was pronounced as [h], this
difference was not signiﬁcant (t(l,24)=.-.017,p=.987). In other words, neither ChWC nor
ChMC children improved on algunos (‘some.PL’) when the plural morpheme was

pronounced as [h], instead ChMC children had a decrease in plural responses.

172

To determine whether there was an age effect, the younger-[s] children were
compared to the older-[s] children. The results show an age effect for the ChWC children
(t(1,23)=-2.222,p<.05) but no age effect for ChMC children (t(1,23)=-.159,p=.875) and
no age effect for MexWC children (t(l,17)=-1.606,p=. 127).

Finally, in order to determine whether the same children always treated the plural
indeﬁnite algunos as plural or whether behavior was variable for each child, we
examined the systematicity in response patterns on algunos for each child, shown in
Table 34. We are deﬁning systematic as having the same response (plural or singular) in
at least 3 out of 4 trials. Variable would be having a singular response in 2 of 4 trials and

also a plural response in 2 of 4 trials.

173

Table 34. Study 1: A Lgunos: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(34 plural) (3-4 singular) Total (2 pl/ 2 sg)_

 

Plural Pronounced as [s]

Younger
MexWC 79% 14% 93% 7%
ChMC 93% 7% 100% 0%
ChWC 33% 60% 93% 7%
Older
MexWC 100% 0% 100% 0%
ChMC 90% 10% 100% 0%
ChWC 80% 20% 100% 0%
Plural Pronounced as [h]
ChMC 55% 45% 100% 0%
ChWC 37% 54% 91% 9%

 

The results show that overall children are systematic in their response patterns, either
always associating algunos with an interpretation of ‘more than one’ or always
associating algunos with an interpretation of ‘one’.

Algunos de los. In the algunos de los condition a one-way ANOVA showed

signiﬁcant differences between younger-[s] ChWC, younger-[s] ChMC, younger-[s]

174

MexWC and Adult subjects (F (3,59)=12.627,p<.001). Post hoc Bonferroni tests showed
that the only younger-[s] groups that behaved signiﬁcantly different from adults were the
younger-[s] ChWC children (p<.001) and the younger-[s] MexWC children (p<.001).
Younger-[s] ChMC (p=l .0) did not behave signiﬁcantly different from adults. Post hoc
Bonferroni tests also showed that while the younger-[s] ChMC children behaved
signiﬁcantly different from both younger-[s] ChWC (p<.001) and MexWC children
(p<.05), the ChWC and MexWC children did not behave differently from each other
(p=1 .0). In other words, younger-[s] ChWC and MexWC children performed the same on
algunos de 10s and differently from the ChMC children. Turning now to the older-[s]
children groups a one-way ANOVA showed signiﬁcant differences between the older-[s]
ChWC, older-[s] ChMC, older-[s] MexWC and Adult groups (F (3,40)=2.941,p<.05).
Post hoc Bonferroni tests showed that only older-ChWC children behaved signiﬁcantly
different from adults (p<.05) while older-ChMC (p=l.0) and older-MexWC (p=l.0)
children did not. However, the three older-child groups did not differ signiﬁcantly
different from each other (older-ChWC vs. older-ChMC: p=.765; older-ChWC vs. older-
MexWC: p=.925; older-ChMC vs. older-MexWC: p=l.0).

The next step is to determine whether there are signiﬁcant differences in the
younger ChWC and younger ChMC child behavior when the plural is pronounced as [s]
as compared to [h]. The plural overt partitive will sound different in the [s] and [h]
studies because both the indeﬁnite algunos (‘some.PL’) and the deﬁnite determiner 10s
(‘the.PL’) will both occur with either [s] or with [h]. For the ChMC children, there is a
signiﬁcant decrease in plural responses on algunos de las (‘some.PL of the.PL’) when the

plural is pronounced as [h] (t(1,22)=4.871,p<.001). For the ChWC children there is no

175

difference in plural responses on algunos de [as when the plural is pronounced as [s] or
[b] (t(1,24)=-.077,p=.939). In other words, neither ChWC nor ChMC children improved
on algunos de los (‘some.PL of the.PL’) when the plural morpheme was pronounced as
[h], rather the ChMC children provided signiﬁcantly less plural responses when the plural
was pronounced as [h]. This is interesting because it suggests that ChMC children prefer
[s] as the plural morpheme and some are treating algunoh de loh (with aspiration) as if
there is not plural morpheme there, as if it is alguna de los (‘one.SG. of the.PL’), which is
possible in adult Spanish.

To determine whether there was an age effect in the algunos de los condition, the
younger-[s] children were compared to the older-[s] children. The results show a
marginal age effect for the ChWC children (t(l,23)=-2.006,p=.057) but no age effect for
ChMC children (t(1,23)=-.074,p=.942) and no age effect for MexWC children (t(1,17)=-
1.910,p=.073).

Finally, in order to determine whether the same children always treated the plural
indeﬁnite algunos de [as as plural or whether behavior was variable for each child, we
examined the systematicity in response patterns on algunos de [OS for each child, as
shown in Table 35. We are deﬁning systematic as having the same response (plural or
singular) in at least 3 out of 4 trials. Variable plural responses are when children assigned

a plural reading in 2 of 4 trials and also a singular reading in 2 of 4 trials.

176

Table 35. Study 1: Algunos de los: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(3-4 plural) (3-4 singular) Total (2 pl/ 2 sg)_

Plural Pronounced as [s]

Younger
MexWC 43% 43% 86% 14%
ChMC 87% 7% 94% 7%
ChWC 33% 60% 93% 7%
Older
MexWC 100% 0% 100% 0%
ChMC 80% 10% 90% 10%
ChWC 60% 20% 80% 20%

Plural Pronounced as [h]
ChMC 11% 78% 89% 1 1%

ChWC 36% 63% 100% 0%

 

Table 35 shows that for the most part all children were systematic in their responses,
either always associating the plural partitive to a ‘more than one’ interpretation or always
associating it to an interpretation of ‘one’. The largest percentage of variable responders
come from the Older-[s] ChWC and Younger MexWC groups where 2 children in each

group were variable in their responses.

177

Turning next to the two singular conditions, un and uno de los, we ﬁnd that all
groups performed at ceiling on the singular indeﬁnite condition un and were systematic

in their responses, as shown in Table 36.

Table 36. Study 1: Un: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(3-4 plural) (3-4 singular) Total (2 pl/ ZSgL

 

Plural Pronounced as [s]

Younger
MexWC 0% 100% 100% 0%
ChMC 0% 100% 100% 0%
ChWC 0% 100% 100% 0%
Older
MexWC 0% 100% 100% i 0%
ChMC 0% 100% 100% 0%
ChWC 0% 100% 100% 0%

Plural Pronounced as [h]
ChMC 0% 100% 100% 0%

ChWC 0% 100% 100% 0%

 

178

Uno de los. For the younger-[s] children a one-way ANOVA showed signiﬁcant
differences between the younger-[s] ChMC, ChWC, and MexWC children and Adults
(F (3,59)=8.690,p<.001). Post hoc Bonferroni tests showed that only the younger-[s]
ChMC children (p<.001), but not the younger-[s] ChWC (p=.806) nor MexWC (p=.438),
performed signiﬁcantly different from adults. The younger-[s] ChMC children performed
signiﬁcantly differently from the other two younger-[s] child groups: ChWC (p<.01),
MexWC (.05). The younger-[s] ChWC and MexWC children did not behave signiﬁcantly
different from each other (p=l .0). For the older-[s] children a one-way ANOVA showed
no signiﬁcant differences between groups (ChWC, ChMC, MexWC, Adults)

(f(3,40)=1 .617,p=.202). Comparing Chilean children who participated in the part of the
study where the plural was pronounced as [h] we ﬁnd that ChWC and ChMC children did
not behave signiﬁcantly different from each other (t(1,18)=.978,p=.34l). Adults were not
included in this statistic because they were not tested on their comprehension of [h] as the
plural morpheme.

The next step is to see whether the Chilean children who heard the plural
morpheme pronounced as [3] performed differently than the Chilean children who heard
it pronounced as [h]. The difference in pronunciation of the plural morpheme affected the
singular condition uno de las (‘one.SG of the.PL’) in that the deﬁnite article [as in this
partitive is plural. However, the pronunciation of uno in the overt partitive was the same
in the [h] and [8] parts of the study. Although there was a slight increase in plural
responses in the [h] experiment for ChWC children, these differences were not signiﬁcant
(t(1,24)=-.928,p=.363). However there were signiﬁcant differences between ChMC

children who heard the plural morpheme pronounced as [s] vs. those who heard it

179

pronounced as [h] (t(1,22)=-2.370,p<.05), with [s]-ChMC children treating uno de 10s as
plural signiﬁcantly more often than [h]-ChMC children. It appears as if the ChMC
children are hypercorrecting, assuming that the researcher is omitting the plural
morpheme in the [8] part of the study but not the [h] part of the study.

Turning now to possible age effects, the results showed that there were no age
effects for the ChWC children (t(1,23)=.208,p=.837) nor the MexWC children
(t(l,17)=.527,p=.605) but there was as signiﬁcant age effect for the ChMC
(t(1,23)=2.798,p<.01). Younger-[s] ChMC children incorrectly associated uno de [OS to
the interpretation of ‘more than one’ more often than older-[s] ChMC children.

Finally, Table 37 shows the percentage of children who had systematic responses

in the uno de Ios condition.

180

Table 37. Study 1: Uno de los: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(3-4 plural) (3-4 singular) Total (2 pl/ 2 sgL

 

Plural Pronounced as [s]

Younger

MexWC 0% 93% 93% 7%
ChMC 20% 47% 67% 33%
ChWC 0% 87% 87% 13%

Older

MexWC 0% 100% 100% 0%
ChMC 10% 90% 90% 10%
ChWC 0% 100% 100% 0%

Plural Pronounced as [h]

ChMC 0% 89% 89% 1 1%
ChWC l 8% 73% 91% 9%

 

There were several younger-[s] ChMC children who were variable in their interpretation

of uno de los (‘one.SG of the.PL’). The variable responders in the younger-[s] group were:

1 MexWC, 5 ChMC, and 2 ChWC. In the older-[s] group only 1 ChMC child had a

variable response pattern. In the younger-[h] group only 1 ChMC and l ChWC child had

a variable response pattern. It is surprising that Chilean children would be variable in

181

their interpretation of uno de 10s given that the same children systematically treated
un/una (‘a/one’) as singular and had a tendency to incorrectly treat the plural unos
(‘some.pl’) as singular. At this point, I am not sure why we found more variable
responders in this condition. In any case, the majority of children tended to treat uno de
[as as singular. It may be that some children are sometimes focusing on uno (‘one.SG’)

and sometimes focusing on [OS (‘the.PL’) in their interpretation and ignoring the rest.

5.1.6 Highlighting Task

Of the younger children the Highlighting Task was administered to 9 ChWC, 4
ChMC and 3 MexWC children. None of the 9 ChWC children improved. All of the
ChMC children and one of the MexWC children improved. Of the older children the
Highlighting Task was administered to 2 ChWC, l ChMC and no MexWC children. The
ChWC children did not improve; the ChMC child did improve. The number of children
administered the Highlighting Task and who treated the plural indeﬁnite as plural are

shown in Table 38.

Table 38. Study 1: Highlighting Task: Plural Responses.

 

 

 

ChWC ChMC MexWC
Younger 0/9 (0%) 4/4 (100%) 1/3 (33%)
Older 0/2 (0%) 1/1 (100%) N/A

 

182

5.1.7 Feminine vs. Masculine Nouns

Given that the plural and singular feminine indefmites differ only by the presence
or absence of the plural morpheme (unas vs. una) while the form of the masculine plural
indeﬁnite is also different from that of masculine singular indeﬁnites (unos vs. an), it is
possible that children would be better at interpreting number on masculine indefmites
than on feminine indefmites because there would still be a difference between the
singular and plural forms even if syllable ﬁnal /s/ were omitted (un auto ‘a/one.SG car.SG’
vs. uno[Z] auto[Z] ‘some.PL cars.PL’]. However, this difference between masculine and
feminine indeﬁnites would not be relevant for the partitive constructions uno de [as and
algunos de los, given that the only difference between the singular and plural forms of
algunos de [as is the plural morpheme /s/ (alguno de las vs. algunos de los) and the plural
form of uno de Ios does not exist in adult Spanish (*unos de los). For the younger-[s]
ChWC children paired sample t-tests showed that for there were no signiﬁcant
differences in plural responses between masculine vs. feminine indeﬁnites in any
condition: unos vs. unas (t (1,13) =-1.472, p = .165), algunos vs. algunas (t (1,13) =
1.883, p = .082), uno de los vs. una de las (t (1,13) = -.366, p = .720), and algunos de 103
vs. algunas de las (t (1,13) = 1.883, p = .082). For the younger-[s] ChMC children paired
sample t-tests also showed no signiﬁcant differences between masculine and feminine
indefmites in any condition: unos vs. unas (t (1,13) =.366, p = .720), algunos vs. algunas
(t (1,13) = 1,147, p = .272), uno de 103 vs. una de las (t (1,13) = .366, p = .720), and
algunos de las vs. algunas de las (t (1,13) = 1.472, p = .165). For younger-[s] Mexican
working class children there were no signiﬁcant differences between masculine and

feminine noun phrases on unos vs. unas (t(l ,13)=.434,p=.671), algunos vs. algunas

183

(t(1,13)=.000,p=1 .0), uno de Ios vs. una de las (t(1,13)=.000,p=1.0), algunos de los vs.
algunas de las (t(l,13)=-.562,p=.583). There were also no signiﬁcant differences for
older-[s] ChWC between feminine and masculine nouns unos vs. unas (identical data
sets), algunos vs. algunas (t(1,9)=-1.0,p=.343), uno de los vs. una de las (t(1,9)=-
.557,p=.591), algunos de los vs. algunas de las (identical data sets) or for older-[s]
ChMC children unos vs. unas (identical data sets), algunos vs. algunas (identical data
sets), uno de los vs. una de las (t(1,9)=1.0,p=.343), algunos de los vs. algunos de las
(t(l,9)=-.429,p=.678) or for older-[s] MexWC children unos vs. unas (t(l,4)=-
1.0,p=.374), algunos vs. algunas (t(1,4)=-1.0,p=.374), uno de las vs. una de las
(t(1,4)=1.633,p=.178), algunos de los vs. algunas de las (t(l,4)=-1.633,p=. 178). Finally,
even when the plural morpheme was pronounced as [b] there were no signiﬁcant
differences between masculine and feminine nouns for the ChWC on unos vs. unas
(t(1,10)=1 .491,p=. 167), algunos vs. algunas (t(1,10)=l.491,p=.167), uno de los vs. una
de las (identical data sets), algunos de las vs. algunas de las (t(1,1'0)=.559,p=.588) nor
for the ChMC children on unos vs. unas (t(1,8)=1.000,p=.347), algunos vs. algunas
(t(1,8)=1 .000,p=.347), uno de Ios vs. una de las (t(1,8)=-1.000,p=.347), algunos de Ios
vs. algunas de las (t(1,8)=-2.000,p=.081). Table 39 shows the percentage of plural

responses in the feminine and masculine noun conditions.

184

Table 39. Study 1: Plural Responses in Feminine and Masculine Trials.

 

 

 

Younger Older Younger

Children [s] Children [s] Children [h]
COND CW C CMC MWC CWC CMC MWC CWC CMC

unos 33% 67% 79% 75% 90% 90% 27% 33%
(10/30) (20/30) (22/28) (15/20) (18/20) (9/10) (6/22) (6/ 18)

unas 30% 63% 75% 75% 90% 100% 18% 28%
(9/30) (19/30) (21/28) (15/20) (18/20) (10/ 10) (4/22) (5/18)

algunos 33% 93% 79% 75% 90% 100% 41% 56%
(10/30) (28/30) (22/28) (15/20) (18/20) (10/10) (9/22) (10/ 18)

algunas 43% 83% 75% 80% 90% 100% 32% 50%
(13/30) (25/30) (21/28) (16/20) (18/20) (10/10) (7/22) (9/ 18)

uno de los 10% 40% 14% 5% 0% 20% 23% 6%
(3/30) (12/30) (4/28) (1/20) (0/20) (2/10) (5/22) (1/18)

una de las 10% 37% 14% 10% 0% 0% 23% 6%
(3/30) (1 1/30) (4/28) (2/20) (0/20) (0/10) (5/22) (1/18)

alg de los 40% 90% 50% 70% 85% 80% 41% 17%
(12/30) (27/30) (14/28) (14/20) (17/20) (80/ 10) (9/22) (3/18)

alg de las 30% 83% 54% 70% 90% 100% 36% 33%
(9/30) (25/30) (15/28) (14/20) (18/20) (10/ 10) (8/22) (6/ 18)

 

5.1.8 Discussion

1 will ﬁrst discuss the results of each experimental group separately: (1) younger

children with plural pronounced as [s], (2) older children with plural pronounced as [s],

and (3) younger children with plural pronounced as [h] and then discuss differences

between the three groups.

The results for the younger-[s] groups revealed that only Mexican children were

able to reach adult levels in the unos condition, while both ChMC and MexWC children

185

performed at adult levels in the algunos condition. In the algunos de los condition, only
ChMC children reached adult levels. However, in any experimental task there are several
reasons for why younger children perform differently from adults (e.g. attention,
memory, task difﬁculty). Given that Mexican and Chilean children were of the same age,
had similar educational and economic backgrounds and were administered identical tests
and given the differences in the production of the plural morpheme in the input to each of
the three child groups, I assume that any differences found between the three groups is
associated primarily with the linguistic input that children are exposed to.

With respect to the plural indefmites, I assume that, if Spanish-speaking children
have an underlying representation for the plural morpheme, they should associate the
plural indeﬁnites to the interpretation of ‘more than one’. The results show that for
younger-[s] children, most MexWC children (12/14) associate unos to the interpretation
of ‘more than one’, while less ChMC children (8/ 15) and even fewer ChWC children
(4/ 15) do. On the other hand, most ChMC children (14/ 15) and MexWC children (1 1/ 14)
associate algunos with ‘more than one’ while very few ChWC children (5/15) do. At this
point, these ﬁndings suggest that there is a strong tendency by Chilean children in general
to treat the indeﬁnite unos as associated to and interpretation of ‘one’. I suggest that this
is due to the fact that the plural indeﬁnite unos/unas is almost identical in form to the
Spanish numeral for ‘one’ (uno/una). If Chilean children have not matched [s] or [b] to
an underlying representation for number, then the plural indeﬁnite becomes identical to
the numeral ‘one’. The presence of word ﬁnal [8] or [h] on unas/unos has no extra
meaning associated with it, in the same way that the ﬁnal non-morphological [s] or [b] on

Iapiz (‘pencil’) does not change the meaning of the word (e. g. [lapis], [lapih], [lapi] =

186

‘pencil’). Given that Chilean children perform well on the controls (quantiﬁers and
numerals) and tend to treat both the singular and plural forms of un/una as singular, I
suggest that they have initially constructed a grammar that relies on lexical items
(quantiﬁers and numerals) to assign number.

However, the results also showed that all ChMC children associated the plural
indeﬁnite unos/unas to the interpretation of ‘more than one’ when the plural morpheme
was highlighted. This indicates that the ChMC children, but not ChWC children, are
constructing a grammar, which matches [s] and [h] to an underlying representation for
number, yet still relies on lexical items (quantiﬁers and numerals) to assign number. As a
result, it seems that ChMC children have two phonological forms that are competing in
the interpretation they assign to the indeﬁnite plural: uno = ‘one’ vs. [s]/[h] = ‘more than
one’.

An unexpected ﬁnding for the younger-[s] group was that, while most ChMC

children (13/15) associated the overt plural partitive algunos de las with ‘more than one’,
only half of the MexWC children (6/14) and very few ChWC children (5/15) did. Both
the plural algunos de Ios (‘some.PL of the.PL’) and the singular alguno de los (‘one.SG of
the.PL’) are grammatical in adult Spanish, although the latter is not felicitous in this
context. It is unclear why Mexican children had difﬁculty with the plural overt partitive
given that the same children associated unos (‘some.PL’) and algunos (‘some.PL’) to
‘more than one’. However, because MexWC children are exposed to an input that
systematically marks the plural and yet many MexWC children did not associate the
plural algunos de Ios to an interpretation of ‘more than one’, it is difﬁcult to interpret the

results for the Chilean children. It was suggested that Mexican children may have found

187

that given the experimental context of only 6 items to choose from (e. g. 6 miniature
spiders) a singular response was sufﬁcient. Based on the data I collected, I will not be
able explain why Mexican children had difﬁculty in this condition and will instead leave
it to future research.

Another unexpected ﬁnding was in the uno de los (‘one.SG of the.PL’) condition.
Ferenz and Prasada (2002) showed that by 2;0 years of age English-speaking children
associate singular partitive constructions, like ‘one of the marbles’, to an interpretation of
‘one’ . However, the results presented here for Spanish-speaking children revealed that
while most of the younger-[s] MexWC children (13/14) and ChWC children (13/15)
associated the singular overt partitive to ‘one’, several ChMC children (7/15) associated
it to a ‘more than one’ interpretation. The ﬁndings for ChMC children are especially
surprising given the fact that these same ChMC children associated the indeﬁnite singular
un/una to an interpretation of ‘one’. At this point, the behavior of the ChMC children is
unclear. This may be a case of hypercorrection. Given the presence of the plural
morpheme on the deﬁnite determiner in the partitive construction along with the fact that
the plural morpheme is often omitted in Chilean Spanish, ChMC children may have
interpreted uno de Ios/una de las as unos de los/unas de las. While the latter plural form
is ungrammatical in Spanish, Chilean children would not necessarily know this if the
plural is sometimes omitted in the input they are exposed to. The ﬁnding that MexWC
children (exposed to systematic plural marking) performed well in this condition
indicates that the behavior of the ChMC children may be associated to the variable nature

of the input to which they are exposed.

188

The plural morpheme is also pronounced as [h] some of the time in adult Chilean
speech. It is possible that Chilean children encode [h] as the plural morpheme instead of
[5]. However, the ﬁndings of Study 1 do not to support this idea. Both ChWC and ChMC
children incorrectly associate all three plural indefmites to the interpretation of ‘one’
more often when the plural morpheme is pronounced as [h], suggesting that they have not
encoded [h] as the plural morpheme.

One interesting ﬁnding for the Chilean children who did associate [h] to ‘more
than one’ was that these children would often correct the researcher’s aspirated
pronunciation of the plural morpheme. In other words, when the researcher read the
experimental sentence with aspiration, “Pon unah bolita/1 en la caja” (“Put some.PL
marbles.FL in the box”), ChMC children would often repeat the sentence back to the
researcher, as if correcting him, with [s] as the plural morpheme instead of [h], “Pon EMS.
Mtg en la caja”. This suggests that these ChMC children associate [h] to ‘more than
one’ but preferred the [s] pronunciation over the [h] pronunciation in the experimental
context.

Up until this point, I have only discussed the results of the younger children.
However, the results for the older children show that all older child groups perform at
adult levels in Study 1. This indicates that while MexWC children associate the plural
morpheme with ‘more than one’ by 4;7 years of age, ChWC children do not associate the
plural morpheme with ‘more than one’ until about 6;7 years of age, two years later.
Moreover, ChMC children hypercorrect on uno de los until approximately 6;0 years of
age. It important to note that not only are the Chilean children much older than the

Mexican children, but they also receive reading and writing instruction in school while

189

the Mexican children do not. In addition, two ChWC children (child 1: 6;10, child 2: 8;0)
still associated all three plural conditions (but not plural control conditions) with ‘one’
and did not improve in the Highlighting Task. This suggests that even at this age, some
Chilean children still have not encoded the plural morpheme. However, we must be
careful in our interpretation of the results as we did not test the older children on the
plural morpheme pronounced as [h].

Given that the experiment in Study 1 was administered to MexWC, ChMC and
ChWC children using the same methods and materials, it is quite interesting to see such
differences in child behavior across the three groups. A study that would have examined
only Chilean or only Mexican children would not have provided an adequate account of
plural morpheme acquisition in Spanish-speaking children in general, especially since
several dialects of Spanish have syllable ﬁnal /s/ lenition.

The ﬁndings from Study 1 support the Variability Delay Hypothesis, which
hypothesizes that variability in the input will delay child comprehension of grammatical
forms when the variability causes an ambiguity (involves omission) and is constrained
not only by linguistic (phonological, grammatical) but also extra-linguistic (SES, age,
sex) factors. MexWC children, who are exposed to systematic plural marking, differ from
ChWC and ChMC children, who are exposed to variable plural marking involving
omission. This indicates that the development of plural morphology in the presence of

systematic vs. variable input is not the same.

5.1.9 Summary of Findings
The following is a summary of the results for the comprehension of plural

morphology in Study 1.

190

l. Chilean children differ from Mexican children in their comprehension of the
plural morpheme in indeﬁnite noun phrases.

2. When the plural morpheme is systematically produced in the adult speech,
Mexican Spanish-speaking children associate the plural morpheme to an interpretation of
‘more than one’ by at least 4;7 years of age.

3. When the plural morpheme has a variable behavior and is also omitted in the
adult speech, ChWC children do not associate the plural morpheme in indeﬁnite noun
phrases to an interpretation of ‘more than one’ until approximately 6;7 years of age.

4. The ChMC children appear to have difﬁculty associating the plural indeﬁnite
unos to an interpretation of ‘more than one’, unless presented with the highlighting task,
suggesting, along with the ﬁndings for uno de Ios (one.SG of the.PL), that ChMC children
are beginning to associate the plural morpheme [s] with the interpretation of ‘more than
one’ by approximately 4;8 years of age but they still rely on other information (e. g.
quantiﬁers, numerals) to assign number to the noun phrase.

5. Almost all Chilean and Mexican children were systematic in their
interpretation of the plural and singular noun phrases.

6. The younger ChWC and ChMC children have difﬁculty associating both [s]
and [h] to an interpretation of ‘more than one’, with [h] resulting in more interpretations
of ‘one’ than [s].

7. The difference in form between the plural and singular masculine determiners
does not result in more correct responses in the masculine targets than on the feminine

targets.

191

5.2 Study 2. Picture Matching Task: Indeﬁnite Noun Phrases

Study 1 showed that both ChWC children and ChMC children, but not MexWC
children, associate the plural indeﬁnite unos to an interpretation of ‘one’ in an Act-out
Task. The purpose of Study 2 is to test child comprehension of unos using a different
experimental methodology (Picture Matching Task) in order to conﬁrm the ﬁndings of
Study 1. Moreover, unlike Study 1, I will test the same children on their comprehension
of both [3] and [h] in Study 2. The purpose of testing the same children on both variants
is to determine whether Chilean children, who do not associate the plural variant [s] to

‘more than one’, will associate the plural variant [h].

5.2.1 Background
Study 2 examines child comprehension of the singular and plural indefmites, as in

(6).

(6) a. gEn cual de las dos tarjetas hay una I botella?
In which of the two cards EXST a/one.SG bottle.SG

‘In which of the two cards is there a/one bottle?’
b. gEn cuél de las dos tarjetas hay unas botellas?
In which of the two cards EXST some.PL bottles.PL

‘In which of the two cards are there some bottles?’

The indeﬁnite in (6a) is singular and the indeﬁnite in (6b) is plural. The existential verb

hay (‘there is/there are’) was used because it does not carry number information. It can be

192

used with both plural and singular nouns. For this reason, the only number information in
(6a) and (6b) is the plural morpheme in the determiner phrase.

Study 2 asks the following questions: (1) Given the differences in the input, will
Chilean children differ from Mexican children in their comprehension of indeﬁnite plural
noun phrases, as they did in Study 1? (2) Will Chilean children with variable input
associate the plural morpheme to an interpretation of ‘more than one’? (3) Will the same
Chilean children treat both [s] and [h] as associated to an interpretation of ‘more than
one’. (4) Given that the masculine plural noun phrase has a different form than the
singular masculine noun phrase (e. g. un auto vs. unos autos), will children perform better

on masculine noun phrases than on feminine noun phrases?

5.2.2 Method and Design

A Picture Matching task was used. Subjects were presented with pairs of pictures
(Figure 10) where one picture card had a singular object and the other card had 6 objects
of the same kind Children were asked questions like (7) and (8) in the context of Figure
10 and answered the question by pointing to the appropriate picture. See Appendix B for
full set of materials. Each child was tested on both the plural and singular indefmites and
for each pair of picture cards, they were only asked a question with either a plural
indeﬁnite noun phrase or a singular indeﬁnite noun phrase. The plural morpheme was
always pronounced as [s]. Chilean children who consistently chose the singular card in
the plural condition were tested a week later with the plural morpheme pronounced as
[b]. There were 4 trials of the plural condition, four 4 of the singular condition, and 4
ﬁllers ﬁ'om another experiment testing child comprehension of the Spanish copulas ser

and estar. In the 8 target trials the initial sound and the gender of each target word was

193

controlled for Chilean subjects: burros ‘donkeys’, manos ‘monkeys’, barcos ‘boats’,
martillos ‘hammers’, bolitas ‘marbles’, manzanas ‘apples’, botellas ‘bottles’, monedas
‘coins’ (see Study 1 for a discussion of how initial sounds were controlled for). The same
words were used for Mexican children except changos was used for ‘monkeys’ and
canicas was used for ‘marbles’ so that we could continue to use the same materials yet
accommodate to the Mexican Spanish lexicon. In addition, half of the indeﬁnites were
feminine and half were masculine. In the feminine indeﬁnites, only the plural morpheme
provides number information. In masculine indeﬁnites the form of the determiner is also

different in the singular vs. plural conditions.

ﬁgure 10. SM 2: Sample Ta_rgetTrit_11.

 

(7) (En cual de las dos tarjetas hay una botella?
In which of the two cards EXST a/one.SG bottle.SG?

...a ‘In which of the two cards is there alone bottle?’

 

(8) (En cual de las dos tarjetas hay unas botellas?
In which of the two cards EXST some.PL bottle.PL?

L
to u .
' if ‘ ' .
“1 J J ‘In whrch of the two cards are there some bottles?’

 

 

 

 

 

 

5.2.3 Subjects
50 children participated in this study: 19 MexWC (4;11-6;2, Mean: 5;4), 17

ChWC (4;9-6;4, Mean: 5;5), 10 ChMC (4;10-6;4, Mean: 5;5) children. In addition, 22

194

Chilean adults and 8 Mexican adults participated in this study. Table 40 shows the

distribution of the child subjects.

Table 40. Study 2: Distribution of Children.

Pronunciation Number Age Range Mean Age

 

MexWC [s] 19 4;] l-6;2 5;4
ChMC [s]/[h] 10 4;10-6;4 5;5
ChWC [s]/ [h] l 7 4;9-6;4 5 ;5

 

 

The Chilean children were recruited from schools in Punta Arenas, Chile and the
Mexican children from a daycare in Mexico City. All children were in preschool and
kindergarten. Chilean adults were undergraduates at the Universidad de Magallanes in
Punta Arenas, Chile and the Mexican adults were undergraduates at the Universidad

Auténoma Metropolitana de Iztapalapa in Mexico City.

5.2.4 Procedure

All subjects were tested by native speakers of Spanish who lived in the same city
as the subjects. The author of this dissertation was present during the testing of all
subjects to ensure that procedures were identical for all subjects. Because of the
simplicity of the task, there was no warm-up and only two controls (one control involved
un solo ‘only one’ and the other muchos ‘many’) that were administered after children
ﬁnished all target questions. The controls were administered after the target questions so

that un solo ‘only one’ would not provide any information to the child about the

195

interpretation of un ‘a/one’. In addition, placement of cards (singular card above plural
card vs. plural card above singular card) was controlled for. Half of the children from
each group was presented with the singular card above the plural card and the other half
was presented with the plural card above the singular card. The order of presentation of
sentences was the same for all subjects. Children who consistently chose the singular
picture card in the plural condition were tested between 1 — 2 weeks later by a different

researcher who pronounced the plural morpheme as [h].

5.2.5 Results

There were no differences between Chilean and Mexican adults (all adults
performed correctly 100% of the time) so their scores were combined. Choosing the card
with multiple items was considered a plural response. Choosing the card with only one
item was considered a singular response. All children treated the controls un solo (‘only
one’) as associated to the interpretation of ‘one’ and muchos (‘many.pl’) as associated to
the interpretation of ‘more than one’ 100% of the time. Although all children performed
well on controls, they did not all perform the same in the target conditions. Table 41
shows the percentage of plural responses when the plural morpheme was pronounced as

[s] for adults, MexWC children, ChMC children, and ChWC children.

196

Table 41. Study 2: Plural Responses in Target Conditions.

 

 

unos (‘some.PL’) un (‘a/one.SG’)

Adults 100% i 0%
MexWC 79% 6%

(60/76) (4/76)
ChMC 33% 0%

(12/40) (0/40)
ChWC 35% 1%

(24/68) (1/68)

 

 

Adults performed at ceiling on both the plural and singular indefmites and children
performed at ceiling on the singular indeﬁnite condition un (‘a/one.SG’). However, in the
plural indeﬁnite condition while MexWC children associated the plural indeﬁnite to an
interpretation of ‘more than one’, the ChMC children and ChWC children associated the
plural indeﬁnite to an interpretation of ‘one’, choosing the picture with only one item in
the plural condition.

Within every group there were children who treated the plural indeﬁnite unos
(‘some.PL’) as meaning ‘more than one’; however, the groups differed. signiﬁcantly in
how many children did this. The number of plural responses in the plural indeﬁnite unos
condition for each child was entered into a one-way AN OVA (adults, MexWC, ChMC,
ChWC). The results showed a signiﬁcant difference between the four groups

(F(3,74)=20.210,p<.001). Post hoc Bonferroni tests showed that only ChMC (p<.001)

197

and ChWC (p<.001), but not MexWC (p=.092), children differed signiﬁcantly from
adults in the number of plural responses assigned to the plural indeﬁnite. MexWC
children also differed signiﬁcantly from ChWC (p<.05) and ChMC (p<.05) children but
there were no signiﬁcant differences between the two Chilean child groups (p=l .0).
Table 42 and Table 43 show the percentage of children who were systematic in
their responses. Systematic responders are categorized as those who chose either the
singular or plural picture in at least 3 out of 4 trials. Variable responders are children who
chose the plural picture in half of the trials (2 out of 4) and the singular picture in half of

the trials.

Table 42. Study 2: Unos: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(3-4 plural) (3-4 singular) Total (2 pl/ 2sg)

 

Plural Pronounced as [s]

MexWC 74% 21% 95% 5%
ChMC 20% 70% 90% 10%
ChWC 35% 65% 100% 0%

 

198

Table 43. Study 2: Un: Systematic Responders.

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(3-4J11ural) (3-4 singular) Total (2 pl/ 28g)_

 

Plural Pronounced as [s]

MexWC 11% 89% 100% 0%
ChMC 0% 100% 100% 0%
ChWC 0% 100% 100% 0%

 

Similar to Study 1, almost all of the children were systematic in their responses. The
children who treated the plural indeﬁnite as associated to ‘one’ did so systematically. In
other words, it is not the case that the there were children who associated the plural
indeﬁnite to ‘one’ in half of the trials and also to ‘more than one’ in half of the trials.
Instead, children either associated the plural indeﬁnite to ‘one’ systematically or they
associated it to ‘more than one’ systematically.

Between 1 — 2 weeks after this experiment was carried out with the plural
morpheme pronounced as [s], Chilean children who chose the singular card in at least
three of the four plural trials were administered the same experiment but this time the
plural morpheme was pronounced as [h]. 11 ChWC children and 7 ChMC children
participated in this part of the experiment. The behavior of all children did not change.
The 11 ChWC children continued to choose the singular card in the plural condition 95%

of the time and the 8 ChMC did so 97% of the time. Paired samples t-test showed that

199

there was no signiﬁcant improvement neither for the 11 ChWC children (t(1,10)=-

l.00,p=.343) nor the 7 ChMC children (t(1,8)=-.552,p=.598).

5.2.6 Feminine vs. Masculine Nouns

It was suggested in Study 1 that children may perform better on the masculine
targets than on the feminine targets given that the only difference between the singular
and plural feminine indefmites (unas ‘some.PL’ vs. una ‘a/one’) is the presence of the
plural morpheme while the form of the plural masculine indeﬁnite (unos ‘some.PL’) and
singular masculine indeﬁnite (un ‘a/one’) is different. However, the results of Study 1
showed that the gender of the indeﬁnite noun phrase did not affect children’s
performance. It appears that gender did not affect child performance in Study 2 either.
The percentage of plural responses on the feminine and masculine indeﬁnite plurals is

shown in Table 44.

Table 44. Study 2: Plural Responses in Feminine and Masculine Trials.

 

ChWC ChMC MexWC

 

unos (‘some.M.PL’) 38% 30% 74%
(13/34) (6/20) (28/38)
unas (‘some.F.PL’) 32% 30% 84%

(1 1/34) (6/20) (32/38)

 

 

Paired samples t-tests showed that there were no signiﬁcant differences in the number of
plural responses between plural masculine and plural feminine indeﬁnite noun phrases

for ChWC children (t(1,16)=-1.852,p=.083), for ChMC children (t(l,9)=-1.0,p=.343), nor

200

for MexWC children (t( 1,18)=l .455,p=. 163). Similar to Study 1, the gender of the noun

does not affect child performance in Study 2.

5.2.7 Discussion

The experimental task in Study 2 seemed to be a bit more difﬁcult for all three
child groups than the experimental task in Study 1, as the number of plural responses
across the three child groups was slightly lower in the plural indeﬁnite condition in Study
2. It is possible that deciding between two pictures that contain the same objects, only
differing in the quantity of the objects, may cause difﬁculty for children if they are
focusing on the presence or absence of the object itself and not on the quantity. In other
words, given the question, “Which card has some marbles”, children may be happy to
choose any card that has marbles, regardless of quantity. However, if this is the case, it is
unclear why the results did not show variable responders in the plural indeﬁnite
condition. Instead, Chilean and Mexican children systematically chose either the singular
or plural card in the plural indeﬁnite condition. If children were'not paying attention to
quantity information, we might expect them to guess (because both cards would be
possible), which would result in more variable responders.

Although the task appears to have been slightly more difﬁcult for all three child
groups, the results still show that, given the same experimental conditions, the Mexican
children still associate the plural indeﬁnite to an interpretation of ‘more than one’ much
more often than the Chilean children, regardless of whether the plural is pronounced as
[s] or [b] for the Chilean children, which suggests that several Chilean children matched
neither [s] nor [h] to an underlying representation for number. Because we did not

administer a highlighting task in this study, we do not lmow for sure whether Chilean

201

children would improve if the plural morpheme were highlighted for them. Nevertheless,
the ﬁndings of Study 2 suggest very strongly that the variable input, which Chilean
children are exposed to, delays their comprehension‘of the plural morpheme, providing
support for the Variability Delay Hypothesis.

Finally, similar to Study 1, the results of Study 2 show that gender had no effect
on children’s ability to associate the plural indeﬁnite to an interpretation of ‘more than

one’. Children performed the same in the masculine and feminine trials.

5.2.8 Summary of Findings

The following is a summary of the results for the comprehension of plural
morphology in Study 2.

1. Chilean children differ from Mexican children in their comprehension of the
plural morpheme.

2. When the plural morpheme is systematically produced in the adult speech,
Mexican Spanish-speaking children associate the plural morpheme to an interpretation of
‘more than one’ by 5;4 years of age.

3. When the plural morpheme has a variable behavior and is also omitted in the
adult speech and no Highlighting Task is provided, ChWC and ChMC children do not
associate the plural indeﬁnite unos to an interpretation of ‘more than one’ at even at 5;5
years of age.

4. Almost all Chilean and Mexican children were systematic in their
interpretation of the plural and singular indeﬁnite noun phrases.

5. ChWC and ChMC children have difﬁculty associating both [s] and [h] in the

plural indeﬁnite unos to an interpretation of ‘more than one’.

202

6. The difference in form between the plural and singular masculine determiners
does not result in more correct responses in the masculine targets than on the feminine
targets.

7. The ﬁndings from Study 2 are consistent with the ﬁndings from Study 1.

5.3 Study 3. Picture Matching Task: Indeﬁnite Noun Phrases

The results of Study 2 showed that overall all three child groups provided slightly
fewer plural responses (i.e. choosing the picture with more than one object) than in Study
1. I suggested that this might possibly be due to the experimental materials in that
children had to decide between two pictures that contained the same objects. It may have
been the case that children were paying attention to the presence or absence of the object
and not to quantity in itself. The goal of Study 3 is to test child comprehension of the
plural indeﬁnite unos in a context where there are other objects in the picture to see
whether the percentage of plural responses (i.e. choosing the picture with multiple
objects) will increase for all child groups and also to test whether signiﬁcant differences

will still be found between Mexican and Chilean children.

5.3.1 Background
Study 3 investigates Chilean and Mexican children’s interpretation of plural and

singular indeﬁnite noun phrases as in (9).
(9) a. gCual niffo tiene unos burros?

Which boy has some.PL donkeys.PL

‘Which boy has some donkeys?’

203

b. gCual niﬁo tiene un burro?
Which boy has a/one.SG donkey.SG

‘Which boy has a/one donkey?’

The indeﬁnite in (9a) is plural and the indeﬁnite in (9b) is singular. Both the plural and
singular indefmites are in object position and the verb does not agree with the indeﬁnite
in number. For this reason, in the feminine determiner phrases the only number
information is the presence or absence of the plural morpheme on the noun and
determiner. The form of the masculine determiners is different in the plural vs. singular
determiner phrases.

Study 3 asks the following questions: (1) Given the differences in the input, will
Chilean children differ from Mexican children in their comprehension of indeﬁnite plural
noun phrases, as they did in Study 1 and Study 2? (2) Will there be an increase in plural
responses for all three child groups in Study 3 as compared to Study 2? (3) Will the
Chilean children treat both [s] and [h] as associated to an interpretation of ‘more than
one’. (4) Given that the masculine plural noun phrase has a different form than the
singular masculine noun phrase (e. g. un auto vs. unos autos), will children perform better

on masculine noun phrases than on feminine noun phrases?

5.3.2 Method and Design

A Picture Matching Task was used. Subjects were presented four short narratives
(Figure 11) about a group of children who were taking a trip together. See Appendix C
for the full set of narratives and pictures. After each short narrative, children were asked

questions like (9a) and (9b) above. For each short narrative, subjects were asked two

204

questions with either plural indeﬁnites unos/unas ‘some.M.PL/some.F.PL’ or singular
indefmites un/una ‘a/one.M.PL/a/one.I-‘.PL’ (as shown in Figln'e 11). The middle question,
gCudl niﬁo no tiene nada? (‘Which boy has nothing?’), was used to draw children’s

attention back to the center of the picture.

Figure 11. Study 3: Sample Target Trial.

 

 

 

 

 

 

205

Researcher: ”Primero Ios niﬁosﬁteron a mirar a Ios animales que habia en

el campo. Habia y . Asr’ es que los nir'ios jugaron con Ios

 

animales. A ver, veamos que tiene cada niﬁo. ” “First the boys went to see

the animals that were in the countryside. There were y . So the

 

boys played with the animals. Let’s see what each boy has.”

(10) a. gCudl nir’io tiene unos manos?
Which boy has some.PL monkey.PL

‘Which boy has some monkeys?’

b. gCudl niﬁo no tiene nada?
Which boy NEG has nothing

‘Which boy doesn’t have anything?’

c. gCudl niﬁo tiene unos burros?
Which boy has some.PL donkey.PL

‘Which boy has some donkeys?’

The boy with multiple objects is the correct answer in the plural condition and the boy

holding only one object is the correct answer in the singular condition. There were 4

singular target sentences, 4 plural target sentences, 4 singular controls, 4 plural controls

206

and 10 ﬁllers ﬁom a study on the development of the Spanish copulas ser and estar in
children.

In the target sentences the initial sound and the gender of each direct object was
controlled for Chilean subjects: burros ‘donkeys’, manos ‘monkeys’, barcos ‘boats’,
martillos ‘hammers’, bolitas ‘marbles’, manzanas ‘monkeys’, botellas ‘bottles’, monedas
‘coins’ (see above for explanation of how initial sounds were controlled for). Again, the
same words were used for Mexican children except changos was used for ‘monkeys’ and
canicas was used for ‘marbles’. In addition, half of the indefmites were feminine and half
were masculine. In the feminine indeﬁnites, only the plural morpheme provides number
information. In masculine indefmites the form of the determiner is also different in the
singular vs. plural conditions. Controls involved muchos (‘many’) and un solo (‘only
one’). In the control condition only gender, but not initial sound, was controlled for.
Subjects could perform well on the controls even if they ignored the plural morpheme

(LCua'l niﬁa tiene una sola llave? ‘Which girl has only one key?’). -

5.3.3 Subjects

52 children participated in this study. 12 MexWC (4;7-5;7, Mean: 5;1), 10 ChWC
(5;1-5;6, Mean: 5;3), 10 ChMC (5;0-5;4, Mean: 5;2) children and 20 adults (10 Chilean
and 10 Mexican) participated in the version of the study where the researcher pronounced
the plural morpheme as [s]. In addition, 11 ChWC (4;6 — 5;11, Mean: 5;1) and 9 ChMC
children (4;6 — 6; 1, Mean: 5;5) participated in the study where the researcher pronounced
the plural morpheme as [h]. Chilean children attended preschools and kindergartens in
Punta Arenas, Chile and the Mexican children attended a preschool in Mexico, D.F. None

of the children had yet received any reading education at school. Chilean adults were

207

undergraduates at the Universidad de Magallanes in Punta Arenas, Chile and the Mexican
adults were undergraduates at the Universidad Autonoma Metropolitana de Iztapalapa in

Mexico, D.F. Table 45 shows the distribution of children.

Table 45. Study 3: Distribution of Children.

 

 

 

 

Pronunciation Number Age Range Mean Age
MexWC [s] 12 4;7-5;7 5;1
ChMC [s] 10 5;0—5;4 5;2
[h] 9 4;6-6;1 5;5
ChWC [s] 10 5;1-5;6 5;3
[h] 11 4;6-5;ll 5;]
5.3.4 Procedure

All subjects were tested by native speakers of Spanish who lived in the same city
as the subjects. The author of this dissertation was present during the testing of all
children to ensure that procedures were identical for all subjects. The experimenter
started the testing session by introducing the main characters in the story to the child. The
experimenter then presented each section of the story to the child, stopping after each
short narrative to ask the target questions. All 8 target sentences were presented ﬁrst,
followed by 10 ﬁllers and then 8 controls. Controls were presented last to ensure that un
solo (‘only one’) would not provide number information for an (‘a/one.SG’) and that

children would not be biased to pick the singular picture in the unos (‘some.PL’)

208

condition because they assumed that muchos (‘many.PL’) and unos (‘some.PL’) could not
refer to the same quantity, that unos (‘some.PL’) must be less. Materials and presentation
of materials was identical for all children. Mexican adults were tested in the same way as

children while Chilean adults were given a paper and pencil version of this task.

5.3.5 Results

There were no differences between Chilean and Mexican adults so their scores
were combined. Although all children performed well on controls, they did not all
perform the same in the target conditions. The dependent variable is the number of plural
responses. Choosing the character with multiple items was considered a plural response.
Choosing the character with only one item was considered a singular response. Table 46
shows the percentage of plural responses for adults, MexWC children, ChMC children,

and ChWC children when the plural was pronounced as [s] and [h].

209

Table 46. Study 3: Plural Responses in Target Conditions.
unos (‘some.PL’) un (‘a/one.SG’)

Plural Morpheme Pronounced as [s]

Adults 100% 0%

MexWC 88% 8%
(42/48) (4/48)

ChMC 60% 4%
(24/40) (1/40)

ChWC 25% 23%
(10/40) (9/40)

Plural Morpheme Pronounced as [h]

ChMC 78% 8%
(28/36) (3/36)

ChWC 59% 21%
(26/44) (9/44)

 

Within every group there were children who associated the plural indeﬁnite unos
‘some.PL’ to an interpretation of ‘more than one’; however, the groups differed
signiﬁcantly in how many children did so. Looking ﬁrst at the children who heard the
plural morpheme pronounced as [s] in the experiment, the number of plural responses for
each child was entered into a 4 (Group: adults, [s]-MexWC, [s]-ChMC, [s]-ChWC) X 2

(Condition: un, unos) mixed design Analysis of Variance (ANOVA) with Condition as a

210

within subjects variable and Group as a between subjects variable. The results revealed a
main effect for Group (F (1,48) = 7.942, p < .001) and a main effect for Condition
(F(l,48) = 211.658, p < .001). There was also a signiﬁcant interaction between Group X
Condition (F(l,48) = 26.761, p < .001). Post hoc Bonferroni t-tests showed that only [s]-
ChWC (p<.001) and [s]-ChMC (p<.05) children differed signiﬁcantly from adults while
[s]-MexWC children (p=1 .0) did not differ signiﬁcantly from adults. Also, [s]-MexWC
children only differed signiﬁcantly from [s]-ChWC children (p<.01) but not [s]-ChMC
children (p=.102). The two [s]-Chilean child groups did not differ signiﬁcantly from each
other (p=1 .0).

Independent samples t-tests showed a signiﬁcant increase in plural responses in
the indeﬁnite plural condition for the ChWC children (t(1,19)=2. 191,p<.05) when the
plural is pronounced as [h] but no signiﬁcant increase in plural responses for ChMC
children (t(1,17)=.993,p=.335). However, because of the increase in plural responses
when the plural was pronounced as [h], it is possible that [h]-Chilean children will not
behave signiﬁcantly different from [s]-MexWC children. For this reason, [h]-ChWC, [h]-
ChMC, [s]-MexWC and [s]-Adu1ts were compared. A one-way ANOVA comparing
these four groups in the plural indeﬁnite condition showed signiﬁcant differences across
the four groups (F (1,5 1)=6.887,p<.001). Post hoc Bonferroni tests showed that only [h]-
ChWC children (p<.001), but not [h]-ChMC children (p=. 197), differed signiﬁcantly
from adults.

Table 47 and Table 48 show the percentage of children who were systematic in
their responses. Systematic responders are categorized as those who chose either the

singular or plural picture in at least 3 out of 4 trials. Variable responders are children who

211

chose the plural picture in half of the trials (2 out of 4) and the singular picture in half of

the trials.

Table 47. Study 3: Unos: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(34 plural) (3-4 signlar) Total (2 pl/ 2sg)_

Plural Pronounced as [s]

MexWC 100% 0% 100% 0%
ChMC 60% 40% 100% 0%
ChWC 20% 70% 90% 10%

Plural Pronounced as [h]
ChMC 7 8% 1 1% 89% 11%
ChWC 64% 36% 100% 0%

 

212

Table 48. Study 3: Un: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(3-4 plural) (3-4 singular) Total (2 pl/ 2sgL

 

Plural Pronounced as [s]

MexWC 8% 92% 100% 0%

ChMC 0% 100% 100% 0%

ChWC 20% 80% 100% 0%
Plural Pronounced as [h]

ChMC 1 1% 89% 100% 0%

ChWC 9% 91% 100% 0%

 

In general, children were systematic in their responses in the plural and singular
indeﬁnite conditions. Only 1 [s]-ChWC child and 1 [h]-ChMC child showed variable

behavior in the plural indeﬁnite condition.

5.3.6 Feminine vs. Masculine Nouns
As in Study 1 and 2, children did not behave signiﬁcantly different in the
feminine vs. masculine trials. Table 49 shows the percentage of plural responses in the

masculine and feminine plural indeﬁnite trials.

213

Table 49. Study 3: Plural Responses in Feminine and Masculine Trials.
[S] [111
CONDITIONS ChWC ChMC MexWC ChWC ChMC

 

 

unos 20% 60% 92% 59% 78%

(‘some.M.PL’) (4/20) (12/20) (22/24) (13/22) (14/18)

unas 30% 60% 83% 59% 78%

(‘some.F.PL’) (6/20) (12/20) (20/24) (13/22) (14/18)

 

 

Paired samples t-tests showed no signiﬁcant difference in the number of plural responses
between the plural masculine vs. plural feminine indeﬁnites for [s]-ChMC children

(t(l ,9)=0.0,p=1.0), [h]-ChMC children (t(l,9)=0.0,p=l .0), [s]-ChWC children (t(1,10)=-
1.0,p=.343), [h]-ChWC children (t(1,10)=0.0,p=l .0), nor for MexWC children

(t(1,11)=.804,p=4.38).

5.3.7 Discussion

Compared to Study 2, which appeared to be slightly harder for children, Study 3
showed an increase in the percentage of plural responses in the plural indeﬁnite condition
for all three child groups. The results of Study 3 show that when the plural morpheme is
pronounced as [s] both ChWC and ChMC children differed signiﬁcantly from adults in
their comprehension of the plural morpheme while Mexican children did not. This
ﬁnding is consistent with the ﬁndings in Study 1 and Study 2. However, Study 3 also
showed that when the plural morpheme was pronounced as [h], only the ChWC children,

but not the ChMC children differed signiﬁcantly from adults. This increase in plural

214

responses in Chilean children who were presented with the plural variant [h] vs. those
presented with the plural variant [s] indicates that more Chilean children associated [h] to
an interpretation of ‘more than one’. However, the results in the [h] version of the study
also reveal that ChWC children, unlike Mexican children, still have not reached adult
levels in their interpretation of the plural morpheme.

It is not clear at this point why Chilean children associated the plural variant [h] to
‘more than one’ in Study 3 but not in Studies 1 or 2. Given that the subjects are different
in each of these three experimental studies, it could be simply due to the children who
participated. However, the production data presented in Chapter 4 revealed that in free
speech, [h] is the most ﬁ'equent plural variant for ChWC adults and children and ChMC
adults and children, with the ChWC adults and children producing more Q than [h]. For
this reason, it is not surprising that more Chilean children might associate [h] to and
interpretation of ‘more than one’ before associating [s] to an interpretation of ‘more than
one’. Instead, it is surprising that in Studies 1 and 2 more children did not associate [h] to
‘more than one’. Finally, as in Studies 1 and 2, gender of the noun phrase did not affect

children’s performance in Study 3.

5.3.8 Summary of Findings

The following is a summary of the results for the comprehension of plural
morphology in Study 3.

l. Chilean children differ ﬁom adults in their comprehension of plural

morphology while Mexican children did not differ from adults.

215

2. When the plural morpheme is systematically produced in the adult speech,
Mexican Spanish-speaking children associate it to an interpretation of ‘more than one’ by
5;] years of age.

3. When no Highlighting Task is provided and the plural morpheme is
pronounced as [s], ChWC and ChMC 5;5-year-old children do not associate the plural
indeﬁnite unos to the interpretation of ‘more than one’.

4. More Chilean children associated [h] to the interpretation of ‘more than one’
than [s].

5. ChMC children, but not ChWC children, reached adult levels when the plural
variant was pronounced as [h].

6. Almost all Chilean and Mexican children were systematic in their
interpretation of the plural and singular indeﬁnite noun phrases.

7. The difference in form between the plural and singular masculine noun phrases
did not result in more correct responses in the masculine targets than on the feminine
targets.

8. The ﬁndings from Study 3 are consistent with the ﬁndings from Studies 1 and
2, with the exception that ChMC children were adult-like when the plural morpheme was

pronounced as [h].

5.4 Study 4. Elicitation Task: Bare Nouns and Indeﬁnite Noun Phrases

Study 4 is not a comprehension task but rather an elicitation task. Nevertheless, it
was placed within this chapter on comprehension because the experimental task was
designed to provide information about whether children associate the plural morpheme to

an interpretation of ‘more than one’. Unlike the production tasks in Chapter 4, the

216

elicitation task presented here involves naming plural and singular objects in order to
determine whether children will use the plural morpheme to distinguish ‘more than one’
from ‘one’. If the only difference between children’s descriptions of plural vs. singular
objects is the plural morpheme, this would provide evidence that children associate the
plural morpheme to the interpretation of ‘more than one’. Hence, the purpose of Study 4
is to determine whether Chilean and Mexican children produce singular/plural minimal
pairs. Finally, given that Chilean children and adults produced the plural variant [s] more
' often in the Naming Task in Chapter 4 than in any other production task, the Naming
Task presented here will allow us to determine whether Chilean children use [5] to

distinguish ‘one’ from ‘more than one’.

5.4.1 Background

Study 4 asks the following questions: (1) Given that Chilean children often assign
an interpretation of ‘one’ to the plural indeﬁnite unos/unas will they use the indeﬁnite
plural to describe plural sets? (2) Will Chilean and Mexican children use the plural [s]
variant to produce plural-singular minimal pairs when describing ‘more than one’ vs.

‘one’?

5.4.2 Method and Design
An Elicitation Task was used where children were asked to name pictures of
animals. Children were shown pictures of single animals and of plural sets of animals, as

shown in Figure 12, followed by question (11). All pictures are shown in Appendix D.

217

Figure 12. Study 4: Sample Target Trial.

461;?

 

 

 

 

 

 

 

(11) gQué hay aqui?
What EXST here

‘What’s here?’

The question in (11) has an existential verb, which is not marked for number and can
occur with either plural nouns or singular nouns. For this reason, any plural nouns that
are elicited from children will not be the result of children makingrthe noun phrase agree
in number with the verb, as may have been the case in the Naming Task discussed in
Chapter 4. There were 16 pictures total: 8 singular pictures and 8 plural pictures. Table
50 shows the singular and plural forms of the nouns that were elicited from children. For
every animal, both a singular picture and a plural picture were shown to the child so that
we could elicit minimal singular-plural pairs. The pictures and order of presentation were

the same for all children and are shown in Appendix D.

218

Table 50. Study 4: Target Words Elicited.

 

/-s/ allomorph
Singular Plural
vaca ‘cow’ vacas ‘cows’
abeja ‘bee’ abejas ‘bees’
chancho/cochino ‘pig’ chanchos/conchinos ‘pigs’
perro ‘dog’ perros ‘dogs’
gato ‘cat’ gatos ‘cats’
elefante ‘elephant’ elefantes ‘elephants’
/-es/ allomorph
pez ‘ﬁsh’ peces ‘ﬁshes’
ratén ‘mouse’ ratones ‘rnice’

 

Of the four nouns used in Study 4, 2 were masculine and 2 were feminine, all 4 requiring
the /-s/ allomorph. There were also 2 masculine nouns requiring the /-es/ allomorph.
Finally, elefante was included because it requires an /-s/ allomorph but its plural form
looks as if it has an /-es/ allomorph. If children have difﬁculty with the /-es/ allomorph, as
previous studies suggest, they should do ﬁne on the plural form of elefante, for example,

but not on the plural form of pez and rato'n.
5.4.3 Subjects

52 subjects participated in this study. 11 MexWC children (4:0-4;7, Mean: 4:2),
12 ChWC children (4:2-4;1 1, Mean: 4:7) and 7 ChMC (4;2-4;11, Mean: 4:4). In addition,

8 Mexican adults and 14 Chilean adults participated in this study. Mexican children and

219

adults were recruited ﬁ'om a preschool in Mexico City and the Universidad Autonoma
Metropolitana de Iztapalapa in Mexico City. The Chilean children and adults were

recruited ﬁom preschools and the Universidad de Magallanes in Punta Arenas, Chile.

Table 51. Study 4: Distribution of Children.

 

Number Age Range Mean Age

 

MexWC l 1 4:0-4:7 4:2
ChMC 7 4;2-4:1 1 4:4
ChWC 12 4;2-4:1 1 4:7

 

 

5.4.4 Procedure

Subjects were tested in a quiet classroom. They were ﬁrst presented with three
warm-up trials to get them used to talking and interacting with the researcher. These
practice tasks involved pointing to the larger or smaller of two circles, naming colors and
naming shapes. After the warm-up task, children were shown the ﬁrst picture followed by
the question in (11). This task was very quick and simple for all children. Based on
recordings, noun phrases were transcribed and coded for whether the plural morpheme

was pronounced as [s] or not.

5.4.5 Results
All but one child produced [s] as the plural morpheme in at least one trial in plural
picture condition, and most children produced [s] as the plural morpheme at least 50% of

the time in the plural picture condition. On the other hand, all children produced the

220

singular indeﬁnite noun phrases un (‘a/one.M.SG’) and una (‘a/one.F.SG’) when describing
singular pictures. Figure 13 shows the percentage of time children and adults pronounced

the plural morpheme as [s] when describing plural pictures.

Figure 13. Study 4: Percentage of Nouns with Plural Variant [s].
100% 99%

 

100% r

   

l
80% fr

1
60% s

l

40% r;—

20% 1

0% a ' '
MexAdults ChAdults MexWC ChMC ChWC

     

 

 

 

 

Figure 13 shows that, although there is variability in frequency with which children and
adults produce the plural morpheme as [s] in the plural picture condition, all child and
adult groups produce the plural morpheme as [5] some of the time. The number of plural
nouns phrases produced by children and adults in the plural picture condition was placed
into one—way AN OVA. The two adult groups were not combined given the differences in
results between them. The results showed signiﬁcant differences between the four groups
(ChWC, ChMC, MexWC, ChAdults, MexAdults) (F(1,51)=7.729,p<.001). Post hoc

Bonferroni t-tests showed only ChWC children differed signiﬁcantly from Chilean adults

221

(p<.001). MexWC children did not differ from Mexican adults (p=1 .0) and ChMC
children did not differ from Chilean adults (p=1 .0). Post hoc Bonferroni t-tests also
showed that between the child groups, there were only signiﬁcant differences between
the ChWC and MexWC children (p<.001), but not beﬂeen the ChWC and ChMC
(p=.344) nor the MexWC and ChMC children (p=.975). These ﬁndings suggest that both
ChMC and MexWC children, but not ChWC children, use [8] to describe plural sets,
hence it must be the case that they associate [s] to ‘more than one’. However, these
ﬁndings only show us that subjects produced [s] as the plural morpheme, they do not tell
us whether they produced the indeﬁnite plural unos/unas to describe plural sets nor
whether they produced plural-singular minimal pairs.

With respect to the types of noun phrases that children and adults produced, the
results showed that, given the experimental question, gQué hay aqui? (‘What’s here’),
there were only a couple of different types of noun phrases produced. When shown the
singular picture all children and adults produced indeﬁnite noun phrases (e. g. una vaca
‘a/one.SG cow.SG’). However, when shown a plural picture, there were at least three
possible answers that children and adults provided: indeﬁnite plural noun phrases (e. g.
unas vacas ‘some.PL cow.PL’), bare plurals (e.g. vacas ‘cow.PL’) and numerals (e.g. tres
vacas ‘three cows.PL’). The example in (12) shows the noun phrase type that all children
and adults produced for the singular picture and the examples in (13) shows the noun

phrase types that were produced for the plural pictures.

222

(12) una vaca
a/one.SG cow.SG

‘a/one cow’

(13) a. unas vacas
some.PL cows.PL

‘some cows’

b. vacas
cows.PL
‘cows’

c. tres vacas

three cows.PL

‘three cows’

The noun phrases in (12) and (13a) are minimal pairs with the only difference between
the two being the presence or absence of the plural morpheme. Since all children used
(12) to describe the singular picture, any children who use (13a) to describe the plural
picture would appear to have plural morphology because they are using the plural
morpheme alone to distinguish a singular set from a plural set. This is a very important
point and was discussed in Chapter 3 where it was unclear in previous studies whether

children were producing minimal pairs or not in a similar task (see Bedore and Leonard

223

2001). The ﬁndings from Study 1, Study 2, and Study 3 indicate that Chilean children,
but not Mexican children, associate the plural indeﬁnite (e. g. unas vacas ‘some.PL
cows.PL’) to and interpretation of ‘one’ in the comprehension tasks. We attribute this to
the variability in the input: if the plural morpheme is sometimes omitted in the input, the
plural indeﬁnite would sound like the word for ‘one’ in Spanish. If Chilean children
associate the plural indeﬁnite to ‘one’, they should not produce indeﬁnite plurals when
describing plural sets. Mexican children, on the other hand, could produce indeﬁnite
plurals (or any other type of plural NP) when describing plural sets since the previous
comprehension studies in this dissertation indicate that Mexican children associated the
plural indeﬁnite to ‘more than one’. Table 52 shows the percentage of noun types

produced by all child and adult groups.

224

Table 52. Study 4: Percentage of Noun Types Produced.

 

 

 

Indeﬁnite Bare NPs Bare NPs
NPs [s] [h/0] Numerals

unas vacas vacas vaca/vacah” 3 vacas

‘some cows’ ‘cows’ ‘cow/cows’ ‘3 cows’ Other
Mexican 10% 22% 0% 68% 0%
Adults (6/63) (14/63) (43/63)
Chilean 4% 29% 2% 63% 2%
Adults (5/1 12) (32/1 12) (3/1 12) (70/112) (2/1 12)
MexWC 80% 1 1% 0% 5% 4%
Children (68/85) (9/85) (4/85) (4/85)
ChMC 4% 53% 19% 24% 0%
Children (2/53) (28/53) (10/53) (13/53)
ChWC 7% 51% 42% 0% 0%
Children (6/87) (44/87) (37/87)

 

The results of noun types produced in Table 52 show that, while Mexican children

produced plural indeﬁnite NPs (e. g. unas vacas ‘some.PL cows.PL’) 80% of the time,

ChWC children only produced plural indeﬁnites 7% of the time and ChMC only 4% of

the time (i.e. only 1 ChWC child and 1 ChMC child produced plural indeﬁnites), which

indicates, along with the ﬁndings in Studies 1, 2, and 3 that Chilean children do not

associate the plural indeﬁnite unos/unas to an interpretation of ‘more than one’ and

furthermore, suggests that Chilean children have different strategies than Mexican

children for encoding plrnality. Mexican children appear to have a preference for the

plural indeﬁnite in the Naming Task presented here, while Chilean children prefer the

 

2° The bare plural NPs were only coded as having the plural pronounced as [s] or not (i.e.
[h] or zero). This decision was primarily made because we were interested to know
whether ChMC and ChWC children would pronounce the plural morpheme as [s].

bare plural. This is expected given that many Chilean children systematically associated
the plural indeﬁnite to an interpretation of ‘one’ in the comprehension tasks presented in
Studies 1, 2, and 3. It should also be noted that the production data in Chapter 4 showed
that in ﬂee speech, Chilean adults produced just as many plural indeﬁnites as Mexican
adults. For this reason, the preference for plural indefmites by Mexican children in Study
4 cannot be due to Mexican adults producing more indeﬁnite plurals in the input
Additionally, with respect to the two nouns pez (‘ﬁsh.SG’) and rato'n (‘rat.SG’) that
required [es] in their plural forms showed that, while Chilean children produced [ratone]
and [ratones] in their plural descriptions, they produced [peses] and [pes] in their plural
descriptions of ‘ﬁsh’. The latter behavior for ﬁsh was also found for some Mexican
children.

Another implication of the results presented in Table 52 is that, while the Mexican
children appear to have matched the plural morpheme to any underlying representation
for number because they are producing plural and singular minimal pairs (e. g. unas vacas
‘some.PL cows.PL’ vs. una vaca ‘a/one.SG cow.SG’) to describe plural and singular
pictures, the results do not demonstrate this for Chilean children because they do not
produce plural-singular minimal pairs but rather bare plurals vs. singular indeﬁnites (e. g.
vacas ‘cows.PL’ vs. una vaca ‘a/one.SG cow.SG’). It may appear the Chilean children are
more like adults than Mexican children because both adult groups produced very few
plural indeﬁnites. However, we must be cautious in this interpretation of the results given

that the adults had a tendency to produce noun phrases headed by numerals.21

 

2’ Both of the adult groups and the ChMC children produced more noun phrases headed
by numerals than the ChWC and MexWC children, which we assume is because ChMC
children and adults have had more counting experiences than the other two child groups.

226

5.4.6 Discussion

The results of Study 4 reveal that both Mexican and Chilean adults and children
use the plural morpheme [s] to describe plural sets. This suggests very strongly that the
plural variant [s] occurs in the input to both Chilean and Mexican children and hence
should be tested, in addition to [h], in the comprehension tasks. However, the results also
showed that ChWC children used [s] signiﬁcantly less so than Chilean adults and also
Mexican children, which is consistent with the production data presented in Chapter 4
and indicates that ChWC children have less [s] in the input that the MexWC and ChMC
children.

Study 4 also indicates that Chilean children do not use the plural indeﬁnite
unos/unas when describing plural sets while Mexican children do use plural indeﬁnites.
Instead, Chilean children used bare nouns (both with and without the plural morpheme
[5]) to describe plural sets. This ﬁnding is consistent with Studies 1, 2, and 3, which
indicate that Chilean children do not associate the plural indeﬁnite to an interpretation of
‘more than one’.

Finally, the ﬁndings from Study 4 show that, while Mexican children use plural-
singular minimal pairs to describe ‘more than one’ and ‘one’ (e. g. unas vacas vs. una
vaca), Chilean children do not. Instead, Chilean children use bare plurals to describe
‘more than one’ and indeﬁnite singulars to describe ‘one’ (e. g. vacas vs. una vaca). For
Chilean children, responses for the plural and singular picture differ with respect to the
presence or absence of the plural morpheme and also of the determiner. Because children
could produce any type of plural noun phrase in the plural picture condition, it is possible

that the high number of bare plurals for Chilean children is simply a preference and

227

Chilean children actually do produce indeﬁnite plurals to describe plural sets. While this
might be the case, it is then unclear why Chilean children overwhelmingly prefer bare
plurals and Mexican children overwhelmingly prefer plural indeﬁnites. Given the
ﬁndings ﬁom Studies 1, 2, and 3 the difference between Chilean and Mexican children in
Study 4 seems to have to do with Chilean children’s tendency to assign an interpretation
of ‘one’ to plural indefmites. These differences between Chilean and Mexican children
suggest that variability in the production of the plural morpheme has an effect on the

strategies children use to encode plurality in their production.

5.4.7 Summary of Findings

The following is a summary of the results for the production of plural morphology
in Study 4.

1. Both Chilean and Mexican adults and children use the plural variant [s] to
varying degrees when describing plural sets.

2. Mexican children produce plural-singular minimal pairs when describing
‘more than one’ vs. ‘one’.

3. Chilean children produce bare plurals vs. singular indefmites when describing
‘more than one’ vs. ‘one’

4. Chilean children prefer bare plurals when describing plural sets and avoid
plural indeﬁnites.

5. The ﬁndings of Study 4 are consistent with those ﬁ'om Studies 1, 2, and 3,
indicating that Chilean children do not appear to associate the plural indeﬁnite unos/unas
to ‘more than one’ and also indicating that ChWC children are exposed to an input with

less [s] than the other two child groups.

228

5.5 Study 5. Picture Matching Task: Bare Singulars vs. Bare Plurals

Study 4 showed that Chilean children overwhelmingly produce bare plurals in
their description of plural sets; however, because they also produce singular indeﬁnites to
describe singular sets their plural vs. singular descriptions do not form minimal pairs with
the only difference being the plural morpheme. For this reason, we cannot conclude that
Chilean children associate the plural morpheme with an interpretation of ‘more than one’.
Rather, it may be that Chilean children associate bare nouns with a ‘more than one’
interpretation. The purpose of Study 5 is to determine whether Chilean children associate
the plural morpheme with an interpretation of ‘more than one’ by testing their

comprehension of bare plurals vs. bare singulars.

5.5.1 Background
Study 5 tests Chilean children on their interpretation of bare plurals and bare

singulars as in (14a) and (14b) below.

(14) a. LCual niﬁa tiene llave?
Which girl has key.SG

‘Which girl has a key/keys?’
b. gCual niﬁa tiene llaves?

Which girl has keys.PL

‘Which girl has keys?’

229

The bare singular in (14a) can refer to an interpretation of ‘one’ or ‘more than one’ while
the bare plural in (14b) only refers to an interpretation of ‘more than one’. Only Chilean
children participated in Study 5, given that it was Chilean children who produced so
many bare plurals in Study 4 and also given the fact that the distribution of bare singulars
is much more restricted in Mexican Spanish than in Chilean Spanish. Bosque (1996)
observes that Spanish bare singulars, in the dialects where they are found, are allowed as
objects in certain contexts: as objects of intensional predicates like buscar (to look for),
querer (to want), and necesitar (to need), as objects under negation, and in constructions
that denote inherent properties of a particular entity or where the object has an unique
interpretation (as in Llevaba chaqueta ‘He was wearing a jacket’, T enia casa en la
montaﬁa ‘He had a house in the mountains’). In our ﬁeld-work, we have found that
Chilean Spanish (more speciﬁcally the dialect of Punta Arenas) also allows non-
intensional verbs to take bare singular objects (Me compré auto ‘I bought myself a car’,
Me consegui perro ‘I got a dog’, Hicimos muralla ‘We put up a ﬁre wall’ (between our
house and the neighbor’s house), Viste concha ‘You saw seashells’, T engo que pagar
cuenta ‘I have to pay bills’).

Study 5 asks the following questions: (1) Will Chilean children distinguish
between bare singular and bare plural noun phrases? (2) Will the Chilean children

associate [s] or all bare nouns to an interpretation of ‘more than one’?

5.5.2 Method and Design
A Picture Matching Task was used. Subjects were presented four short narratives
(Figure 14) about a group of children who were taking a trip together. See Appendix E

for the full set of narratives and pictures. After each short narrative, children were asked

230

questions like (153) and (15c). For each short narrative, subjects were asked two
questions with either bare plurals Ilaves/libros ‘keys.F.PI/books.M.PL’ or bare singular
[lave/libro ‘key.F.PL/book.M.PL’. Questions like (15b), gCudl niﬁa no tiene nada?
(‘Which girl has nothing?’), were used to draw children’s attention back to the center of
the picture. Although the pictures used in Study 3 and Study 5 are the same, the
experimental questions and also children tested are not the same. All target trials involved
nouns that can optionally occur bare in object position of tener (‘to have’) in Chilean
Spanish: llave (‘key’), libro (‘book’), pelota (‘ball’), toalla (‘towel’), moneda (‘coin’),
carta (‘letter’), gato (‘cat’), perro (‘dog’). There were 4 trials in the bare singular
condition (like (8a) above), 4 trials in the bare plural condition (like (8b) above), 4
singular controls (un solo ‘only one’) and 4 plural controls (muchos ‘many’). In addition,
there were 10 ﬁllers that came from another study examining the development of the
copulas ser vs. estar in child Spanish. Although both feminine and masculine nouns were
used, this did not matter for this particular study given that the nouns were bare and
hence only the presence or absence of the plural morpheme provided number
information. In the control condition, subjects could perform well even if they ignore the

plural morpheme (gCudl niﬁo tiene un solo mono? ‘Which boy has only one monkey?’).

231

Fi e 14. Stud 5: Sam le Tar et Trial.

 

 

 

 

 

 

Researcher: “Primero, ﬁleron a conocer un castillo enorme porque en ese
castiIIo habia cosas muy interesantes. A ver, habia _ y _. AI salir del
castillo las niftas se Ilevaron algunas cosas. A ver, veamos que tiene cada
niﬁa. ” "First, they went to visit a big castle because in this castle there were
many interesting things. Let’s see, there were _ and _. Upon leaving

the castle the girls took some things. Let’s see what each girl has.”

232

(15) a. g Cudl niiia tiene libro?
Which girl has book

‘Which girl has (a) book/books?’

b. gCudl niﬁa no tiene nada?
Which girl NEG has nothing

‘Which girl doesn’t have anything?’

c. gCudl niﬁa tiene llaves?
Which girl has keys.PL?

‘Which girl has keys?’

Given this experimental story and picture, when subjects are presented with a sentence
like (15a) they should choose the girl with only one book or with several books, when
subjects are presented with a sentence like (15c) they should choose only the girl with

several keys.

5.5.3 Subjects

20 Chilean children participated in this study. 10 children were ChWC (4:4-5:8,
Mean: 5:1) and 10 children were ChMC (4;6-5:11, Mean: 5;2). In addition, 10 Chilean
adults participated. Children were recruited from preschools in Punta Arenas, Chile and
adults were undergraduate college students at the Universidad de Magallanes in Punta

Arenas, Chile.

233

5.5.4 Procedure

All subjects were tested by native speakers of Spanish who lived in the same city
as the subjects. The author of this dissertation was present during the testing of all
children to ensure that procedures were identical for all subjects. The plural morpheme
was pronounced as [s] for all children. We chose to pronounce the plural morpheme as [s]
for several reasons: (1) there was no determiner on the nouns; hence, the morpheme
would be the only clue to number and [s] is acoustically more salient than [h], (2) Chilean
children from Study 1 and Study 2 did not associate [h] to ‘more than one’ (3) ChMC
children who did recognize [h] as the plural morpheme, often corrected the researcher
when he pronounced the plural as [h] and (4) adults and Chilean children produced bare
nouns with the plural morpheme [s] more than half the time in Study 4. All 8 target trials
were presented ﬁrst, followed by the ﬁllers and then the 8 controls, as shown in

Appendix E.

5.5.5 Results

While ChWC and ChMC children behaved the same in the control conditions,
assigning a plural reading to muchos (‘many.PL’) 100% of the time and a singular reading
to un solo (‘only one’) 100% of the time, they did not behave the same in the target
conditions. The percentage of ‘more than one’ interpretations that children assigned to

the bare singular and bare plural is shown in Table 53.

234

Table 53. Study 5: Plural Responses in Target Conditions.

 

Bare Singulars Bare Plurals
vaca (‘cow’) vacas (‘cows.PL’)

Chilean Adults 2% 100%

(2/40) (40/40)
ChWC Children 68% 83%

(27/40) (33/40)
ChMC Children 48% 95%

(19/40) (38/40)

 

 

The results show that, while adults preferred to associate the bare singular to an
interpretation of ‘one’, the Chilean children did not. Rather, Chilean children allowed
both a plural and singular reading for the bare singular. The adult behavior is surprising
given that bare singulars can be associated to an interpretation of ‘one’ and also to an
interpretation of ‘more than one’, making the experimental question ambiguous. It may
be the case that adults were comparing the bare singular to the bare plural, which only
allows a plural reading and assumed that a singular response was being elicited from
them. In any case, both an interpretation of ‘one’ and ‘more than one’ are grammatical.

A mixed design ANOVA with condition (bare plural, bare singular) as a within
subjects variable and group (Adults, ChMC, ChWC) as a between subjects variable
showed a main effect for group (F (l,27)=6.022,p<.01) and for condition

(F(l,27)=76.164,p<.001) and a signiﬁcant interaction for group x condition

235

(F (1,27)=14.910,p<.001). Post hoc Bonferroni t-tests showed that both ChWC (p<.05)
and ChMC (p<.05) children differed signiﬁcantly from adults but not ﬁ'om each other
(p=1 .0).

For the child groups the results showed that, while ChMC children assigned a
‘more than one’ interpretation to the bare plural 95% of the time, they only assigned a
‘more than one’ interpretation to the bare singular 48% of the time. Planned comparison
paired samples t-tests showed that ChMC children treated bare singulars signiﬁcantly
different from bare plurals (t(l,9)=4.146,p<.01). On the other hand, the results showed
that ChWC children assigned a ‘more than one’ interpretation to bare plurals 83% of the
time and a ‘more than one’ interpretation to bare singulars 68% of the time. Unlike for
ChMC children, planned comparison paired samples t-tests showed that ChWC children
did not treat bare singulars differently from bare plurals (t(l,9)=1.108,p=.297) These
results suggest that ChMC children distinguish between plural and singular bare nouns
while ChWC children do not, instead ChWC children associate bare singulars and bare
plurals to an interpretation of ‘more than one’.

The results of Studies 1, 2 and 3 showed that overall children were systematic in
their comprehension of the plural morpheme. The percentage of subjects who were
systematic in their interpretation of bare plurals and bare singulars in Study 5 is shown in

Table 54 and Table 55.

236

Table 54. Study 5: Bare Plural: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(3-4 plural) (3—4 singular) Total (2 pl/ 2§gL
Plural Pronounced as [s]
ChMC 100% 0% 100% 0%

ChWC 80% 10% 90% 10%

 

Table 55. Study 5: Bare Singplar: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(3-4 plural) (3-4 singular) Total (2 pl/ ZSgL
ChMC 30% 30% 60% 40%

ChWC 60% 30% 90% ' 10%

 

Table 54 shows that in the Bare Plural condition, ChWC and ChMC children

systematically treated the bare plural as associated to an interpretation of ‘more than one’,

while only 1 ChWC child systematically associated the bare plural to an interpretation of

‘one’ and l ChWC child treated the bare plural variably. On the other hand, Table 55

shows that in the Bare Singular condition almost half of the ChMC children were variable

in their responses, while only 1 ChWC child was variable in his response patterns. The

patterns of variability for the ChMC children is telling especially considering that ChMC

237

children were not variable, but rather systematic, in Studies 1, 2, and 3. This suggests that
ChMC children associate the bare singular to both an interpretation of ‘one’ and ‘more
than one’. This cannot be concluded for the ChWC children given that they were not
variable in their interpretation of the bare singular. However, we must be careful in this
interpretation as both an interpretation of ‘more than one’ and ‘one’ are compatible with
the bare singular.

Because Study 3 and Study 5 were very similar with the exception that Study 3
tested singular indefmites vs. plural indeﬁnites (un mono vs unos manos ‘a/one.SG
monkey.SG vs. some.PL monkeys.PL’) while Study 5 tested bare singulars vs. bare plurals
(libro vs. libros ‘bookSG vs. book.PL’), the results of the two studies will be compared.
Note that the subjects who participated in Study 3 were different from the subjects in
Study 5 and we are comparing only the results of Study 3 where the plural morpheme
was pronounced as [s], given that [h] was not tested in Study 5. Table 56 shows the ages

of the Children in Study 3 and Study 5.

238

Table 56. Comparison of Ages in Study 3 and Study 5.

Pronunciation Number Age Range Mean Age

 

Study 3
ChMC [s] 10 5:0-5;4 5;2
ChWC [s] 10 5:1-5;6 5;3
Study 5
ChMC [s] 10 4:4-5:8 5:1
ChWC [s] 10 4:6-5:1 1 5:2

 

The percentage of plural responses for the plural indefmites and singular
indeﬁnites in Study 3 and the bare plurals and bare singulars in Study 5 for both the

ChWC and ChMC children is shown in Figure 15 .

239

Figure 15. Comparison of Plural Responses in Study 3 and Study 5.

 

 

 

 

 

 

 

 

 

100%
80%
60%- -
40% _ I ChMC
20% —

0% -

Bare Bare Plural Singular Plural
Singular Indeﬁnite Indeﬁnite
Study 5 Study 3

 

 

 

 

This comparison shows that when the plural morpheme is pronounced as [s], both ChWC
and ChMC children assign a plural reading to bare plurals (libros ‘books.PL’) and a
singular reading to the singular indeﬁnites (un mono ‘a/one.SG monkey.SG’), similar to
what they produced when describing plural vs. singular sets in Study 4. For this reason,
the results of Study 3 and Study 5, taken together, indicate that both ChWC and ChMC
children associate the singular indeﬁnite to an interpretation of ‘one’ and the bare plural
to an interpretation of ‘more than one’. This ﬁnding, however, does not show that

Chilean children associate the plural morpheme to an interpretation of ‘more than one’.

240

The results of both studies also show that Chilean children associate the bare plural to an
interpretation of ‘more than one’ more often than they associate the plural indeﬁnite to an
interpretation of ‘more than one’. An independent t-test showed that Chilean children
assigned a plural interpretation signiﬁcantly more often to the bare plural than to the
plural indeﬁnite, even though the ‘more than one’ interpretation was the only correct
answer in the contexts provided (ChWC: (t(1,18)=-4.445,p<.001), ChMC: (t(1,18)=-

2.510,p<.05)).

5.5.6 Discussion

The results of Study 5 differ quite drastically from the results of Study 3 in that
Chilean children overwhelmingly provided plural responses in Study 5 yet they
overwhelmingly provided singular responses in Study 3. Given that the only difference
between the two experimental studies was that Study 3 tested plural indefmites and
singular indeﬁnites while Study 5 tested bare plurals and bare singulars, suggests that
Chilean children often associate the indeﬁnite plural to an interpretation of ‘one’ while
they associate the bare plural and bare singular to an interpretation of ‘more than one’.
While the results of Study 5 indicate that ChMC children distinguish between bare
singulars and bare plurals, it appears that ChWC children do not, which suggests that
ChWC children do not associate the plural morpheme to an interpretation of ‘more than
one’ rather they associate bareness to an interpretation of ‘more than one’. Further
evidence supporting this idea comes from the production data presented in the Naming
Task in Chapter 4, which showed that ChWC children produced bare singulars when

describing plural sets 53% (90/171) of the time and bare plurals 47% (81/171) of the

241

time, while ChMC children only produced bare singulars 15% (6/40) of the time, ChWC

Adults 9% of the time (2/22) and ChMC adults 13% (6/48) of the time.

5.5.7 Summary of Findings

The following is a summary of the results for the comprehension of plural
morphology in Study 5.

1. ChWC children do not treat bare plurals differently ﬁ'om bare singulars while
ChMC children do.

2. Chilean children overwhelmingly treat bare plurals as plural and singular
indeﬁnites as singular, which is consistent with their production in the Naming Task in
Study 4.

3. ChWC children associate both bare singulars and bare plurals with an
interpretation of ‘more than one’. They appear not to associate the plural variant [s] to an
interpretation of ‘more than one’.

4. ChMC children associate the plural variant [8] to an interpretation of ‘more
than one’.

5. The ﬁndings of Study 5 are consistent with Studies 1, 2, 3, and 4.

5.6 Study 6. Act-out Task: Inalienable Possession

All of the studies presented up until now have dealt with indeﬁnite and bare noun
phrases. Study 6 examines children’s ability to associate the plural morpheme to an
interpretation of ‘more than one’ in deﬁnite noun phrases. It is important to test deﬁnite
noun phrases, in order to determine whether Chilean children, especially ChWC children,
incorrectly associate the plural morpheme to an interpretation of ‘one’ in other contexts

or whether it is the lexical nature of the plural indeﬁnite (its similarity in form to the

242

numeral ‘one’) that is causing difﬁculty for Chilean children. Deﬁnite noun phrases in
constructions involving inalienable possession are tested in Study 6 and referential

deﬁnite noun phrases are tested in Study 7.

5.6.1 Background
Study 6 investigates Chilean and Mexican children’s interpretation of the plural
feminine deﬁnite noun phrases: Ia vs. las (the.F.SG vs. the.F.PL) in constructions involving

inalienable possession, as in (16).

(16) a. Tocale la rodrilla a Carlita.
Touch.her the.SG knee.SG to Carlita

‘Touch Carlita’s knee.’

b. Tocale las rodillas a Carlita,
Touch.her the.PL knees.PL to Carlita

‘Touch Carlita’s knees.’

In (16a) la rodilla (‘the.SG knee.SG’) refers to only one of Carlita’s knees while in (16b)
las rodillas (‘the.PL knees.PL’) refers to both of Carlita’s knees. Study 6 asks the
following questions: (1) Given the differences in the input, will Chilean children differ
from Mexican children in their comprehension of deﬁnite plural noun phrases? (2) Will
Chilean children with variable input associate the plural morpheme to the interpretation
of ‘more than one’? (3) Will Chilean children treat both [s] and [h] as associated to an

interpretation of ‘more than one’.

243

5.6.2 Method and Design
An Act Out Task was used. Subjects were shown a dolls (Figure 16) and asked to

carry out activities as stated in (17a) and (17b).

 

(17) a. Tocale la rodilla.
Touch.her the.F.SG knee.F.SG

‘Touch her knee.’

b. Tocale las rodillas.
Touch.her the.F.PL knees.F.PL

‘Touch her knees.’

The plural morpheme was always pronounced as [s] in the ﬁrst experiment
administered to Mexican and Chilean children and as [h] in the second experiment
administered only to Chilean children. The deﬁnite nouns phrases were always feminine

so that the only difference between the plural and singular noun phrases was the presence

244

or absence of the plural maker. Experimental Sentences are shown in APPENDIX F. In
the target experimental sentences only the plural morpheme provides number information
(la rodilla ‘the.SG knee.SG’ vs. las rodillas ‘the.PL knee.PL’). Controls involved feminine
deﬁnite noun phrases that were referential (Dame las vacas ‘Give me the.PL cows.PL’ vs.

Dame la vaca ‘Give me the.SG cow. SG’).

5.6.3 Subjects

81 subjects participated in this study. 12 MexWC (4;7-5;6, Mean: 5:1), 16 ChWC
(4;5-5;11, Mean: 5:3), 10 ChMC (4;6-5:11, Mean: 5;2) children and 20 adults (12
Chilean and 10 Mexican) participated in this study with the plural morpheme pronounced
as [s]. Additionally, 14 ChWC (4;8-6:4, Mean: 5:4) and 7 ChMC (4;11-6;l, Mean: 5:6)
children participated in this study with the plural morpheme pronounced as [h]. The
Chilean children were recruited from preschools and daycares in Punta Arenas, Chile and
the Mexican children from a daycare in Mexico City. Chilean adults were undergraduates
at the Universidad de Magallanes in Punta Arenas, Chile and the Mexican adults were
undergraduates at the Universidad Autonoma Metropolitana de Iztapalapa in Mexico

City. Table 57 shows distribution of child subjects.

245

Table 57. Study 6: Distribution of Children

 

 

 

Pronunciation Number Age Range Mean Age
MexWC [s] 12 - 4;7-5:6 5:1
ChMC [s] 10 4:6-5:1 1 5:2
[h] 7 4:11-6;1 5;6
ChWC [s] 16 4;5-5;11 5;3
[h] 14 4;8-6:4 5;4
5.6.4 Procedure

All subjects were tested by native speakers of Spanish who lived in the same city
as the subjects. The author of this dissertation was present during the testing of all
children to ensure that procedures were identical for all subjects. All children and adults
were able to perform well on controls and no children were discarded ﬁom this study.
Subjects were presented ﬁrst with the 4 singular deﬁnite target and control trials followed
by the 4 plural deﬁnite target and control trials. Experimental sentences were presented in
this order to avoid priming a plural response in Chilean children. Given that the plural
morpheme can be omitted in Chilean adult Spanish, it might be the case that children who
ﬁrst hear a plural target trial might believe that subsequent singular trials are actually
plural trials where the researcher is omitting the plural morpheme. Masculine deﬁnites
were not tested because the singular masculine deﬁnite article is different in form from

the plural masculine deﬁnite (el ‘the.M.SG’ vs. Ios ‘the.M.PL’).

246

5.6.5 Results

Although all children performed 100% on controls, they did not all perform the
same in the target conditions. The dependent variable is the number of plural answers
children gave. Moving two body parts was considered a plural response. Moving only
one body part was considered a singular response. Table 58 shows the overall percentage

of plural responses for MexWC, ChMC, and ChWC children.

Table 58. Study 6: Plural Responses in Target Conditions.

las (‘the.PL’) la (‘the.SG’)

 

Plural Morpheme Pronounced as [s]

MexWC 71% 0%
(34/48) (0/48)

ChMC 50% 10%
(20/40) (4/40)

ChWC 58% 19%
(37/64) (12/64)

Plural Morpheme Pronounced as [h]

ChMC 93% 0%
(26/28) (0/28)

ChWC 25% 5%
(14/56) (3/56)

 

247

The number of plural responses in the plural deﬁnite condition was entered into a one-
way AN OVA. The results showed a signiﬁcant difference between the four [s]-groups
(F(l,59)=5.582,p<.01). [s]-ChMC (p<.05) and [s]-ChWC (p<.01) children behaved
signiﬁcantly different ﬁ'om adults but [s]-MexWC children (p=. 150) did not. However,
the three child groups did not differ signiﬁcantly ﬁ'om each other: [s]-MexWC vs.[s]—
ChWC (p=1 .0), [s]-MexWC vs. [s]-ChMC (p=l.0), [s]-ChMC vs. [s]-ChWC (1.0).

Table 58 shows an increase in plural responses for ChMC children when the
plural was pronounced as [h] but a decrease in plural responses for the ChWC children
when the plural was pronounced as [h]. Independent samples t-tests showed a marginally
signiﬁcant decrease in plural responses in the [h] part of the study for ChWC children
(t(l ,28)=—2.030,p=.052) but no signiﬁcant increase in the number of plural responses in
the [h] part of the study for ChMC children (t(l,15)=1.597,p=. 131). The inability to ﬁnd
signiﬁcant differences here is most likely due to the few number of subjects in the [h]
version of the study.

The number of plural responses for the [h]-ChWC, [h]-ChMC, [s]-MexWC and
[s]-Adults was entered into a one-way ANOVA. The results showed signiﬁcant
differences between the three groups (F(l,54)=21.347,p<.001). Post hoc Bonferroni tests
showed a signiﬁcant difference between adults and [h]-ChWC children (p<.001) but not
between adults and [h]-ChMC (p=1 .0). [h]-ChWC children differed signiﬁcantly from
both [s]-MexWC children (p<.001) and [h]-ChMC children (p<.001); however, [h]-
ChMC and [s]-MexWC children did not differ signiﬁcantly from each other (p=.644).
When the plural morpheme is pronounced as [h] for the ChMC children, they behave

more closely to adults than MexWC children do on deﬁnite plurals. These ﬁndings

248

suggest that ChMC children associate [h] to an interpretation of ‘more than one’ while
ChWC children do not.

Studies 1, 2, and 3 showed that children were systematic in their responses, while
Study 5 on bare plural and singular nouns showed that ChMC children were not
systematic, but rather appear to be variable in their responses on the bare singular, which
was an acceptable response. Table 59 and Table 60 show the percentage of systematic

responders for Study 6.

Table 59. Study 6: Las: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(3-4 plural) (3-4 siggular) Total (2 pl/ 2sg)_

Plural Pronounced as [s]

MexWC 67% 25% 92% 8%

ChMC 60% 40% 100% ' 0%

ChWC 56% 38% 94% 6%
Plural Pronounced as [h]

ChMC 86% 0% 86% 14%

ChWC 21% 71% 92% 8%

 

249

Table 60. Study 6: La: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(3-4 plural) (3-4 singular) Total (2 pl/ Zng

 

Plural Pronounced as [s]

MexWC 0% 100% 100% 0%

ChMC l 1% 89% 100% 0%

ChWC 19% 81% 100% 0%
Plural Pronounced as [h]

ChMC 0% 100% 100% 0%

ChWC 0% 93% 93% 7%

 

Table 59 and Table 60 show that the majority of children were systematic in their
responses. There were only 4 variable responders in the plural deﬁnite condition, 1
MexWC, 1 [s]-ChWC, 1 [h]-ChMC, and l [h]-ChWC child in the plural deﬁnite

condition and l [h]-ChWC child in the singular deﬁnite condition.

5.6.6 Discussion

First of all, Study 6 seemed to be more difﬁcult for children than the previous
experiments, which may have to do with the fact that it involved inalienable possession.
Given the sentence T irale las orejas (‘Pull his ears’) it is not necessarily an incorrect
response to just pull one car. This may explain why more Mexican children treat the
plural as singular; however, note that MexWC children did not behave signiﬁcantly

different from adults. In any case, given that all child groups were tested on the same

250

experiment, it is interesting to note that while [s]-ChMC, [s]-ChWC, and [h]-ChWC
children behaved signiﬁcantly different from adults, MexWC and [h]-ChMC children did
not.

Another interesting ﬁnding is that ChMC children provide more plural responses
(even more so than the MexWC children) for the plural deﬁnite when it is pronounced
with aspiration [h], while the opposite is true for the ChWC children. This ﬁnding is
interesting given that ChMC children had fewer plural responses in Studies 1 and 2 when
the plural was pronounced as [h], preferring more [8]. Instead this ﬁnding is consistent
with Study 3, which found ChMC children allowing both [s] and [h] as the plural
morpheme. It is not clear why we ﬁnd these differences existing between the studies.
Because these studies involve different children, it appears that within the Chilean child
groups there is a lot of variability in whether children prefer the [s] or [h] variant, or both.
The ﬁndings of Study 6 indicate that, while MexWC children associate the plural
morpheme on the deﬁnite noun phrases to an interpretation of ‘more than one’, ChWC
children do not and ChMC children do only for the plural variant [h] but not for the
variant [s]. These ﬁndings for ChWC children support the Variability Delay Hypothesis,
which proposes that variability in the input will delay child comprehension of
grammatical forms when the variability causes an ambiguity (involves omission) and is
constrained not only by linguistic (phonological, grammatical) but also extra-linguistic
(SES, age, sex) factors. However, it is not clear that the ﬁndings for ChMC children
support the Variability Delay Hypothesis, given that the [h]-ChMC children performed at
ceiling on this task. From the production data presented in Chapter 4 and in Cepeda

(1995) it appears that there are more omissions in the speech of working-class adults than

251

in the speech of middle-class adults. This may explain why ChMC children associate [h]
to an interpretation of ‘more than one’ , at least in deﬁnite noun phrases more often than

ChWC children.

5.6.7 Summary of Findings

The following is a summary of the results for the comprehension of plural
morphology in Study 6.

1. [s]—ChWC, [h]-ChWC and [s]-ChMC children do not in general associate the
plural morpheme in deﬁnite noun phrases to an interpretation of ‘more than one’.

2. [h]-ChMC and MexWC children associate the plural morpheme in deﬁnite
noun phrases to an interpretation of ‘more than one’.

3. ChMC children associate [h] to an interpretation of ‘more than one’ but not [s].
ChWC children associate neither plural variant to a ‘more than one’ interpretation.

4. The ﬁndings of Study 6 are not consistent with Studies 1 and 2, which show
that Chilean children do not associate [h] to an interpretation of ‘more than one’;
however, the ﬁndings for the [h]-ChMC children are consistent with Study 3, which do
show that Chilean children associate [h] to an interpretation of ‘more than one’.
Differences in ﬁndings appear to be related to the fact that Studies 1, 2, and 3 focused on

indefmites and Study 6 on deﬁnites.

5.7 Study 7. Act-out Task: Deﬁnite Noun Phrases

Thus far the ﬁndings ﬁom Studies 1-6 have shown that overall there are
differences between Chilean and Mexican children in the development of plural
morphology. While Mexican children consistently associate the plural morpheme to an

interpretation of ‘more than one’, regardless of noun phrase type, Chilean children vary

252

in their interpretation of the plural morpheme depending on noun phrase type (indeﬁnite
plurals vs. bare plurals vs. deﬁnite plurals) and also depending on whether their plural is
pronounced as [s] or [h] in the experimental task. These ﬁndings suggest that overall
ChWC children do not associate the plural morpheme to an interpretation of ‘more than
one’ while ChMC children do. In addition, the behavior of ChMC children is dependent
on noun phrase type, with better performance on bare nouns and deﬁnites than on
indeﬁnite noun phrases.

On the same token, the ﬁndings of Study 6 showed that in general the task may
have been slightly more difﬁcult than previous studies given the fewer number of ‘more
than one’ responses in Mexican children. I suggested that the inalienable construction
may have been responsible for this difference given that in some cases a singular
response in a plural condition is acceptable. For this reason, the purpose of Study 7 is to
test Chilean and Mexican child interpretation of the plural morpheme in deﬁnite noun

phrases with a referential interpretation.

5.7.1 Background
The ﬁnal study in this dissertation, Study 7, tests children on deﬁnite noun

phrases, as in (18).
(18) a. Dame la mufleca durrniendo a1 lado de la casa.

Give.me the.F.SG doll.F.SG sleeping next to the house

‘Give me the doll sleeping next to the house.’

253

b. Dame las muﬂecas durmiendo a1 lado de la casa.
Give.me the.F.PL dolls.I-‘.PL sleeping next to the house

‘Give me the dolls sleeping next to the house.’

Study 7 asks the following questions: (1) Given the differences in the input, will Chilean
children differ ﬁ'om Mexican children in their comprehension of plural deﬁnite noun
phrases? (2) Will there be an increase in plural responses for all three child groups in
Study 7 as compared to Study 6, given that the deﬁnite is not in a construction involving

inalienable possession?

5.7.2 Method and Design
An Act-out Task was used. A display as diagrammed in Figure 17 was presented

to subjects followed by commands as in (18a) and (18b) above.

Figure 17. Study 7: Diagram of Sample Target Trial.

 

 

 

 

 

 

 

 

 

House Barn
Doll A Doll 1
Doll B Doll 2
Doll C Doll 3

 

 

 

254

There were two controls: una (‘a/one.sg’) and todas (‘all.PL’). As a result, there were 4
trials for each of the 4 noun phrase types (todas ‘all.PL’, una ‘a/one.PL’, las ‘the.PL’, la
‘the.SG’). In addition, there were 8 ﬁllers that involved questions about the animals in the
scene. All nouns were feminine so that the only difference between the plural and
singular forms was the presence or absence of the plural morpheme. The nouns were:
muﬁecas ‘dolls’, vacas ‘cows’, hormigas ‘ants’, arar'ias ‘spiders’. All toys were miniature
so that children could easily pick up more than one toy at a time.

The materials and methods were the same for all children except there were 20
Mexican children who received a different version of the target sentences, as shown in
(19a) and (19b). 10 Mexican children received the target sentences as in (18a) — (18b)
above and 20 received the target sentences with participles as (19a) and (19b) below.
This was done because the Mexican researchers preferred sentences with participles like

(19) while Chilean researchers preferred the sentences with gerrmds (18).

(19) a. Dame la mufleca dormida a1 lado de la casa.
Give.me the.SG doll.SG asleep.SG next to the house

‘Give me the doll (that are) asleep next to the house.’
b. Dame las mufiecas dormidas a1 lado de la casa.

Give.me the.PL dolls.PL asleep.PL next to the house

‘Give me the dolls (that are) asleep next to the house.’

255

If children hear the sentence with the plural deﬁnite (18b or 19b), they should choose the
maximal set of toys next to the house (i.e. Dolls A, B, and C). If they hear the sentence
with the singular deﬁnite (18a orl9a), they should choose the one toy that is closest to the
house (i.e. Doll A). The dolls in ﬁ'ont of the barn (Dolls 1, 2, and 3) are needed to make
the plural deﬁnite condition felicitous. In other words, a modifying relative clause is only
needed if there is an alternative set of dolls the speaker could be referring to.

Responses were counted as plural if the subject chose ‘more than one’ toy and
singular if subjects chose only one toy. The plural morpheme was pronounced as [s] for
all children. The plural morpheme pronounced as [h] was not tested in this study. All

experimental sentences are shown in Appendix G.

5.7.3 Subjects

74 subjects participated in this study. 30 MexWC children (3;2-6;2, Mean: 4:6),
10 ChMC children (4:2-6;1, Mean: 5:0), and 12 ChWC children (4;3-5;11, Mean: 4:9)
and 22 adults (14 Chilean and 8 Mexican) participated in Study 7. Mexican children and
adults were recruited from a preschool and the Universidad Autonoma Metropolitana de
Iztapalapa in Mexico City. The Chilean children and adults were recruited from

preschools and the Universidad de Magallanes in Punta Arenas, Chile.

5.7.4 Procedure

The procedure and materials were the same for all children. Subjects were tested
by native-speakers who were ﬁ'om the same town as the subjects. The author of this
dissertation was present at all testing to ensure that procedures were the same. Children
were ﬁrst tested on their understanding of the preposition al lado de (‘next to’). All

children showed understanding of the prepositional phrase. Next, children were presented

256

with one or two practice trials involving todos (‘all.PL’) and un (‘a/one’) to ensure that
they understood that they had to pay attention to the quantity of objects the researcher
was asking for. Next, the researcher presented the main part of the experiment, as shown

in Figure 18.

_Figure 18. Study 7: Sample Experimental Story.
Researcher: “Mira. Muﬁecas. Y durmiendo, durmiendo, durmiendo, durmiendo,

 

durmiendo, y durmiendo (researcher says “durmiendo ” as he lays down each
doll). Yo te voy a decir cudntas me tienes que dar y til me da la cantidad que yo
te digo. gT e parece? ” “Look Dolls. And sleeping, sleeping, sleeping, sleeping,
sleeping, and sleeping (researcher says each gerund as he lays down each doll).
I am going to tell you how many you have to give me and you give me the

quantity that I tell you. Ok?”

 

 

 

Following this presentation, the researcher preceded with the experimental questions.

5.7.5 Results

All children behaved the same on the control todas (‘all.PL’), giving the maximal
set to the researcher 100% of the time and the control una (‘a/one.SG’), giving only one
item almost all of the time. However, children did not behave the same in all of the target
conditions. Table 61 shows the percentage of plural responses that subjects provided for

the plural and singular defmites.

257

Table 61. Study 7: Plural Responses in Target Conditions.

 

 

la las
(‘the.F.SG’) (‘the.F.PL’)
Adults 0% 100%
MexWC 8% 96%
(10/120) (115/120)
ChMC 18% 80%
(7/40) (32/40)
ChWC 50% 90%
(24/48) (43/48)

 

 

Looking ﬁrst at the plural deﬁnite condition we see that all four groups associated the

plural deﬁnite to a ‘more than one’ interpretation more than 80% of the time. A one-way

AN OVA showed that all four groups (Adults, MexWC, ChMC and ChWC) differed

signiﬁcantly in the number of plural responses they provided in the deﬁnite plural

condition las (F(3,73)=4.668,p<.01). Post hoc Bonferroni tests revealed that only ChMC

children differed signiﬁcantly from adults (p<.01), while ChWC (p=.327) and MexWC

(p=1 .0) children did not. Post hoc Bonferroni tests also showed that ChMC differed

signiﬁcantly from MexWC children (p<.05) but not from ChWC children (p=.817) on the

deﬁnite plural. ChWC and MexWC children did not differ from each other (p=l.0).

Looking just at the plural deﬁnite condition, it appears that all four groups associate the

258

plural variant [s] on plural deﬁnites with an interpretation of ‘more than one.’ However,
the results for the singular deﬁnites must also be considered.

In the singular deﬁnite condition, while the MexWC and ChMC children in
general associated the singular deﬁnite to an interpretation of ‘one’, the ChWC children
associated the singular deﬁnite to an interpretation of ‘more than one’ about 50% of the
time. In the singular deﬁnite condition, a one-way AN OVA showed that all forn' groups
differed signiﬁcantly in the number of plural responses they provided
(F(3,73)=16.723,p<.001). Post hoc Bonferroni tests showed that only ChWC children
differed signiﬁcantly from adults (p<.001) but not ChMC (p=. 167) nor MexWC (p=.904)
children. ChWC children also behaved signiﬁcantly different from the ChMC (p<.01)
and MexWC (p<.001) children by providing more plural responses in the singular
deﬁnite condition. ChMC and MexWC children did not differ from each other (p=1 .0).
The results show that, while all groups treated the plural deﬁnite as plural, they differ in
their interpretation of deﬁnite singulars. ChMC and MexWC children associate the
singular deﬁnite to ‘one’ while ChWC children associate the deﬁnite singular to an
interpretation of ‘more than one’ approximately 50% of the time.

Studies 1, 2, 3 and 6 showed that children were systematic in their response
patterns. The only time children were not systematic in their response patterns was in
Study 5 where ChMC children associated the bare singular to both a ‘more than one’ and
‘one’ interpretation. The response patterns of children in Study 7 are shown in Table 62

and Table 63.

259

Table 62. Study 7: Las: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

 

(3-4 plural) (3-4 singular) Total (2 pl/ 28g)_

Plural Pronounced as [s]

MexWC 100% 0% 100% 0%
ChMC 80% 20% 100% 0%
ChWC 83% 0% 83% 17%

 

Table 63. Study 7: La: Systematic Responders.

 

Systematic Systematic Variable
Plural Singular Plural
Response Response Systematic Response

(3-4 plural) (0-1 singular) Total (2 pl/ 2sgL

 

Plural Pronounced as [s]

MexWC 7% 93% 100% 0%
ChMC 10% 80% 90% 10%
ChWC 42% 33% 75% 25%

 

The results in the above tables show that in both the plural deﬁnite and singular deﬁnite
conditions children were generally systematic in their response patterns. However, the
tables also show that 2 ChWC children were variable in their response patterns in the
deﬁnite plural condition and 3 ChWC children were variable in their response patterns in

the plural deﬁnite condition.

260

5.7.6 Discussion

Study 7 showed that ChWC children, but not ChMC children, differ from
MexWC children in their comprehension of plural and singular deﬁnites. While ChMC
children and MexWC children associate the plural deﬁnite with an interpretation of ‘more
than one’ and the singular deﬁnite with an interpretation of ‘one’, approximately half the
ChWC children associated both the singular deﬁnite and the plural deﬁnite 'with an
interpretation of ‘more than one’. These ﬁndings indicate that while ChMC and MexWC
children associate the plural morpheme in deﬁnite noun phrases with an interpretation of
‘more than one’, ChWC children do not. Rather, the increase in plural responses in the
plural deﬁnite condition is accompanied by an increase in plural responses in the singular
deﬁnite condition, which suggests that the plural response has to do with the deﬁniteness
of the noun phrase and not the plural morpheme.

The ﬁndings of Study 7 also show that, unlike Study 6, ChMC children associate
the plural variant [8] with an interpretation of ‘more than one’. It is unclear why we ﬁnd
differences among the various comprehension studies as to which of the two plural
variants ChMC children associate with ‘more than one’. It may simply be due to
differences between the children who participated in each of the studies. Future research
is needed to clarify these ﬁndings.

Finally, Study 7 showed a slight increase in plural responses for MexWC children
when compared to Study 6 with construction involving inalienable possession, which

suggests the experimental task in Study 7 may have been easier for children.

261

5.7.7 Summary of Findings

The following is a summary of the results for the comprehension of plural
morphology in Study 7. i

1. [s]-ChWC children do not associate the plural morpheme in deﬁnite noun
phrases to an interpretation of ‘more than one’.

2. [s]-ChMC and MexWC children associate the plural morpheme in deﬁnite
noun phrases to an interpretation of ‘more than one’.

3. The ﬁndings of Study 7 are consistent with Studies 1 and 2, which show that
ChWC children do not associate [s] with an interpretation of ‘more than one’; however,
they are inconsistent with Study 6, which shows that ChMC children do not associate [s]
to an interpretation of ‘more than one’. In Study 7 ChMC children do associate [s] to
‘more than one’. It appears that there is more variability in the comprehension of plural

morphology by ChMC children than for ChWC children.

5.8 Summary and Discussion

The ﬁndings of the comprehension studies in Chapter 5 show different results for
the three groups of children. For the MexWC children the results show that they associate
the plural morpheme to an interpretation of ‘more than one’ at 5 years of age. Studies 1,
2, and 3 show that they assign a ‘more than one’ interpretation to indeﬁnite noun phrases
unos/unas (‘some.PL’) and algunos/algunas (‘some.PL’); however, Study 1 indicates that
MexWC children do not assign a ‘more than one’ interpretation to overt partitives
algunos de los (some.PL of the.PL). It is unclear why MexWC children had difficulty with
plural overt partitives. This will be left to future research. The results of Studies 6 and 7

show that MexWC children associate the plural morpheme to an interpretation of ‘more

262

 

than one’ in deﬁnite noun phrases. Finally, the results of Study 4 show that MexWC
children prefer to produce singular-plural minimal pairs in their description of singular
and plural sets (e. g. una vaca ‘a/one.SG cow.SG’ vs. unas vacas ‘some.PL cows.PL’).
Taken together, Studies 1—7 show that MexWC children associate the plural morpheme to
an interpretation of ‘more than one’.

For the ChWC children the results indicate that most children do not associate the
plural morpheme to an interpretation of ‘more than one’ in any of the experimental tasks
carried out. In Studies 1, 2, and 3 ChWC children assign an interpretation of ‘one’ to both
plural and singular indeﬁnite noun phrases unos/unas (‘some.PL’) and algunos/algunas
(‘some.PL’) and in Studies 6 and 7 they assign a ‘more than one’ reading to both plural
and singular deﬁnite noun phrases. The results also indicate that ChWC children do not
distinguish between bare singulars and bare plurals, instead they assign a plural reading
to all bare nouns. Moreover, the results show that ChWC children do not reach adult
levels regardless of whether the plural is pronounced as [s] or [h] in the experimental
study. The ﬁndings in Studies 1-7 show that ChWC children do not associate the plural
morpheme to an interpretation of ‘more than one’. The results for ChWC children, taken
together with the results for MexWC children, support the Variability Delay Hypothesis.

Finally, the results for the ChMC children indicate that they associate the plural
morpheme to an interpretation of ‘more than one’ in the experimental tasks carried out:
however, their performance at times does not reach adult levels. Studies 1, 2 and 3 show
that ChMC children assign an interpretation of ‘one’ to the plural indeﬁnite unos/unas
(‘some.PL’). However, when the plural indeﬁnite is highlighted, as in Study 1, ChMC

children associate the plural indeﬁnite to ‘more than one’. Study 1 also showed that

263

ChMC children associated the plural indeﬁnite algunos (‘some.pl’) to ‘more than one’
but that many ChMC also incorrectly associated the singular overt partitive uno de los
(one.SG of the.PL’) to an interpretation of ‘more than one’. This suggests that ChMC
children were hypercorrecting, assuming that the researcher omitted the plural morpheme
on uno (‘one.SG’). Moreover, Studies 6 and 7 show that ChMC children associate plural
deﬁnites to ‘more than one’ and singular deﬁnites to ‘one’. Study 5 indicates that ChMC
children distinguish between bare singulars and bare plurals. While ChMC children, like
MexWC children, appear to associate the plural morpheme to an interpretation of ‘more
than one’, they behave differently from MexWC children on the plural indeﬁnites
unos/unas and the overt partitive uno de los. Moreover, while MexWC children prefer the
plural indeﬁnites in their description of plural sets in Study 4, ChMC children behave like
ChWC children in preferring bare plurals. This suggests that MexWC children and
ChMC children differ in their development of plural morphology. ChMC children appear
to pay attention to both lexical information of the determiner and also the plural
morpheme. When there is a conflict between the two, as is the case with the indeﬁnite
plural unos/unas, which is the numeral ‘one’ uno/una + the plural morpheme, ChMC

children have difﬁculty with their interpretation.

264

CHAPTER 6

SUIVIMARY AND CONCLUSIONS

6.1 Goal of Dissertation

This dissertation assmnes that language acquisition involves two components: an
innate component, often referred to as a language acquisition device (LAD), and the
input. While I assume that the LAD is invariable across typically developing populations,
the input that children are exposed to varies. The purpose of this dissertation was to
examine how different types of input affects the acquisition of grammatical morphemes
in children. Speciﬁcally, this research asked whether children exposed to systematic vs.
variable input would differ in their development of grammatical morphology. In order to
answer this question, the acquisition of plural morphology in two varieties of Spanish
was examined, Mexican Spanish (Mexico City) where plural morphology is
systematically produced by adult speakers and Chilean Spanish where plural morphology
has a variable behavior and is sometimes omitted by adult speakers.

With respect to production, it was predicted that children exposed to an input that
systematically marks plural morphology would also systematically mark plural in their
own production and that children exposed to variable plural marking would be variable in
their own production. This prediction is based on previous research showing that by 3
years of age children produce plural morphology systematically when exposed to
systematic input (Berko 1958, Keman and Blount 1966, Brown 1973, Cazden 1968,
Perez-Pereira 1989, Mervis and Johnson 1991, Kvaal et a1. 1988, Marrero and Aguirre
2003) and that children exposed to variable input are also variable in their own speech

even as early as 4 years of age (Ramer and Rees 1973, Moore 1979, Kovac and Adamson

265

1981, Washington and Craig 1994, Roberts 1994, 1997). However, it was also predicted
that children exposed to variable input may not have acquired all the linguistic and extra-
linguistic constraints governing that variability even at 7 years of age. This prediction is
consistent with research by Roberts (1994, 1997) and Smith et al. (2006).

There are almost no studies examining the effect of variable input on the
comprehension of grammatical morphemes and very few experimental studies on the
comprehension of plural morphology in children exposed to systematic input. The few
studies that do exist have been carried out recently in English Kouider et al. 2006, Munn
et al. 2006). As far as we know, there are no studies that have examined the
comprehension of plural morphology in Spanish-speaking children exposed to systematic
input. Given what little we know about the comprehension of plural morphology in
children, there were two possible outcomes of the experimental studies presented in this
dissertation: (1) No differences would exist between children exposed to systematic vs.
variable input. Rather, as long as the plural morpheme is available in the input at least
some of the time, both Mexican and Chilean children would comprehend the plural
morpheme by 5 years of age. (2) Differences would exist between children exposed to
systematic vs. variable input. Children exposed to systematic input would acquire plural
morphology earlier than children exposed to variable input. In other words, variable input
involving omission would create an ambiguity and this ambiguity would cause a delay in
comprehension, in associating the plural morpheme to an interpretation of ‘more than
one”. Based on research suggesting that variable input involving omission delays

comprehension of grammatical morphemes (Moore 1979, Johnson 2005), we predicted

266

outcome (2), that differences would be found between Chilean vs. Mexican children and

we proposed the Variability Delay Hypothesis, which is based on Yang (2000, 2002).

(1) Variabiligg Delay Hymthesis (based on Yang 2000, 2002): Variability in
the input will delay child comprehension of grammatical morphemes when the variability
causes an ambiguity in the input (involves a zero form) and is constrained not only by

linguistic (phonological, grammatical) but also extra-linguistic (SES, age, sex) factors.

Differences in comprehension of plural morphology between Mexican and Chilean child
groups would lend support to the Variability Delay Hypothesis. If no differences are

found, the variability hypothesis would be rejected.

6.2 Summary and Discussion of Findings

The research presented in this dissertation collected experimental data on both the
production and comprehension of plural morphology by Chilean and Mexican Spanish-
speaking children. The ﬁndings of the production data will ﬁrst be discussed followed by
the ﬁndings of the comprehension data.

Production data was collected from both Chilean and Mexican adults and
children. There were three experimental tasks that varied in degree of formality: a Free
Speech task, a Repetition Task and a Naming Task. The main ﬁndings of the production
data showed that ( 1) Mexican children and adults systematically produce the plural
morpheme as [s] (and [2]) while Chilean adults and children have variable production,
producing the plural morpheme as [s], [h] and also omitting the plural morpheme on

semantically plural elements within the determiner phrase; (2) Chilean working-class

267

children and adults both omit the plural morpheme more often than Chilean middle-class
children and adults; (3) While Chilean working-class and middle-class children are
variable in their production of the plrn‘al morpheme, they have not yet acquired all of the
linguistic and extra-linguistic constraints governing that variability at 5 years of age.
These ﬁndings suggest that systematic input results in systematic production in children
while variable input results in variable production in children, even if children have not
yet acquired all of the constraints governing the variability. The ﬁndings also suggest that
at 5 years of age, children are still acquiring the constraints that govern variation in their
language. These ﬁndings are consistent with our predictions and with what has
previously been reported in the literature on child production of plural morphology and
child production of variation.

This dissertation presented 8 experimental studies on child comprehension of
plural morphology. The comprehension studies examined whether children exposed to
systematic vs. variable input distinguished between plural and singular indeﬁnite noun
phrases, deﬁnite noun phrases and bare nouns. Mexican children were tested on the plural
morpheme /s/ (realized as [s] and [2]) while Chilean children were tested on both [s] and
[h]. The overall ﬁndings of the comprehension data showed that (1) Chilean children
differ ﬁ'om Mexican children in their comprehension of the plural morpheme on
indeﬁnite noun phrases. Mexican children associate plural indeﬁnites to an interpretation
of ‘more than one’ while many Chilean middle-class children and several Chilean
working-class children do not: (2) Chilean children, who have plural morphology, may
associate either [8] and/or [h] to an interpretation of ‘more than one’: (3) Mexican

children consistently produce singular-plural indeﬁnite noun phrase minimal pairs (mm

268

vaca ‘a/one.SG cow.SG’ vs. unas vacas ‘some.PL cows.PL) when describing singular and
plural sets while Chilean children do not. Instead, Chilean children consistently produce
bare plurals (and sometimes bare singulars) vs. singular indefmites (vacas ‘cows.PL’ or
vaca ‘cowﬁ’ vs. una vaca ‘a/one.SG cow.SG’); (4) Chilean working-class children treat
bare singulars and bare plurals the same, associating both to an interpretation of ‘more
than one’ while Chilean middle-class children treat bare singulars and bare plurals
differently, associating the bare plural to an interpretation of ‘more than one’ and
allowing both a singular and plural interpretation of the bare singular; (5) Mexican
children associate the plural morpheme to an interpretation of ‘more than one’ in deﬁnite
noun phrases involving inalienable possession, while many Chilean middle-class children
associate [h] but not [s] to an interpretation of ‘more than one’ and many Chilean
working-class children do not associate neither [3] nor [h] to an interpretation of ‘more
than one’ on deﬁnite noun phrases involving inalienable possession: and (6) Mexican
children and Chilean middle-class children distinguish between‘plural vs. singular
deﬁnite noun phrases that are referential in nature when the plural is pronounced as [s]
while Chilean working-class children do not.

Based on these overall ﬁndings for comprehension, we can conclude that by 5
years of age Mexican working-class children have acquired plural morphology. In other
words, Mexican children associate the plural morpheme to an interpretation of ‘more than
one’. In this sense, Mexican child comprehension patterns with Mexican adult
comprehension. On the other hand, the results strongly suggest that most 5-year-old
Chilean working-class children do not yet associate plural morphology (pronounced as

[s] or [h]) to an interpretation of ‘more than one’. This seems to be the case regardless of

269

the type of noun phrase (e. g. indeﬁnite, deﬁnite, or bare) that Chilean working-class
children are tested on. Finally, the results for the Chilean middle-class children are a bit
more difﬁcult to interpret. The ﬁndings on indeﬁnite noun phrases suggests that Chilean
middle-class children have a strong tendency to initially assign an interpretation of ‘one’
to plural indeﬁnites un/una (‘a/one.SG’) and unos/unas (‘some.PL’) and that they avoid
using plural indefmites when expressing an interpretation of ‘more than one’. Instead,
they rely on bare plurals. However, Chilean middle-class children appear to associate the
plural morpheme [s] and [h] to an interpretation of ‘more than one’ on deﬁnite noun
phrases and bare nouns. Taken together, the ﬁndings for Chilean middle-class children
suggest that most of them associate the plural morpheme [s] and/or [h] to an
interpretation of ‘more than one’ but that indeﬁnite noun phrases cause some extra
distraction, which is most likely due to the fact that the indeﬁnite determiner is also the
word for ‘one’ in Spanish. Importantly, the indeﬁnite noun phrase does not cause any
distraction for Mexican working-class children, which suggests that the variable input to
Chilean children still affects Chilean middle-class children in their development of plural
morphology to some extent.

The ﬁndings of this dissertation also reveal that production of plural morphology
appear to precede comprehension of the plural morpheme. While Chilean children
produced the plural morpheme in their own speech some of the time, they did not always
associate the plural morpheme to an interpretation of ‘more than one’. This ﬁnding is
consistent with previous research that has shown that English-speaking children start

producing the plural morpheme at around 1:10 years of age (Ferenz and Prasada 2002)

270

but do not associate the plural morpheme to an interpretation of ‘more than one’ until 3
years of age (Kouider et al. 2006).

A ﬁnding of this dissertation that is difﬁcult to explain is Chilean child
performance in comprehension tasks when the plural morpheme was pronounced as [s]
vs. [h]. While the overall ﬁnding was that several Chilean children do not associate [s]
nor [h] to an interpretation of ‘more than one’, among those children who did appear to
have plural morphology, the percentage of times that [s] or [h] was associated to ‘more
than one’ varied between experimental studies. In Study 1 and Study 2 (Section 5.1 and
5.2), which tested children on their interpretation of plural and singular indefmites, more
Chilean children associated the plural variant [5] to an interpretation of ‘more than one’
than [h] to this interpretation. However, in Study 3 (Section 5.3), which also tested
children on their comprehension of plural and singular indefmites, the opposite pattern
was found, where more Chilean children associated [h] to an interpretation of ‘more than
one’ than [s]. In addition, in Study 6 (Section 5.6), which tested plural deﬁnite noun
phrases in structures involving inalienable possession, we ﬁnd that only ChMC children,
but not ChWC children, associated the plural variant [h] to an interpretation of ‘more
than one’ more often than the plural variant [s]. For this reason, it is unclear whether one
of the two variants is initially acquired as the plural morpheme and how Chilean children

initially acquire lenition. Future research will address these questions.

6.3 Conclusions
The ﬁndings reported in this dissertation lend support to the Variability Delay
Hypothesis. Variability in the input that involves omission of the plural morpheme causes

a delay in the comprehension of plural morphology by many Chilean children. While it

271

might have seemed reasonable to predict that as long as Chilean adult speakers produced
the plural morpheme on semantically plural nouns some of the time, that would be
enough for Chilean children to acquire the plural morpheme and associate it to an
interpretation of ‘more than one’. However, this is not the case. Instead, it appears that
the ambiguous nature of the input that Chilean children are exposed to causes a delay in
their comprehension of the plural morpheme.

The ﬁndings reported in this dissertation also lend support to the theoretical
proposal that languages change due to a mismatch between the adult I-language (the
input to children) and the adult E—language (Kroch 1994, 2001, Yang 2000, 2002). This
dissertation assumes that child language acquisition involves the interaction between an
innate component that allows all humans to acquire language (e. g. the LAD) and the
linguistic input (or adult speech) that children are exposed to. Yang (2000, 2002)
proposes that languages change when the input that adults provide children (adult E-
language) does not match the adult grammar (I-language) and as a result children acquire
a grammar (I-language) that differs slightly from their parents. Over time, this may result
in language change as children grow into adults and provide a slightly different input to
their own children than what was provided to them

The experimental data presented in this dissertation show that while Chilean
adults always associate the plural morpheme to an interpretation of ‘more than one’ in
comprehension tasks, they do not always produce the plural morpheme on semantically
plural nouns in their own speech. In this sense, we might conclude that the Chilean adult
I-language has a representation for plural morphology while the adult E-language does

not always unambiguously match that representation. Because Chilean children do not

272

have access to the adult I-language but must rely solely on the adult E-language for
constructing their own grammars, they may initially construct and grammar that does not

associate the plural morpheme to an interpretation of ‘more than one’.

6.4 Future Research

I have presented the view that variable input involving omission causes a delay in
the acquisition of grammatical morphemes that are affected by that variation and I have
used the acquisition of plural morphology in Chilean and Mexican Spanish to investigate
and provide support for this view. While the experimental studies presented in this
dissertation provide support for such a delay, there are a number of questions still
remaining. It is essential that we look at much younger Chilean and Mexican children in
order to examine more closely differences in production between the two groups of
children. For example, it appears that 2-year-old Madrileﬁo Spanish-speaking children,
who receive unambiguous input for plural morphology, initially place the plural
morpheme only on nouns and not on determiners. Would this be true for Chilean
Spanish-speaking children who are exposed to an input where the plural morpheme is
placed more often on the determiner and omitted more often on the noun? Moreover,
would Chilean children rely more on the determiner than on the noun when assigning
number to determiner phrases? Another important question has to do with verbal
morphology. It has been reported that English-speaking children have difﬁculty using
verbal morphology to assign number to subject determiner phrases. However, given that
verbal morphology is unambiguous in Chilean Spanish while nominal morphology is
ambiguous, would Chilean children rely on verbal morphology to assign number to

subjects DPs? Miller and Schmitt (in press) provides evidence that they do. Finally, other

273

properties of plural morphology must be investigated in Chilean and Mexican Spanish.
For example, does the omission of the plural morpheme in Chilean Spanish affect
children’s ability to use plural morphology to distinguish between count vs. mass nouns.
We plan to investigate such questions about Chilean Spanish in ﬁrture research.

It is also important to investigate how variable input involving omission might
affect other areas of the grammar in various languages. Johnson (2005) has looked at
child comprehension of the third person singular marking in African American
Vernacular English-speaking children and found that variable input involving omission
appears to delay the comprehension of the third person singular marker, which is
consistent with the Variability Delay Hypothesis. Another area of interest would be the
acquisition of auxiliaries in English-speaking children who are exposed to a variety of
English where adults often omit the auxiliary in their own speech (see Chapter 3). These
types of studies will ultimately tell us more about the validity of the Variability Delay
Hypothesis. Thus, building on this research in Chilean Spanish and Mexican Spanish and
investigating the acquisition of other areas of the grammar which are affected by variable
input involving omission, will ultimately tell us more about how different types of input
(variable vs. consistent) and the frequency of these different types of input affects

language acquisition.

274

APPENDICES

275

APPENDIX A
Production: Study 1. Repetition Task

All children were presented with the following sentences in the same order as shown in
Table 64. Sentences were accompanied by colored pictures representing each sentence.

Table 64. APPENDIX A: Study 1 Repetition Task.

Researcher: “Este juego se llama “Escucha y Repite gTu sabes como jugar? Yo te
voy a decir alga sobre el dibujo y tzi tienes que repetir lo que yo digo. gT e parece? ”
“This game is called “Listen and Repeat”. Do you know how to play? I am going to
say something about the pictures here and you have to repeat what I say. Do you want
to play?”

 

PR La nifia tiene un chaleco verde.
The girl has a green sweater.

PR La niﬁa tiene un chaleco azul.
The girl has a blue sweater.

1. Algunos bomberos estan comiendo manzanas.
Some.PL ﬁremen.PL are.3.PL eating apples.PL

2. Estoy comiendo chocolates. gQuieres alguno?
I am eating chocolates.PL Do you want one.SG

3. Un monito (chango)22 esta comiendo una ' pera.
A/One.SG monkey.SG is.3.SG eating a/one.SG pear.SG
4. Unas bolitas estan en la silla.

Some.PL marbles.FL (canicas) are.3.PL on the chair

5. Hay muchas galletas, gQuieres alguna?
There are many.PL cookies.PL Do you want one.SG?

6. Algunas muffecas estan en el piso.
Some.PL dolls.PL are.3.PL on the ﬂoor

7. Un bombero esta comiendo una manzana.
A/One.sg ﬁreman.SG is.3.SG eating a/one.SG apple.SG

8. Algunos monitos (Changos) estan comiendo ﬁ'utillas.
Some.PL monkeys.PL are.3.PL eating strawberries.PL

 

22 Words in parenthesis are those that were used with Mexican subjects due to lexical
differences between the two dialects.

276

Table 64. APPENDIX A: Study 1 Rggetition Task, cont’d.
9. Una bolita (canica) esta en la silla.
A/One.SG marble.SG is.3.SG on the chair

10. Unas munecas estan en la cama.
Some.PL dolls.PL are.3.PL on the bed

11. Un monito (chango) esta comiendo una frutilla.
A/One.SG monkey.SG is.3.SG eating a/one.SG strawberry.SG
12. Algunas bolitas (canicas) estan en la mesa.
Some.PL marbles.FL are.3.PL on the table
13. Un bombero esta comiendo una papa.

A/One.SG ﬁreman.SG is.3.SG eating a/one.SG potato.SG

12. Unos monitos (Changos) estan comiendo peras.
Some.PL monkeys.PL are.3.PL eating pears.PL

13. Una muﬁeca esta en la cama.
A/One.SG doll.SG is.3.SG on the bed

14. Unos bomberos estan comiendo papas.
Some.PL ﬁremen.PL are.3.PL eating potatoes.PL

277

APPENDIX B
Study 2. Picture Matching Task: Indeﬁnite Noun Phrases
All children were presented with pictures in the following order. The experimental
question that accompanied each pair of pictures is shown in the box. Placement of

pictures was controlled for. All pictures were in cOlor.

Figure 19. APPENDIX B: Study 2 Barcos ‘ships’.

 

 

gEn cuél de las dos tarjetas hay unos barcos?
In which of the two cards EXST. some.M.PL ship.M.PL
§ ‘In which of the two cards is/are there some ships?’

 

 

 

 

 

 

 

 

Figure 20. APPENDIX B: Study 2 Monedas ‘coins’.

 

 

gEn cuél de las dos tarjetas hay una moneda?
In which of the two cards EXST. a/one.F.SG coin.F.SG
@ ‘In which of the two cards is/are there one coin?’

 

 

 

 

 

 

 

278

Figure 21. APPENDIX B: Study 2 Botellas ‘bottles’.

 

 

gEn cual de las dos tarjetas hay unas botellas?
In which of the two cards EXST. some.F.PL bottle.F.PL
‘In which of the two cards is/are there some bottles?’

 

 

 

 

 

 

 

Figure 22. APPENDIX B: Study 2 Monos (Changos for Mexico City) ‘monkeys’.

 

 

gEn cual de las dos tarjetas hay un mong?
In which of the two cards EXST. a/one.M.SG monkey.M.SG
‘In which of the two cards is/are there alone monkey?’

 

 

 

 

 

 

 

 

 

279

Figure 23. APPENDIX B: Study 2 Burros ‘donkeys’.

 

 

 

’1;

 

 

 

 

 

 

gEn cual de las dos tarjetas hay un burro?
In which of the two cards EXST. a/one.M.SG donkey.M.SG?
‘In which of the two cards is/are there a/one donkey?’

 

Figure 24. APPENDIX B: Study 2 Martillos ‘hammers’.

 

 

 

 

 

 

 

 

gEn cual de las dos tarjetas hay unos martillos?
In which of the two cards EXST. some.M.PL hammer.M.PL
‘In which of the two cards is/are there some hammers?’

 

280

 

 

Figure 25. APPENDIX B: Study 2 Bolitas (Canicas in Mexico City) ‘marbles’.

 

 

 

gEn cual de las dos tarjetas hay una bolita?
In which of the two cards EXST. a/one.F.SG marble.F.SG
‘In which of the two cards is/are there a/one marble?’

 

0

 

 

 

 

Figure 26. APPENDIX B: Study 2 Manzanas ‘apples’.

 

 

 

gEn cual de las dos tarjetas hay unas manzanas?
In which of the two cards EXST. some.l-‘.PL apple.l-‘.PL
‘ ‘In which of the two cards is/are there some apples?’

 

 

:2:

 

 

 

 

281

 

 

APPENDIX C
Study 3. Picture Matching Task: Indeﬁnite Noun Phrases
All children were presented with pictures in the following order. The experimental

question that accompanied each pair of pictures is shown in the box. All pictures were in
color.

Fl 6 27. APPENDIX C: Stud 3 Tar et Sto : Introduction to Characters.

   

 

Estos niﬁos son muy buenos
amigos. Un dia decidieron salir
de paseo. gA ver, ga do'nde
fueron? These boys are very good
ﬁiends. One day they decided to
go for a walk. Let’s see, where
did they go?

 

 

 

282

 

Figure 28. APPENDIX C: Study 3 Target Story: Narrative 1.

 

Primero Ios niﬁosﬁteron a mirar a
los animals que habia en el campo.
Habia y . Asi es que
las niiiosjugaron con las animales.
A ver, veamos que tiene cada niﬁo.
First the boys went to see the
animals that were in the
countryside. There were y

. So the boys played with the
animals. Let’s see what each boy
has.

 

 

 

 

 

 

 

 

 

gCudl niﬁo tiene unos manos?
Which boy has some.PL monkey.PL
‘Which boy has some monkeys?’

gCudI niﬁo no tiene nada?
Which boy NEG has nothing
‘Which boy doesn’t have anything?

gCudI niiio tiene unos burros?
Which boy has some.PL donkey.PL
‘Which boy has some donkeys?’

 

283

 

 

Figure 29. APPENDIX C: Study 3 Target Story: Narrative 2.

 

Despuésﬁteron a conocer una
tienda dejuguetes. A ver, en esa
tienda habia y . Los
nifzos se compraron varias cosas. A
ver, veamos que tiene cada niﬁo.
Next, they went to a toy store. Let’s
see, in this store there were

and . The children bought
several things. Let’s se what each
boy has.

 

 

 

 

 

 

 

 

 

gCudl niﬁo tiene una bolita?
Which boy has a/one.SG marble.SG
‘Which boy has a/one marble?’

gCual niﬁo no tiene nada?
Which boy NEG has nothing
‘Which boy doesn’t have anything?

gCudI niﬁo tiene una manzana?
Which boy has an/one.SG apple.SG
‘Which boy has an/one apple?’

 

284

 

 

Figure 30. APPENDIX C: Study 3 Target Story: Narrative 3.

 

Después dejugar se quedaron muy
cansados asi es queﬁreron a la casa
de la abuela. La abuela tem'a
muchas cosas interesantes en su
casa. Habia y . La
abuela siempre [es regalaba cosas a
los niﬁos. A ver, veamos que tiene
cada niﬁo. After playing the boys
were very tired so they went to their
grandma’s house. The grandma had
lots of interesting things in her
house. There were and

Let’s see what each boy has.

 

 

 

 

 

 

 

gCudl nifto tiene unas monedas?
Which boy has some.PL coins.PL
‘Which boy has some coins?’

gCudl niiio no tiene nada?
Which boy NEG has nothing
‘Which boy doesn’t have anything?’

gCudl niﬁo tiene unas botellas?
Which boy has some.PL bottle.PL
‘Which boy has some bottles?’

 

 

 

 

Figure 31. APPENDIX C: Study 3 Target Story: Narrative 4.

 

 

 

Después seﬁteron a la tienda. Alli en
la tienda habia varias coasas. Habia
. Los niﬁos compraron
varias cosas. A ver, veamos que tiene
cada niﬁo. Afterwards, the boys went
to the store. There, in the store there
were lots of things. There were

and . The boys bought
several things. Let’s see what each
boy has.

 

 

 

 

 

 

 

g'Cudl niﬁo tiene un martillo?
Which boy has a/one.SG harnmer.SG
‘Which boy has a/one hammer?’

gCudl niﬁo no tiene nada?
Which boy NEG has nothing
‘Which boy doesn’t have anything?’

gCual nifro tiene un barco?
Which boy has a/one.SG ship.SG
‘Which boy has a/one ship?’

 

286

 

 

 

Figure 32. APPENDIX C: Study 3 Control Story: Introduction to Characters.

 

 

 

 

Estas niﬁas son muy buenas
amigas. Un dia decidieron salir
de paseo. gA ver, ga do'nde
fueron? These girls are very good
friends. One day they decided to
go for a walk. Let’s see, where
did they go?

 

287

 

Figure 33. APPENDIX C: Study 3 Control Story: Narrative 1.

 

Primero, ﬁleron a conocer un
castillo enorme porque en ese
castillo habia cosas muy
interesantes. A ver, habia y

. Al salir del castillo las ninas
se Ilevaron algunas cosas. A ver,
veamos que tiene cada niﬁa. First,
they went to visit a big castle
because in this castle there were
many interesting things. Let’s see,
there were and . Upon
leaving the castle the girls took some
things. Let’s see what each girl has.

 

 

 

 

 

 

 

 

gCudl niﬁa tiene un solo libro?
Which girl has one only book
‘Which girl has only one book?’

gCua'l niﬁa no tiene nada?
Which girl NEG has nothing
‘Which girl doesn’t have anything?’

gCual niﬁa tiene una sola llave?
Which girl has one only key?
‘Which girl has only one key?’

 

288

 

 

Figure 34. APPENDIX C: Study 3 Control Story: Narrative 2.

 

Después seﬁteron a la playa. Para
ir a la playa es bueno traer algunas
cosas, coma par ejempla y

. Las niftas trajeran muchas
cosas a la playa. A ver, veamos que
tiene cada nifta. Next, they went to
the beach. When you go to the beach
it is a good idea to bring some things
along, like and . Let’s

see what each girl has.

 

 

 

 

 

 

 

 

gCudl niria tiene muchas pelatas?
Which girl has many balls
‘Which girl has many balls?’

gCudI niﬁa no tiene nada?
Which girl NEG has nothing
‘Which girl doesn’t have anything?’

gCudl niﬁa tiene muchas toallas?
Which girl has many towels?
‘Which girl has many towels?’

 

289

 

 

Figure 35. APPENDIX C: Study 3 Control Story: Narrative 3.

 

 

 

 

Después de ir a la playa, estuvieron
muy cansadas asi es queﬁteron a la
casa para descansar. Al llegar a la
casa, justo llego' el cartera y traia

y para las niﬁas. A
ver, veamos que tiene cada niﬂa.
After going to the beach, the girls
were very tired so they went home
to rest. When they got home, the
postman came and brought y

for the girls. Let’s see what
each girl has.

 

 

 

 

 

gCudl niiia tiene una sala moneda?
Which girl has one only coin
‘Which girl has only one coin?’

gCudl nir’ia no tiene nada?
Which girl NEG has nothing
‘Which girl doesn’t have anything?’

gCudI niﬁa tiene una sala carta?
Which girl has one only letter?
‘Which girl has only one letter?’

 

290

 

 

Figure 36. APPENDIX C: Study 3 Control Story: Narrative 4.

 

 

Alﬁnal, las niﬁasﬁteron a mirar a
las animals que habia en una tienda
de mascotas. Habia y
Asi es que las niﬁas compraron
algunas mascotas. A ver, veamos
que tiene cada niﬁa. Finally, the
girls went to look at the animals in
the pet shop. There were
.Let’ 5 see what each girl has.

 

 

 

 

 

 

 

 

gCuél niﬁa tiene muchos gatos?
Which girl has many cats
‘Which girl has many cats?’

gCudl niiia no tiene nada?
Which girl NEG has nothing
‘Which girl doesn’t have anything?’

gCudI niﬁa tiene muchos perros?
Which girl has many dogs?
‘Which girl has many dogs?’

 

291

 

 

APPENDIX D
Study 4. Elicitation Task: Bare Nouns and lndefmite Noun Phrases

All children were presented with pictures in the following order. The experimental
question that accompanied each picture is shown in the box. All pictures were in color.

Figure 37. APPENDIX D: Study 4 Elicitation Task.

 

Researcher: Mira. gQué es? gY, de que'
color es? gEs azul a raja? Look. What’s
this? And, what color is it (red circle)? Is
it blue or red?

 

 

Researcher: Mira. gQué es? gY, de que’
color es? gEs azul a rajo? Look What’s
this? And, what color is it (blue circle)? Is
it blue or red?

 

 

 

Researcher: Mira. Aqut’ hay dos circulas.
gCudl circula es mds grande? gCua'l
circula es mas chico? Look. Here are two
circles. Which circle is bigger? Which
circle is smaller?

 

 

 

 

292

 

Figure 37. APPENDIX D: Study 4 Elicitation Task, cont’d.

 

Researcher: gQué hay aqui? What’s here?

 

Researcher: gQué hay aqui? What’s here?

 

Researcher: gQué hay aqui? What’s here?

 

 

 

Researcher: gQué hay aqui? What’s here?

“(A '1‘ ”Q

41"

 

 

293

 

 

_F_igure 37. APPENDIX D: Study 4 Elicitation Task, cont’d

 

Researcher: gQué hay aqui? What’s here?

 

Researcher: gQué hay aqui? What’s here?

 

Researcher: gQué hay aqui? What’s here?

 

Researcher: gQué hay aqui? What’s here?

 

 

 

 

 

294

F igure 37. APPENDIX D: Study 4 Elicitation Task, cont’d

 

Researcher: gQué hay aqui? What’s here?

 

Researcher: gQué hay aqui? What’s here?

 

Researcher: gQué hay aqui? What’s here?

 

 

 

Researcher: gQué hay aqui? What’s here?

 

 

295

 

Figure 37. APPENDIX D: Study 4 Elicitation Task, cont’d.

 

Researcher: gQué hay aqui? What’s here?

 

Researcher: gQué hay aqui? What’s here?

 

 

Researcher: gQué hay aqui? What’s here?

 

 

 

Researcher: gQué hay aqui? What’s here?

 

 

296

 

APPENDIX E
Study 5. Picture Matching Task: Bare Noun Phrases
All children were presented with pictures in the following order. The experimental

question that accompanied each pair of pictures is shown in the box. Placement of
pictures was controlled for. All pictures were in color.

 

Figure 38. APPENDIX E: Study 5 Target Story: Introduction to Characters.

Estas niﬁas son muy buenas
amigas. Un dia decidieron salir
de paseo. 4A ver, ga dande
ﬁreron? These girls are very good
friends. One day they decided to
go for a walk. Let’s see, where
did they go?

 

 

 

 

 

 

297

Figure 39. APPENDIX E: Study 5 Target Story: Narrative l.

 

Primero, ﬁteran a conocer un
castillo enorme parque en ese
castillo habia cosas muy
interesantes. A ver, habia y
. AI salir del castillo las niﬁas
se Ilevaron algunas cosas. A ver,
veamos que tiene cada niiia. First,
they went to visit a big castle
because in this castle there were
many interesting things. Let’s see,
there were and . Upon
leaving the castle the girls took
some things. Let’s see what each

 

 

 

 

 

 

 

 

gCual niﬁa tiene libras?
Which girl has books
‘Which girl has books?’

gCudl nir’ta no tiene nada?
Which girl NEG has nothing
‘Which girl doesn’t have anything?’

gCua’l niﬁa tiene llaves?
Which girl has keys?
‘Which girl has keys?’

 

298

 

Figure 40. APPENDIX E: Study 5 Target Story: Narrative 2.

 

 

Después se ﬁteron a la playa. Para
ir a la playa es bueno traer algunas
cosas, coma par ejempla y

. Las niﬁas trajeron muchas
cosas a la playa. A ver, veamos que
tiene cada niﬁa. Next, they went to
the beach. When you go to the
beach it is a good idea to bring some
things along, like and
Let’s see what each girl has.

 

 

 

 

 

 

 

 

gCudl niﬁa tiene pelota?
Which girl has ball
‘Which girl has (a) ball?’

gCudl nina no tiene nada?
Which girl NEG has nothing
‘Which girl doesn’t have anything?’

gCudI nir'ia tiene taalla?
Which girl has towel?
‘Which girl has (a) towel?’

 

299

 

Figure 41. APPENDIX E: Study 5 Target Story: Narrative 3.

 

 

Después de ir a la playa, estuvieron
muy cansadas asi es quefueron a la

_casa para descansar. Al llegar a la

casa, justa llego' el cartera y traia
y para las niﬁas. A
ver, veamos que tiene cada niﬁa.
After going to the beach, the girls
were very tired so they went home
to rest. When they got home, the
postman came and brought
for the girls. Let’s see what
each girl has.

 

 

y

 

 

 

 

 

 

 

gCudI niﬁa tiene monedas?
Which girl has coins
‘Which girl has coins?’

gCuril niﬁa no tiene nada?
Which girl NEG has nothing
‘Which girl doesn’t have anything?’

gCudl niﬁa tiene cartas?
Which girl has letters?
‘Which girl has letters?’

 

300

 

 

Figure 42. APPENDIX E: Study 5 Target Story: Narrative 4.

 

 

Alﬁnal, las nir'iasfueron a mirar a
las animals que habia en una tienda
de mascotas. Habia y

 

[Asi es que las niﬁas campraron

algunas mascotas. A ver, veamos
que tiene cada niﬁa. Finally, the
girls went to look at the animals in
the pet shop. There were y

. Let’s see what each girl has.

 

 

 

 

 

gCudl niﬁa tiene gato?
Which girl has cat
‘Which girl has (a) cat?’

gCudl niﬁa no tiene nada?
Which girl NEG has nothing
‘Which girl doesn’t have anything?’

gCudI niﬁa tiene perra?
Which girl has dog?
‘Which girl has (a) dog?’

 

 

301

 

 

 

Figure 43. APPENDIX E: Study 5 Control Story: Introduction to Characters.

 

 

 

 

 

Estos niﬁos son muy buenos
amigos. Un dia decidieron salir
de paseo. gA ver, ga dénde
ﬁteron? These boys are very
good friends. One day they
decided to go for a walk. Let’s
see, where did they go?

 

302

 

Figure 44. APPENDIX E: Study 5 Control Story: Narrative l.

 

 

Primero las niﬁasﬁzeron a mirar a
las animals que habia en el campo.
Habia y . Asi es que
las niﬁasjugaran con los animales.
A ver, veamos que tiene cada niiio.
First the boys went to see the
animals that were in the
countryside. There were y

. So the boys played with the
animals. Let’s see what each boy
has.

 

 

 

 

 

 

 

 

gCudI niﬁo tiene muchos manos?
Which boy has many monkeys
‘Which boy has many monkeys?’

gCudI nino no tiene nada?
Which boy NEG has nothing
‘Which boy doesn’t have anything?

gCuél niﬁo tiene muchos burros?
Which boy has many donkeys
‘Which boy has many donkeys?’

 

303

 

 

Figure 45. APPENDIX E: Study 5 Control Story: Narrative 2.

 

Despuésfueran a conocer una
tienda dejuguetes. A ver, en esa
tienda habla y . Las
niiios se compraron varias cosas. A
ver, veamos que tiene cada niﬁo.
Next, they went to a toy store. Let’s
see, in this store there were

and . The children bought
several things. Let’s se what each
boy has.

 

 

 

 

 

 

 

 

gCudl niﬁo tiene una sola bolita?
Which boy has only one marble
‘Which boy has only one marble?’

gCudl niﬁo no tiene nada?
Which boy NEG has nothing
‘Which boy doesn’t have
anything?’

gCudl niﬁa tiene una sala
manzana?

Which boy has only one apple
‘Which boy has only one apple?’

 

304

 

Figure 46. APPENDIX E: Study 5 Control Story: Narrative 3.

 

 

Después dejugar se quedaron muy
cansadas asi es quefueron a la casa
de la abuela. La abuela tenr’a
muchas cosas interesantes en su
casa. Habia y . La
abuela siempre les regalaba cosas a
las ninos. A ver, veamos que tiene
cada niﬁa. After playing the boys
were very tired so they went to their
grandma’s house. The grandma had
lots of interesting things in her
house. There were and

. Let’s see what each boy has.

 

 

 

 

 

 

 

 

 

gCudl niﬁa tiene muchas monedas?
Which boy has many coins
‘Which boy has many coins?’

gCudl niﬁo no tiene nada?
Which boy NEG has nothing
‘Which boy doesn’t have anything?’

gCudI niﬁo tiene muchas botellas?
Which boy has many battles
‘Which boy has many bottles?’

 

305

 

Figure 47. APPENDIX E: Study 5 Target Story: Narrative 4.

 

Después seﬁleron a la tienda. Alli
en la tienda habia varias coasas.
Habia y . Las niflas

. compraron var ias COSOS. A ver,

veamos que tiene cada niiio.
Afterwards, the boys went to the
store. There, in the store there were
lots of things. There were and
. The boys bought several
things. Let’s see what each boy has.

 

 

 

 

 

 

 

gCuél niﬁo tiene un solo martiIIa?
Which boy has only one hammer
‘Which boy has only one hammer?’

gCudI niﬁo no tiene nada?
Which boy NEG has nothing
‘Which boy doesn’t have anything?’

gCudI nir'ro tiene un solo barca?
Which boy has only one ship
‘Which boy has only one ship?’

 

306

 

APPENDIX F
Study 6. Act-out Task: Deﬁnites in Inalienable Possession

All children were presented with sentences in the following order.

Table 65. APPENDIX F: Study 6 Act-out Task. '

 

Researcher: Mira. Este es Carlita (Show female doll). Yo te voy a decir lo que tienes que
hacer con Carlita y tu tienes que hacer/o. gEntiendes? Look. This is Carlita. I am going
to tell you what you have to do with Carlita and you have to do it. Understand?

l Levantale la piema
Lift.her the.F.SGleg.F.SG
‘Lift her leg’

Control Dame la vaca
Give.me the.F.SG cow.F.SG
‘Give me the cow’

2 Tirale la oreja
Pull.her the.F.SG ear.F.SG
‘Pull her ear’

3 Tocale la rodilla
Touch.her the.F.SG knee.F.SG
‘Touch her knee’

Control Dame la arar’la

Give.me the.F.SG spider.F.SG
‘Give me the spider’

4 Levantale la mano
Liﬁ.her the.F.SG handrso
‘Lift her hand’

 

307

Table 65. APPENDIX F: Study 6 Act-out Task, cont’d.

Researcher: Mira. Este es Pedro (Show male doll). Ya te voy a decir la que tienes que
hacer con Pedro y tu tienes que hacerla. gEntiendes? Look. This is Pedro. I am going to
tell you what you have to do with Pedro and you have to do it Understand?

5. Levantale las piemas
Lift.him the.F.PLlegs.F.PL
‘Lift his legs’

Control Dame las vacas
Give.me the.F.PL COWS.F.PL
‘Give me the cows’

6. Tirale las orejas
Pull.him the.F.PL earS.F.PL
‘Pull his ears’

7. T6cale las rodillas
Touch.him the.F.PL knees.F.PL
‘Touch his knees’

Control Dame las araflas
Give.me the.F.PL spiders.F.PL
‘Give me the spiders’

8. Levantale las manos
Lifthim the.F.PL hands.F.PL
‘Lift his hands’

 

308

APPENDIX G
Study 7. Act-out Task: Deﬁnite Noun Phrases

All children were presented with sentences in the following order.

Table 66. APPENDIX G: Study 7 Act-out Task.

 

l Dame todas las vacas durmiendo a1 lado de la casa. Control-all
Give.me all.PL the.PL cows.PL sleeping next to the house
‘Give me all of the cows sleeping next to the house.’

2 Dame las vacas durmiendo al lado de la granja. Target-DefPl
Give.me the.PL cows.PL sleeping next to the barn
‘Give me the cows sleeping next to the barn.’

Return cows to original positions.

3 gCrees que las vacas comen pasto? F iller-Distractor
Think.2.SG that the.PL cows.PL eat.3.PL grass?
‘Do you think that cows eat grass?’

4 (“Son grandes o pequeﬁas las vacas? Filler-Distractor
Are.3.PL big.PL or small.PL the.PL cows.PL
‘Are cows big or small?’

5 Dame unas vacas durmiendo al lado de la caSa. Target-IndefPl
Give.me some.PL cows.PL sleeping next to the house
‘Give me some cows sleeping next to the house.’

6 Dame la vaca durmiendo al lado de la granja. Target-Def.S g
Give.me the.SG cow.SG sleeping next to the barn
‘Give me the cow sleeping next to the barn.’
Return cows to original positions.
7 Dame una vaca durmiendo al lado de la granja. Control-one
Give.me a/one.SG cow.SG sleeping next to the barn
‘Give me a/one cow sleeping next to the barn.’
Take away cows and replace with spiders.
8 Dame una aral’la durmiendo a1 lado de la granja. Control—one

Give.me a/one.SG spider.SG sleeping next to the barn
‘Give me a/one spider sleeping next to the barn.’

309

Table 66. APPENDIX G: Study 7 Act-out Task, cont’d.

9 Dame todas las araﬁas durmiendo a1 lado de la casa. Control-all
Give.me all.PL the.PL spiders.PL sleeping next to the house
‘Give me all of the spiders sleeping next to the house.’

Return spiders to original positions.
10 (Has visto alguna vez una tela de araﬁas? Filler-Distractor

Have.2.SG seen some time a web of spiders.PL
‘Have you seen a spider web before?’

11 LQué comen las arar’fas? Filler-Distractor
What eat.3.PL the.PL spiders.PL
‘What do spiders eat?’

12 Dame unas araflas durmiendo a1 lado de la granja. Target-IndefPl

Give.me some.PL spiders.PL sleeping next to the barn
‘Give me some spiders sleeping next to the barn.’

13 Dame la araﬁa durmiendo al lado de la casa. Target-Def.Sg
Give.me the.SG spider.SG sleeping next to the house
‘Give me the spider sleeping next to the house.’

Return spiders to original positions.
14 Dame las araﬁas durmiendo al lado de la granja. Target-Def.Pl
Give.me the.PL spiders.PL sleeping next to the barn
‘Give me the spiders sleeping next to the barn.’
Take away spiders and replace with dolls.
15 Dame las munecas durmiendo al lado de la casa Target-DefPl
Give.me the.PL dolls.PL sleeping next to the house
‘Give me the dolls sleeping next to the house.’
16 Dame la muﬁeca durmiendo a1 lado de la granja Target-Def.Sg
Give.me the.SG doll.SG sleeping next to the barn

‘Give me the doll sleeping next to the barn.’

Return dolls to original positions.

 

310

Table 66. APPENDIX G: Study 7 Act-out Task, cont’d.

 

l7 gTe gustan estas muﬁecas?
You please.3.PL these.PL dolls.PL
‘Do you like these dolls?’

18 (De que color es su ropa?
Of what color is.3.SG their clothes
‘What color are their clothes?’

19 Dame unas munecas durmiendo a1 lado de la granja.
Give.me some.PL dolls.PL sleeping next to the barn
‘Give me some dolls sleeping next to the barn.’

20 Dame una muﬁeca durmiendo al lado de la casa.
Give.me a/one.SG doll.SG sleeping next to the house
‘Give me a/one doll sleeping next to the house.’

Return dolls to original positions.

21 Dame todas las munecas durmiendo al lado de la casa.
Give.me all.PL the.PL dolls.PL sleeping next to the house
‘Give me all of the dolls sleeping next to the house.’

Take away dolls and replace with ants.

22 Dame la hormiga durmiendo al lado de la granja.
Give.me the.sg ant.sg sleeping next to the barn
‘Give me the ant sleeping next to the barn.’

23 Dame una hormiga durmiendo a1 lado de la casa.
Give.me a/one ant sleeping next to the house
‘Give me a/one ant sleeping next to the house.’

Return ants to original positions.

24 gSaben las hormigas hablar?
Know.3.PL the.PL ants.PL Speak.NONFIN
‘Do ants know how to speak?’

25 (Son grandes o pequeffas las hormigas?
Are.3.PL big.PL or sma11.PL the.PL ants.PL
‘Are ants big or small?’

Filler-Distractor

Filler-Distractor

Target-IndefPl

Control-one

Control-all

Target-Def.S g

Control-one

Filler-Distractor

Filler-Distractor

 

311

Table 66. APPENDIX G: Study 7 Act-out Task, cont’d.

26 Dame unas hormigas durmiendo a1 lado de la casa. Target-Indef.Pl
Give.me some.pl ants.pl sleeping next to the house
‘Give me some ants sleeping next to the house.’

27 Dame todas las hormigas durmiendo a1 lado de la granja. Control-all
Give.me all.PL the.PL ants.PL sleeping next to the barn
‘Give me all of the ants sleeping next to the barn.’
Return ants to original positions.
28 Dame las hormigas durmiendo a1 lado de la casa. Target-Def.Pl

Give.me the.PL ants.PL sleeping next to the house
‘Give me the ants sleeping next to the house.’

312

 

BIBLIOGRAPHY

313

BIBLIOGRAPHY

Alonso A., and Lida, R. (1940). El espaﬂol de Chile. Buenos Aires: Biblioteca de
Dialectologia de Buenos Aires. ‘

Bamer, D., and Snedeker, J. (2005). Quantity judgments and individuation: evidence that
mass nouns count. Cognition 97: 41-66.

Barrutia R., and Schwegler, A. (1994). F onética y F onologia Espaﬁolas: Teoria y
Practica. New York: Wiley and Sons.

Bedore L., and Leonard, L. (2001). Grammatical Morphology Deﬁcits in Spanish-
speaking Children with Speciﬁc Language Impairment. Journal of Speech, Hearing
and Language Research 44: 905-924.

Berko, J. (1958). The child’s learning of English morphology. Word 14:150-177.

Berman, R. (1981). Language development and language knowledge: evidence from the
acquisition of Hebrew morphophonology. Journal of Child Language 8: 609-626.

Bosque, I. (1996). El Sustantivo Sin Determinacién. La Ausencia de Determinacién en la
Lengua Espaftola. Madrid: Visor Libros.

Bradley, T. (2006). Spanish Rhotics and Dominican Hypercorrect /s/. Probus 18, (l): 1-
33. '

Brown, R. (1973). A ﬁrst language: the early stages. Cambridge, Mass: Harvard
University Press.

Canﬁeld, D. (1982). Spanish Pronunciation in the Americas. The Modern Language
Journal 66, (3): 350-351.

Cantr’r-Sénchez, M., and Grinstead, J. (2004). Nominal number and gender agreement in
child Spanish SLI. Poster presented at Generative Approaches to Language
Acquisition in North America (GALANA), December 19, University of Hawai’i at
Manoa.

Carlson, G. (1977). Reference to Kinds in English. University of Massachusetts, Amherst
Dissertation, [Published 1980 New York: Garland Press.].

Cazden, C. (1968). The Acquisition of Noun and Verb Inﬂections. Child Development
39: 433-448.

314

 

 

Cepeda, G. (1995). Retention and Deletion of Word-ﬁnal /s/ in Valdivian Spanish
(CHILE). Hispanic Linguistics 6/7: 329-353.

Chomsky, N. (1986). Knowledge of Language. It ’s Nature, Origin and Use. Praeger
Press.

---- (1995). The Minimalist Program. The Massachusetts Institute of Technology Press.

Chierchia, G. (1998). Reference to Kinds Across Languages. Natural Language
Semantics 6: 339-405.

de Villiers, J ., and de Villiers, P. (1973). A cross sectional study of the acquisition of

grammatical morphemes in child speech. Journal of Psycholinguistic Research 2: 267-
278.

de Villiers, P., and de Villiers, J. (2004). Assessing Pragmatics: Who, What and How?
Paper presented at the American Speech-Language-Hearing Association (ASHA)
Convention.

Diaz Campos, M. (2005). The emergence of adult-like command of sociolinguistic
variables: A study of consonant weakening in Spanish-speaking children. In D.
Eddington (ed.), Studies in the Acquisition of the Hispanic Languages: Papers from
the 6th Conference on the Acquisition of Spanish and Portuguese as First and Second
Languages. Somerville, MA: Cascadilla Press: 56-65.

Haeussler, I., and Marchant, T. (1988). Test de Desarrollo Psicamotor 2-5 aﬁas TEPSI.
Santiago, Chile: Universidad Catolica de Chile Press. '

Hudson Kam, C., and Newport, E. (2005). Regularizing unpredictable variation: The
roles of adult and child learners in language formation and change. Language Learning
and Development 1: 151-195.

Fasold, R., and Wolfram, W. (1970). Some linguistic features of Negro dialect. In R.
F asold and R. Shuy (eds.), Teaching Standard English in the Inner City. Washington,
DC: Center for Applied Linguistics: 41-86.

Ferenz, K., and Prasada, S. (2002). Singular or plural? Children’s knowledge of the
factors that determine the appropriate form of count nouns. Journal of Child
Language 29: 49-70.

Gathercole, V. (1986). The acquisition of the present perfect: explaining differences in
the speech of Scottish and American children. Journal of Child Language 13: 537-
560.

Gouvea, A., Aldana, G., Bell, T., Cody, K., de Groat, C., Johnson, C., McCabe, D.,
Zimmerman, L., and Kim, J. (2005). 18 Month Old Infants’ Sensitivity to Number

315

 

 

 

Agreement Inside the Noun Phrase. In: A. Brugos. M. Clark-Cotton, and S. Ha (eds.),
BUCLD 29: Proceedings of the 29th annual Boston University Conference on
Language Development. Cascadilla Press: 240-247.

Grimshaw, J. (1991). Extended projection. Ms., Brandeis University, Waltharn, Mass.

Guy, G. (1980). Variation in the group and the individual: The case of ﬁnal stop deletion.

In W. Labov (ed.), Locating language in time and space. New York: Academic Press:
1-36.

Guasti M. and Rizzi, L. (2002). Agreement and tense as distinct syntactic positions:
Evidence from acquisition. In G. Cinque (ed.), Functional structure in DP and IP: The
cartography of syntactic structures 1, Oxford University Press.

Pavez, M. (2003). Test Explarataria de Grdmatica Espaiiala de A. T aranta: Aplicacidn
en Chile. Santiago, Chile: Universidad Catolica de Chile Press.

---- (2004). Test para la Comprensio’n Auditiva del Lenguaje de E. Carrow. Santiago,
Chile: Alfabeta Artes Graﬁcas.

Johnson, V. (2005). Comprehension of Third Person Singular /s/ in AAE-Speaking
Children. Journal of Language, Speech, and Hearing Services in Schools 36: 116-
124.

Johnson, V., dc Villiers, J ., and Seymour, H. (2005). Agreement without understanding?
The case of third person singular /s/. First Language 25: 317-330.

Hudson Kam, C., and Newport, E. (2005). Regularizing Unpredictable Variation: The
Roles of Adult and Child Learners in Language Formation and Change. Language
Learning and Development 1: 151-195.

Keman, K., and Blount, B. (1966). The Acquisition of Spanish Grammar by Mexican
Children. Anthropological Linguistics 8, (9): 1-14.

Kouider, S., Halberda, J ., Wood, J ., Carey, S. (2006). Acquisition of English Number
Marking: the Singular-Plural Distinction. Language Learning and Development 2, l: 1-
25.

Kopcke, K. M. (1998). The acquisition of plural marking in English and German
revisited: schemata versus rules. Journal of Child Language 25: 293-319.

Kovac, C. and Adamson, H.D. (1981). Variation Theory and First Language Acquisition.
In: D. Sankoff and H. Cedergren (eds.), Variation Omnibus: 403-410.

Kroch, A. (1994). Morphosyntactic Variation. In: K. Beals et al. (eds.), Proceedings of
the Thirtieth Annual Meeting of the Chicago Linguistic Society 2: 180 - 201.

316

 

 

 

---- (2001). Syntactic Change. In: M. Baltin and C. Collins (eds.), Handbook of
Contemporary Syntactic Theory. Blackwell Press.

Kvaal, J ., Shipstead-Cox, N., Nevitt, S.G., Hodson, B., Launer, P. (1988). The
Acquisition of 10 Spanish Morphemes by Spanish-Speaking Children. Language
Speech, and Hearing Services in Schools 19: 384-394.

Kupisch, T. (2006). The acquisition of determiners in bilingual German-Italian and
German-French children. LINCOM Studies in Language Acquisition 17.

Labov, W. (1989). The child as linguistic historian. Language Variation and Change.
1(1): 85-98.

Lenz, R (1940). El espaﬁol en Chile. Buenos Aires: Biblioteca de Dialectologia
Hispanoamericana, Instituto de Filologia de la Universidad de Buenos Aires.

Lillo-Martin, D., and Snyder W. (2002). Cross-linguistic study of early syntax.
Proceedings of the IV Congressa Intemacianal de Lingua e Literatura de Mercasul,
Universidade Luterano do Brasil: 1-10.

Lipski, J. (1985). /s/ in Central American Spanish. Hispania 68: 143-9.

---- (1986). Reduction of /s/ and /n/ in Spanish. Canadian Journal of Linguistics 31: 139-
56.

—--- (1999). The many faces of Spanish /s/-weakening: (re)alignment and ambisyllabicity.
In: J. Gutiérrez-Rexach and F. Martinez-Gil (eds.), Advances in Hispanic linguistics
Somerville, MA: Cascadilla Press: 198-213.

---- (1994). Latin American Spanish. London: Longmann.

Marrero, V., and Aguirre, C. (2003). Plural acquisition and development in Spanish. In:
S. Montrul and F. Ordoﬁez (eds.), Linguistic Theory and Language Development in
Hispanic Languages. Somerville, Cascadilla Press, 275-296.

Masullo, P. (1992). Incorporation and Case Theory in Spanish: A crosslinguistic
perspective. University of Washington Doctoral Dissertation.

Mervis, C., and Johnson, K. (1991). Acquisition of the plural morpheme: a case study.
Developmental Psychology 27 (2): 222-235.

Miller, K., and Schmitt, C. (2004). An Experimental Study on Child Comprehension of

Spanish Indeﬁnites and Bare Singulars. Language Acquisition: A Journal of
Developmental Linguistics 12 (3/4): 247-256.

317

 

Miller, K., and Schmitt, C. (in press). Variable vs. Consistent Input: Comprehension of
Plural Morphology and Verbal Agreement in Children.

Munn, A., Miller, K., Schmitt, C. (2006). Plurality and Maximality in Children's
Comprehension of Deﬁnite Noun Phrases. In: D. Bamman, T. Magnitskaia, and C.
Zaller (eds.), BUCLD 30: Proceedings of the 3 0’” annual Boston University Conference
on Language Development, Somerville, Mass: Cascadilla Press: 365-3 76.

Moore, M. (1979). Acquisition of plural inﬂection by black American children. Michigan
State University Doctoral Dissertation.

Moreno de Alba, J. (1994). La pranunciacion del espaﬁal en Mexico (Estudios de
Dialectologia Mexicana, V). Santa Teresa, Mexico: El Colegio de Mexico.

Morgan, T. (1998). The linguistic parameters of /s/ insertion in Dominican Spanish: A
case study in qualitative hypercorrection. Perspectives on Spanish Linguistics 3: 79-96.

Ojeda, A. (1998). The Semantics of Collectives and Distributives in Papago. Natural
Language Semantics 6: 245-270.

Oroz, R (1960). La lengua castellana en Chile. Santiago, Chile: Editorial Universitaria.

Paolillo, J. (2002). Analyzing Linguistic Variation: Statistical Models and Methods.
Standford, California: CSLI Publications.

Park, T. (1978). Plurals in Child Speech. Journal of Child Language 5: 237-250.

Pavez, M. (2003). Test Exploratorio de Gramatica Espaﬁola de A. T oranta: Aplicacio'n
en Chile. Santiago, Chile: Ediciones Universidad Catolica de Chile.

Pavez, M. (2004). Test para la camprensia'n auditiva del lenguaje de E. Carrow:
Aplicacio’n en Chile. Santiago, Chile: Ediciones Universidad Catolica de Chile.

Pérez-Leroux, A-T. (2005). Number Problems in Children, Proceedings of the 2005
Annual Conference of the Canadian Linguistics Association: 1-12.

Perez-Pereira, M. (1989). The Acquisition of Morphemes: Some Evidence ﬁom Spanish.
Journal of Psycholinguistic Research 18, (3): 289-312.

Ramer, A., and Rees, N. (1973). Selected Aspects of the Development of English
Morphology in Black American Children of Low Socioeconomic Background. Journal
of Speech and Hearing Research 16, (4): 569-577.

Rand, D., and Sankoff, D. (1990). Goldvarb: a variable rule applicatianfar the
Macintosh. Centre de recherches mathematiques, Universite de Montreal, Canada.

318

 

 

Roberts, J. (1994). Acquisition of Variable Rules: (-t,d) Deletion and (ing) Production in
Preschool Children. University of Pennsylvania Doctoral Dissertation.

Roberts, J. (1997). Acquisition of Variable Rules: A Study of (-t,d) Deletion in Preschool
Children. Journal of Child Language 24, (2): 351-172.

Smith J ., Durham, M., Fortune, L. (2006). Caregiver and Child in the Acquisition of
Sociolinguistic Norms in a Scottish Dialect. In D. Bamman, T. Magnitskaia, and C.
Zaller (eds.), BUCLD 30: Proceedings of the 3 0’” annual Boston University Conference
on Language Development, Somerville, Mass.: Cascadilla Press: 572-5 83.

Sauerland, U. (2003). A new semantics for number. The Proceedings of SALT 13. Cornell
University, Ithaca, N.Y.: CLO-Publications.

Munn, A., and Schmitt, C. (2003). Bare nouns and the morpho-syntax of number. In: T.
Satterﬁeld, C. Tortora and D.Cresti (eds.), Current Issues in Romance Languages:

Selected papers from the 29th Linguistic Symposium on Romance Languages (LSRL):
225-239.

Wolfram, W., and Schilling-Estes, N. (1998). American English: Dialects and Variation,
Oxford: Basil Blackwell.

Schwarzschild. R. (1996). Pluralities. Kluwer Academic Publishers.

Schutze. C. (2006). What it means (not) to know (number) agreement. Paper presented at
the GLOW Workshop: Acquisition of the Syntax and Semantics of Number Marking.

Seymour, H., and Pearson, B. (2004). Evaluating language variation: Distinguishing
development and dialect ﬁ'om disorder. Special issue of Seminars in Speech and
Language, New York: Thieme Medical Publishers.

Singleton, J., and Newport, E. (2004). When learners surpass their models: The
acquisition of American Sign Language ﬁom inconsistent input. Cognitive Psychology
49: 370-407.

Slobin, D. (1985). Crosslinguistic Evidence for the Language-Making Capacity. In D.
Slobin (ed.), The crosslinguistic study of language acquisition 2. Hillsdale, NJ:
Lawrence Erlbaum Associates: 1157-1256.

Terrell, T. ( 1981). Diachronic reconstruction by dialect comparison of variable
constraints: S-aspiration and deletion in Spanish. In: D. Sankoff and H. Cedergren
(eds.), Variation Omnibus. Edmonton: Linguistic Research: 115-124.

Tobler, L. (1883). Ueber den Begriff und besondere Bedeutungen des Plurals bei
Substantiven, Zeitschriﬁ fur Volkerspsychologie 14: 410-434.

319

 

 

Vasconcellos Lopes, R. (2006). Bare nouns and DP number agreement in the acquisition
of Brazilian Portuguese. In: N. Sagarra and A. Toribio (eds.), Selected Proceedings of
the 9th Hispanic Linguistics Symposium. Somerville, MA: Cascadilla Proceedings
Project: 252-262.

Vivas, D. (1979). Order of Acquisition of Spanish Grammatical Morphemes: Comparison
to English and Some Cross-linguistic Methodological Problems. Kansas Working Papers
in Linguistics 4: 77-106.

Wackemagel, J. (1920). Vorlesungen uber Syntax, Erste Reihe [ﬁrst series], Birkhauser,
Basel.

Washington, J ., and Craig, K. (1994). Dialectal forms during discourse of urban, African
American preschoolers living in poverty. Journal of Speech and Hearing Research 37:
816—823.

Washington, J ., Craig, H., and Kushmaul, J. (1998). Variable use of African American
English across two language sampling contexts. Journal of Speech, Language, and
Hearing Research 41: l 1 15—1124.

Wilson, J ., and Henry, A. (1998). Parameter setting within a socially realistic setting.
Language and Society 27, 1-21.

Wolfram, W. (1969). A Sociolinguistic Description of Detroit Negro Speech.
Washington, DC: Center for Applied Linguistics.

Yang, C. (2000) Knowledge and Learning in Natural Language. Massachusetts Institute
of Technology Doctoral Dissertation.

---- (2001). lntemal and external forces in language change. Language Variation and
Change 12: 231-250.

---- (2002). Knowledge and Variation in Natural Language. Oxford University Press,
Oxford.

320

‘1 LlBR'

S
_H
E
W
N
U
E
A
1
S

 

MICHIGAN

 

 

 

MU itijti‘ﬂi‘lbzms 91 .

l
r