Ux’fil‘li 9601...“:qu V..." Er ..-

 

ll»inl.|.

7H,...

r715

a.
at.

”m3:

9.
Wuhan“

,.. 5 ha!“

 

o 'l ,
in... a.
url ‘

I! C6
3%..

S:
5... . i!
:1 blah.

Islirlllllblllln .lll» :

 

u

.h

 

This is to certify that the
dissertation entitled

DEVELOPMENT OF TELICITY INTERPRETATON:
SENSITIVITY TO VERB—TYPE AND DETERMINER-TYPE

presented by

DIANE ALICE OGIELA

has been accepted towards fulfillment
of the requirements for the

degree In Audiolggg 8. Speech Sciences

 

 

%//% LC

Major Pro ssor’s Sigma’tfilre Jw

fléfl 7

Date

MSU is an affirmative-action, equal-opportunity employer

0--.--.—.-v--a--n---tOu-.----4--a-n-.-.—-------~-----o-n-o-n-nccuu-I-u-vOaQa-o-n-------.-----n-s---n----.

 

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 p:lC|RC/DateDue.indd-p.1

 

DEVELOPMENT OF TELICITY INTERPRETATION:
SENSITIVITY TO VERB-TYPE AND DETERMINER-TYPE

By

Diane Alice Ogiela

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Audiology and Speech Sciences

2007

ABSTRACT

DEVELOPMENT OF TELICITY INTERPRETATION:
SENSITIVITY TO VERB-TYPE AND DETERMINER-TYPE
By
Diane Alice Ogiela
Previous research (van Hout, 1998) has shown that children have difficulty
understanding whether a verb phrase (VP) is to be interpreted as telic (with a
logical end point) or atelic (with no logical end point). To determine a VP’s telicity,
both nominal and verbal features of the VP must be taken into account. This
dissertation examined the interpretation of VP telicity by expanding the linguistic
variables examined. Four verb types and 2 determiner types were used. The
verb types were (a) non-partitive, quantity-sensitive verbs (build-type); (b)
partitive, quantity-sensitive verbs (eat-type); (c) quantity-insensitive verbs (push-
type); and (d) the latter two with resultative particles (eat up-type and push over-
type). The determiner types in the object determiner phrase (DP) were (a) the
definite determiner the and (b) the cardinal number two. Study I examined how
adults interpret telicity in VPs with different verb types and determiner types.
Study Il examined how 3-, 4-, 5-, and 6-year-old children interpret VP telicity
under the same conditions. Both studies used a modified truth-value judgment
task.
Study 1 hypothesized that, for adults, verb-type and determiner-type
would interact to produce different response patterns to questions about non-

culminating events. We with build-type verbs with a cardinal number in the

object DP, were predicted to induce the most telic interpretations, followed by
build-type verbs with a definite determiner, eat-type verbs with a cardinal
number, and finally, eat-type verbs with a definite determiner. This ranking was
predicted by two linguistic factors: (a) the potential partitivity of eat-type verbs,
and (b) the possible variability due to discourse-based interpretation of the
definite determiner vs. a specific amount indicated by the cardinal number.
Another hypothesis was that resultative particles serve somewhat different
functions with push-type verbs than with eat-type verbs. Overall, the results of
Study I supported the predictions.

The hypotheses and predictions for Study II were similar, with two added
dimensions: (a) older children should demonstrate greater sensitivity to verb-type
and determiner-type with regard to telicity than younger children and (b) because
resultative particles are thought to be unambiguous indicators of telicity, children
should be most sensitive to resultative particles for interpreting We as telic.
There were group differences between the 3-year-olds and the 5- and 6-year-
olds. There were no significant differences for VPs with eat-type vs. build-type
verbs. There were more telic interpretations for VPs with the cardinal number
than with the definite determiner. Although the particle was found to contribute to
a telic interpretation when the object determiner was the definite determiner, its
contribution to a telic interpretation was no greater than the cardinal number’s
contribution without the particle. The results suggest that even by age 6, children
may not have a complete adult-like understanding of the definite determiner and

of the finer distinctions between verb-types as they contribute to VP telicity.

Copyright by
DIANE ALICE OGIELA

2007

DEDICATION

This dissertation is dedicated to my parents, Zdzislawa and WIncenty Ogiela;

to my husband, John Ader; and to my children, Logan and Marlena Ader.

ACKNOWLEDGMENTS

Completing my doctorate and this dissertation would not have been
possible without the help, support, and influence of many wonderful people who I
am fortunate to count among my friends and colleagues. I have many people to
thank, it is hard to know where to start. Perhaps the beginning is a good place.

I have a tremendous amount of appreciation for my undergraduate
mentor, Carl Stach. Although he had died long before I began my doctoral
studies, I would never have gotten to this point without his influence. Carl loved
language in all of its forms, from phrase structure trees to poetry. He showed me
that language is fascinating, cool and important, both in theory and in practice.
His excitement was contagious and it made me change my career plans. At the
time, I didn’t know where my new interest would take me, but it felt right. Carl
continues to influence my thinking and my pursuits. And I owe him a great many
thanks.

I am very grateful to Laurence Leonard, my master‘s thesis advisor at
Purdue University. He gave me the opportunity to be part of his lab and to try my
hand at research. Under his mentorship, I learned a great deal about research
and realized that I wanted to be a researcher. He encouraged me to pursue a
doctorate. Although I chose to first go into clinical practice (temporarily), his
encouragement helped me make a personal commitment to return to school and

complete a doctorate.

vi

I want to express my deep appreciation to my doctoral and dissertation
committees for their hard work, encouragement, patience and flexibility.
Fernanda Ferreira and Jeff Marler were on my doctoral program committee while
they were on the faculty at MSU. Fernanda gave me the chance to learn more
about language processing by inviting me to her psycholinguistics reading group
and Jeff exposed me to the world of electrophysiology. Both of them challenged
me to think about how we measure and study language. I greatly appreciate the
efforts and feedback that I received from my dissertation committee. I am grateful
to Ida Stockman for posing difficult, but thought-provoking questions about the
potential implications and applications of my research area down the road,
specifically for children with language impairments. Although Alan Munn officially
became part of my committee quite late, in reality he had been their all along,
from Syntax class to meetings in the Child Language Lab. He was always willing
to listen to a presentation, give an opinion, help work out a problem and, perhaps
most importantly, to look at the project with a critical eye. His tough questions
strengthened the project at many points along the way. I am also grateful to him
for technical computer support and for opening my eyes to the magical world of
“real” word processing.

The co-chairs of my committee, Mike Casby and Cristina Schmitt were
extremely generous with their knowledge, time, energy, feedback and resources.
They were both very supportive of me personally and academically. Despite both
of them being extremely busy, they made time to think, meet, talk, read and give

feedback. I have been very fortunate to have them as mentors. Throughout my

vii

time at MSU, Mike Casby eagerly helped me pursue various experiences and
interests. He wrote letters and edited papers and posters with lightening speed. I
really enjoyed our discussions and debates and appreciated that he respected
my ideas and decisions. His ideas and input about the methodology used in the
dissertation studies were extremely valuable. I never thought I’d say it, but thank
you very much for insisting on the first pilot study, and the second, and the third,
and... (you get the idea). They were truly necessary and helpful.

Shortly after I mme to MSU, Cristina Schmitt welcomed me into her
linguistics courses and Child Language Lab. Through our chats about
morphology (my original obsession) she introduced me to the concepts of aspect
and telicity and taught me most, if not all, of what I know about them. Cristina is
very knowledgeable and was a very patient teacher. Her knowledge and
guidance were instrumental developing the ideas and questions addressed in
this dissertation. Over the course of our many discussions about the 3'‘1 person
singular, social and personal responsibility, why the verb eat is so special, how
children should be treated and educated; types of verb phrases, life in academia,
bounded and unbounded W5, and the occasional political opinion; Cristina also
became a dear friend to me. I am very grateful to her for her friendship as well as
her scholarship. She is truly a wonderful “creature” (to borrow a term).

I have a special appreciation for Angeliek van Hout, whose experimental
work on telicity served as a foundation for some of the issues examined in this
dissertation. My discussions with her at the BU conference and the Telicity

Workshop she organized at IASCL in Berlin were very helpful in focusing the

viii

specific hypotheses and determining the verb-types that were examined in these
studies.

My fellow Ph.D. students, and their families, were a terrific source of
friendship, support, solidarity, encouragement and fun. They also helped me by
being actors for my videos and allowing their children to be pilot subjects in the
study. Many thanks to Johanna and Derrick Boult, Mary Jo Cooley Hidecker,
Ayla and My-Lanie Hidecker, and Greg, Monica and Maddie Robinson. Thank
you also to the linguistics students who agreed to be actors in the videos.

I also want to express my appreciation to all of those in the MSU Child
Language Lab who listened to and commented on several versions of these
studies, especially Karen Miller, Hsiang Hua Chang, and Kim Lewis. I would not
have been able to create all of the stimuli or collect all of the data in a reasonable
amount of time without the help of several research assistants who helped out at
various stages of the project. Thank you to Katerina French, Amy Geojo, Joe
Jalbert, Jamie Ludin, Adam Mayo, Andrew Sanford, Archie Soelaeman and Gina
Tower. Thank you also to Randy Fotiu and Sandra Herman for their statistical
expertise for the data analyses.

I also want to extend my gratitude to the Bamford-Lahey Children’s
Foundation, the American-Speech-Language—Hearing Foundation and the MSU
Graduate School for awarding me funding that made it possible for me to
complete my degree.

Very importantly, I need to thank the MSU students and the children who

participated in these studies. I greatly appreciate their time and effort.

Furthermore, I owe many thanks to the faculty and staff of the preschools and
elementary schools from which I recruited participants. Their assistance was vital
to the success of the study. I especially need to thank the staff at Happy
Elephant Child Care - Linden Road. Not only were they incredibly generous with
their time, space and cooperation, but more importantly, throughout my doctoral
program, they also provided a warm, caring, nurturing and fun environment in
which my children could learn and play.

I want to express my gratitude to those closest to me for making it
possible for me to pursue my goals. My parents, Zdzislawa and Wincenty Ogiela
always encouraged me to pursue my edumtion to the fullest. They always
supported me, even when they did not completely understand the choices I
made. I greatly appreciate that. I owe both my parents and my parents-in-law,
John and Joanne Ader. a great debt for traveling hundreds of miles to come and
help take of my children at several “crunch times” during my doctoral studies. It
meant so much to me to know that my children were in their grandparents’ care
while I was away. John and Joanne were also a great source of encouragement.
All of my family and my husband’s family have my heartfelt appreciation for all of
their help and very effective cheerleading skills. I also want to say thank you to a
good friend, Jaime Rodriguez, who was so generous with his time and spirit and
assisted my family in so many ways.

My children, Logan and Marlena Ader. have brought so much joy to me
regardless of how stressful things may have been at times. They have been ever

so patient and loving and wonderful. Logan has really been an integral part of

this dissertation. He was my number one pilot subject at several stages of the
project, he was my expert prize-picker-outer, and when l was collecting data at
his pre-school he was both a recruiter (“Your mom can sign you up for my mom’s
project and you can watch her movies”) and my clean up crew. Thank you
Logan, I couldn’t have done it without you. Although Marlena was baby for most
of this, she too had a role to play. On several occasions she entertained the
actors who came to record the video stimuli and she always entertained me.
Finally, my deepest gratitude is to my husband, John Ader. He did so
much to make it possible for me to earn my PhD, at an expense to his own
career and pursuits. At the same time, he kept our life full of humor and never let
too much time go by without making sure we had some fun. My gratitude is
beyond words, so I will keep it simple. You know what you did for me. Thank you

so very much.

xi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................... xiii
LIST OF FIGURES ............................................................................................. xv
CHAPTER 1 .......................................................................................................... 1
Introduction and Background ............................................................................ 1
Linguistic Choices and Telicity ....................................................................... 3
Linguistic Background .................................................................................... 6
CHAPTER 2 ........................................................................................................ 26
Linguistic Variables, Hypotheses and Predictions ........................................... 26
Linguistic Variables ...................................................................................... 26
Hypotheses and Predictions ........................................................................ 42
CHAPTER 3 ........................................................................................................ 53
Methods .......................................................................................................... 53
Study l: Adult Interpretations of Telic and Atelic VPs ................................... 53
Study ll: Children’s Interpretations of Telic and Atelic VPs .......................... 64
CHAPTER 4 ........................................................................................................ 77
Results ............................................................................................................ 77
Study l: Adults” Interpretation of Telicity ....................................................... 77
Study ll: Children’s Interpretation of Telicity ................................................. 92
Follow-up Data Analyses ........................................................................... 112
CHAPTER 5 ...................................................................................................... 117
Discussion ..................................................................................................... 1 17
Study I: Aduits’ Interpretation of Telicity ..................................................... 117
Study ll: Children's Interpretation of Telicity ............................................... 127
Clinical Implications and Future Research ................................................. 150
APPENDIX A .................................................................................................... 156
Linguistic Stimuli ............................................................................................ 156
APPENDIX B .................................................................................................... 158
Results of t-tests Comparing Token Verbs of Each Verb-type ...................... 158
APPENDIX C .................................................................................................... 159
Characteristics of Child Participants .............................................................. 159
APPENDIX D .................................................................................................... 165
Qualified Responses ..................................................................................... 165
REFERENCES ................................................................................................. 176

xii

LIST OF TABLES

Table 1: Results for English-Speaking Children and Adults: Mean Proportion of
Telic Answers as a Function of Sentence Type ........................................... 21

Table 2: Results for Dutch-speaking Children and Adults: Mean Proportion of
Telic Answers as a Function of Sentence Type ........................................... 21

Table 3: Verbs and Object DPs Included in the Experimental Task, by Verb-type
and Deterrniner—type .................................................................................... 56

Table 4: Means and Standard Deviations of the Counts of Telic Interpretations
per VP Condition for Build-type and Eat-type Verbs for Adults .................... 78

Table 5: Summary of ANOVA Results for Adults: Verb-type by Deten'niner-type
for Eat-Type and Build-Type Conditions ...................................................... 79

Table 6: Means and Standard Deviations of the Counts of Telic Interpretations

per VP Condition for Eat-type, Eat up-type, Push-type and Push over-type
Verbs for Adults ........................................................................................... 83

Table 7: Summary of ANOVA Results for Adults: Verb-type by Deterrniner-Type
for Eat-Type and Eat up-Type Conditions .................................................... 85

Table 8: Summary of ANOVA Results for Adults: Push-Type vs. Push over-Type
..................................................................................................................... 88

Table 9: Means and Standard Deviations of the Counts of Telic Interpretations
per VP Condition for Build-type and Eat-type Verbs for Children ................. 93

Table 10. Summary of ANOVA Results for Children: Eat-Type vs. Build-Type...94

Table 11: Means and Standard Deviations of the Counts of Telic Interpretations
per VP Condition for Eat-type, Eat up-type, Push-type and Push over-type

Verbs for Children ...................................................................................... 101
Table 12: Summary of ANOVA Results for Children in Eat-type and Eat up-type
Conditions .................................................................................................. 102
Table 13: Summary of ANOVA Results for Children for Push-type vs. Push over-
type Verb conditions .................................................................................. 107
Table 14: Summary of Results for Children for Build-type vs. Eat-type vs. Push-
type Verb Conditions. ................................................................................ 113
Table 15: Examples of Children's Indirect No Responses ................................ 147

xiii

Table 16: Examples of Children’s Qualified Responses ................................... 149
Table A1: Linguistic Stimuli that Served as Target Questions for Analyses ...... 156

Table 81: Results of t-Tests Comparing Responses to Questions with the Token

Verbs of Each Verb-Type within Each Experimental Condition ................. 158
Table C1: Test Scores and Demographic Characteristics of Child

Participants ................................................................................................ 1 59
Table D1: Children's Qualified Responses ........................................................ 165

xiv

LIST OF FIGURES
Figure 1. Basic telicity ingredients ....................................................................... 16
Figure 2. Ingredients of telicity expanded to distinguish between partitive and
non-partitive quantity sensitive verbs and to distinguish between cardinal

numbers and definite deten'niners serving as quantizers ............................. 40

Figure 3. Mean counts of telic interpretations by verb-type for adults in build-type
and eat-type conditions. ............................................................................... 79

Figure 4. Mean counts of telic interpretations by determiner-type for adults in
build-type and eat-type conditions. .............................................................. 80

Figure 5. Mean counts of telic interpretations for verb-type by determiner-type for
adults in eat-type and build-type conditions. ................................................ 82

Figure 6. Mean counts of telic interpretations for verb-type for adults in eat-type
and eat up-type conditions ........................................................................... 85

Figure 7. Mean count of telic interpretations by detenniner—type for adults in eat-
type and eat up-type conditions. .................................................................. 86

Figure 8. Mean counts of telic interpretations by verb-type and determiner-type
for adults in eat-type and eat up-type conditions. ........................................ 86

Figure 9. Mean counts of telic interpretations by verb-type for adults in push—type
and push over-type conditions. .................................................................... 89

Figure 10. Mean counts of telic interpretations by determiner-type for adults in
push-type and push over-type conditions. ................................................... 89

Figure 11. Mean counts of telic interpretations by verb-type and determiner-type
for adults in push-type and push over-type conditions. ................................ 90

Figure 12. Mean counts of telic interpretations by verb-type and determiner-type
for adults in eat/eat up—type and push/ push over-type conditions. .............. 92

Figure 13. Mean count of telic interpretations by verb-type for children in build-
type and eat-type conditions. ....................................................................... 96

Figure 14. Mean count of telic interpretations by determiner-type for children in
build-type and eat-type conditions. .............................................................. 96

Figure 15. Mean count of telic interpretations by verb-type and determiner-type
for children in build-type and eat-type conditions by age group ................... 97

Figure 16. Mean counts of telic interpretations by verb-type for children in eat-
type and eat up~type conditions. ................................................................ 103

Figure 17. Mean counts of telic interpretations by determiner-type for children in
eat-type and eatup-type conditions. ........................................................... 103

Figure 18. Mean count of telic interpretations by verb-type and determiner-type
for children in eat-type and eat up-type conditions by age group ............... 104

Figure 19: Mean count of telic interpretations by verb—type for children in push-
type and push over-type conditions. .......................................................... 108

Figure 20. Mean counts of telic interpretations by determiner-type for children in
push-type and push over-type conditions. ................................................. 108

Figure 21. Mean count of telic interpretations by verb-type and determiner-type
for children in push-type and push over-type conditions by age group. ..... 110

Figure 22. Mean counts of telic interpretations by verb-type for children in build-
type, eat-type, and push-type conditions. .................................................. 114

Figure 23. Mean counts of telic interpretations by verb-type and determiner-type
presented by age group for children and adults for build-type and eat-type
conditions ................................................................................................... 130

Figure 24. Mean counts of telic interpretations by verb-type and determiner-type
presented by age group for children and adults for eat-type and eat up—type
conditions ................................................................................................... 143

Figure 25. Mean counts of telic interpretations by verb-type and determiner-type

presented by age group for children and adults for push-type and push over-
type conditions. .......................................................................................... 145

xvi

CHAPTER 1

Introduction and Background

The goal of this dissertation was to contribute to the understanding of verb
phrase (VP) telicity. The term felicity is used to refer to the temporal property of
linguistic descriptions associated with logical endpoints. If a description of a
situation includes a logical endpoint, it is said to be a telic description; and if it
does not, it is said to be an atelic descfiption’.

Past research has indicated that cross-linguistically, there is a relationship
between verb morphology and VP aspect in the language of young typically
developing (TD) children (e.g., Antinucci & Miller, 1976; Bloom, Lifter 8 Hafitz,
1980; Bronckart & Sinclair; Hamer, 1981; Johnson & Fey, 2006; Shirai &
Andersen, 1995; Wagner, 2001; Weist, Wysocka, WItkowska-Stadnik,
Buczowska & Konienczna, 1984). Specifically, it has been shown that TD
children exhibit sensitivity to VP aspect in their early differential use of verb
morphology. They tend to first use the progressive morpheme in activity VPs and
the past tense morpheme in event VPs. Telicity is one of the features embedded
within VP aspect. Activity VPs are atelic and event VPs are telic. The feature of
VP aspect that seems to be responsible for this relationship is telicity. Children
appear to be matching the telic or atelic characteristic of the VP aspect with the

perfective (i.e, -ed) and imperfective (i.e., -ing) characteristics of viewpoint

 

I Some authors prefer the terms bounded and unbounded rather than telic and atelic. The terms
telic and atelic will be used here and are considered essentially synonymous to bounded and
unbounded.

aspect,2 respectively. Thus, it can be said that TD children seem to show
sensitivity to verb telicity in their early differential use of verb morphology.

Although there have been many studies on the relationship between tense
and aspect as cited above, there has been relatively little research conducted on
children’s ability to interpret telicity itself, without the influence of verb
morphology. In order to better understand this relationship it is important to study
children’s interpretation of telicity itself. Focusing on the compositionality of
telicity, van Hout (1998) conducted a study in which she examined adults’ and
children’s and interpretations of telicity in a variety of syntactic conditions not
involving bound verb morphology. This dissertation builds on that work by
examining the interpretation of telicity in adults and children using different verb-
types and deterrniner-types.

This project was comprised of two related studies. The first examined
adults’ interpretation of telicity under various verb-type and detenniner—type
conditions and the second examined children’s interpretation of telicity in a cross-
sectional developmental study of 3-, 4-, 5-, and 6-year-olds. Chapter 1 begins
with examples of telic and atelic descriptions and provides relevant linguistic
background concerning telicity and its relationship to other temporal properties of
language. The first chapter concludes with a description of an important earlier
study that experimentally examined theoretical predictions regarding the

interpretation of telicity in both adults and typically developing (l’ D) children (van

 

2 The motivation for the relationship between VP aspect and tense morphology has been the
topic of much discussion, but is beyond the scope of the present paper. See Li and Shirai (2000)
for a detailed discussion regarding various past accounts that have been proposed as the source
of this correlation. See Bohnemeyer and Swift (2004) for a more in depth account.

Hout, 1998). Chapter 2 discusses the specific linguistic variables, hypotheses,
and predictions examined in the studies. Chapters 3 and 4 report the
methodology used and the results obtained. Finally, chapter 5 discusses the

interpretation of the results and makes suggestions for future research.

Linguistic Choices and Telicity

There are many ways in which we can describe any given situation. When
we speak, we make many linguistic choices3 that convey our specific intended
meaning. When we describe an event, the semantic and syntactic choices we
make interact to indicate whether or not the event, as we are describing it, has a
logical endpoint or not. Whether or not a description includes a logical endpoint is
referred to as telicity. If a description indicates that the situation described
culminates in a logical endpoint, it is telic; if it does not indicate that it culminates
in a logical endpoint, it is atelic. For example, the sentence in (1a) is telic; the
logical endpoint is reached when John arrives at the park. The sentence in (1b) is
atelic. There is no time at which walking in the park will culminate at a logical
endpoint. There are several tests used for determining whether a verb phrase
(VP) is telic or atelic. One of the most important tests is the contrast between
adverbials such as, forX time, and in X time. If I say (1a), and modify it with for
two hours, we don’t know whether John ever arrived at the park. In contrast, if I
say (1a) with the adverbial in two hours, I am asserting that the walking to the

park took two hours. Now consider the sentence in (1b). If I say (1 b), and modify

 

3 The use of the word choices here is not meant to refer to conscious choices; rather it refers to
the selection of particular linguistic forms in the course of speaking.

it with the adverbial for two hours, I am asserting that the activity of walking in the
park continued for a period of two hours, but I am not asserting that some logical
endpoint was reached. If I say (1b) and modify it with in two hours, the sentence
is aMcward because there is no logical endpoint to walking in the park. In two
hours provides the temporal interval within which the event is to be completed

but walk in the park does not have a completion point.

(1). a. John walked to the park (#for two hours/in two hours).

b. John walked in the park (for two hours/*in two hours).

Another test used to determine if a VP is telic or atelic is the completion
entailments test. The test goes as follows: Suppose a particular VP was in
progress (using the imperfective -ing form of the verb), and it was suddenly
discontinued. Is it true that it occurred? If the answer is yes, then the VP is atelic.
If the answer is no, then the VP is telic. Consider the sentences in (2) and (3). In
order to determine if the VP in (2a) is telic or atelic, the sentence can be put into
its imperfective form as in (2b), and one can ask, “If Logan was running, and
suddenly stopped running, did he run?" The answer is yes; therefore the VPs in
the sentences in (2) are atelic. The result is different for the VP in (3). If (3a) is
put into its imperfective form as in (3b) and one asks, “If Logan was drawing a
circle, and suddenly stopped drawing a circle, did he draw a circle?” The answer
is no, therefore the VPs in the sentences in (3) are telic. In other words, if a telic

event is discontinued suddenly (i.e., before it reaches its logical endpoint) then

the VP that describes it is not true. The behavior of telic predicates became

known as the imperfective paradox (Dowty, 1979).

(2) a. Logan ran. (atelic VP)

b. Logan was running.

(3) a. Logan drew a circle. (telic VP)

b. Logan was drawing a circle.

The telicity of a VP is not overtly indicated by a simple word or morpheme.
Rather, it is dependent on an interaction between semantic and syntactic
features of the elements in the VP. When we choose a particular verb,
determiner, verb particle, or prepositional phrase, etc., among other choices, we
are essentially choosing whether or not the sentence we produce includes an
endpoint. Although we are most likely unaware of it, when we choose particular
elements to include in a sentence, we are making a decision about its telicity. In
the case of the sentences in (1) the type of prepositional phrase determined
whether the description was telic or atelic. The VP elements impacting telicity
that were examined for this project are: verb-type, determiner-type and

resultative verb particles.

Linguistic Backgmund
Temporal properties of verb phrases

Prior to discussing the particular elements that were examined here, it is
important to place the notion of telicity within the larger context of the temporal
properties of VPs. Telicity is just one of the temporal properties of VPs. In order
to provide a more complete description of telicity, I will outline other temporal
notions that are encoded in language. Specifically, I will briefly discuss tense, VP
aspect (which includes telicity), viewpoint aspect, and how they relate to each
other.

Sentences describe states, activities or events. The term eventuality
description, from Bach (1986), will be used to encompass states, activities, and
events. Eventualities happen in time. In natural languages, tense encodes the
relationship between the time at which an eventuality takes place (i.e., event
time) and the time at which it is talked about (i.e., speech time). It is a deictic
relation that depends on the speech time, that is, the time of speaking about an
eventuality (Comrie, 1976). Eventualities, independent of being identified as past,
present, or future, also have other temporal properties. Aspect is the term used
for the various ways in which we linguistically encode the internal temporal
contour of a situation, that is, how a situation unfolds, regardless of the speech
time. (e.g., Comrie, 1976; Klein, 1994; Smith, 1991).

Languages vary as to how richly they overtly encode tense and/or aspect.4

Natural languages make use of only a small set of the possible temporal

 

‘ Note that aspect is a larger construct than telicity, but that the telic/atelic distinction is important
to aspectual understanding.

eventuality contours. This project will concentrate on the division of eventualities
on the basis of telicity, that is, based on whether the eventuality description has
an inherent logical endpoint, and is therefore telic; or whether its description does
not indicate an inherent logical endpoint, and is therefore atelic.

There are at least two layers with which we construct the contour of a
situation syntactically prior to locating an event as occurring before or after the
speech time. First, an eventuality description is built by combining a verb with its
internal arguments; the result of which is a telic or an atelic VP. This layer, which
we will call the VP layer, is generally referred to as lexical aspect or aktionsart,
since it heavily depends on the lexical properties of the verb. After the internal
contour of the VP is constructed, other aspectual markers such as the
progressive or the perfect are added to the VP. This second aspectual layer is
generally referred to as grammatical or viewpoint aspect.

The aspect of the first layer (i.e., the VP layer) is compositional, resulting
from the interaction between the lexical properties of the verb, particles, and the
internal arguments (e.g., determiner phrases). Therefore, it is necessary to
address the properties of the entire VP, or predicate, and not just the verb itself in
determining telicity and categorizing the aspect of the VP layer (Krifl<a, 1989;
Verkuyl, 1989, 1993). The term lexical aspect implies that aspect is a property of
a specific lexical item rather than a property of the whole VP, obtained

compositionally. However, it is not just the properties of the verb that determine

aspect/telicity. The properties of the verb’s determiner phrase5 (DP)
complements play a particularly important role in aspectual interpretation (Kriflta,
1989; Schmitt, 1996; Verkuyl, 1989, 1993). Although the term lexical aspect is
often used in the literature as referring to compositional aspect, for clarity, I will
refer to the aspect of the VP layer as VP aspect rather than lexical aspect. Telic
VPs will be referred to as events and non-stative atelic VPs will be referred to as
activities.‘5

Because of the compositionality of VP aspect, any given verb will interact

with the syntax of the rest of the VP. Thus, in one syntactic context a particular
verb may be part of an activity/atelic VP, and in another syntactic context the
same verb may be in an event/telic VP. The examples in (4) below illustrate that
the presence of determiner phrases of varying types in the object position, or a
resultative particle, can alter the telicity of a VP. Note that the characterizations in
(4) are pure semantic descriptions and do not reflect exactly how these
sentences are processed and evaluated within a context, as will be addressed in
the research presented here.

(4). a. He ate. (intransitive; atelic/activity; If he was engaged in eating,
the sentence is true whether or not he finishes all of what he was
eating).

b. He ate apples. (bare plural DP; ateliclactivity; If he engaged in

eating more than one apple, the sentence is true whether or not he
finished eating any of the apples).

 

5 Some readers may be more familiar with the term noun phrase rather than determiner phrase.
In generative linguistics the determiner is the head of the determiner phrase and contains the
noun phrase.

States are also atelic, but will not be discussed here.

c. He ate the/two apples. (determiner/cardinal number + noun;
telic/event; If he finished eating the given apples, the sentence is
true).

d. He ate up the/two apples. (resultative verb particle; telic/event; If
he finished eating all of the given apples, the sentence is true).

In optionally intransitive constructions, as in (4a), the VP is atelic. The
described eventuality does not have an inherent logical endpoint. Note that with a
VP that has a bare DP as in (4b) there is a direct object, but it carries no quantity
(i.e., how many) information. Therefore, there is no logical endpoint to the
eventuality, and it is atelic (e.g., Borer, 2005; Schmitt, 1996; Verkuyl, 1993). In a
VP that has a determiner + noun as a direct object as in (4c), the eventuality is
rendered telic because we have quantity information from the object. In a VP that
has a resultative particle, as in (4d), the eventuality is telic because the particle
forces a result state, indicating the completion or culmination of the event.

The VP layer is fundamental for the whole interpretation of the aspectual
properties of a clause because the viewpoint aspect layer takes the VP layer as
its input and modifies its eventuality description. With regard to vieWpoint aspect,
a speaker chooses a particular portion of the eventuality to assert (Klein, 1994)
and it is this interval that is actually located by tense as being before, during, or
after the speech time. The description can refer to the entire eventuality or just a
part of it. This type of aspect is usually indicated through the use of linguistic
elements such as inflectional morphemes. For example, consider the
descriptions of the 'bridge building’ eventuality in (5). We can assert that it is or

was in progress, as in (5a), or we can assert that the event has culminated as in

(5b). Both VPs are telic and have logical endpoints. The difference is that the
sentence in (5a) specifies that the event islwas in progress at the time that is
being referred to and (5b) specifies that it islwas already completed at the time
that is being referred to. In other words, in (5a) the speaker is choosing to talk
about only a portion of the eventuality, whereas in (5b) the speaker is choosing to

talk about the entirety of the eventuality.

(5) a. John is building a bridge/Was building a bridge.
b. John has built a bridge/had built a bridge.

In many languages the overt marking of viewpoint aspect is a much more
integral part of the grammar than it is in English. In English, the only morphemes
that explicitly and overtly mark viewpoint aspect are the progressive
(imperfective) be + -ing, and the perfect have + -en/-ed. Some languages such
as Mandarin, mark viewpoint aspect but do not mark tense. Others, such as
Polish and Russian, mark both tense and aspect obligatorily. In English, tense
and aspect interact in predictable and interesting ways. The third-person singular
present-tense and the past-tense morphemes do not explicitly mark viewpoint
aspect, but they do affect aspectual interpretation by triggering inferences wlled

implicatures. For example, consider the sentences in (6).

(6). a. John walks heme.

b. John walked home.

10

c. John walked home while Mary drove to work.

d. John isMas walking home.

In (6a), the present tense third-person singular morpheme, -s, is
associated with a generic/habitual interpretation in which the activity is most likely
not taking place in the present moment, although it may be. It is imperfective. In
(6b) the past-tense morpheme, -ed, imparts a perfective reading that implies that
the activity of walking home has been completed. However, this is not
necessarily so. As can be seen in (6c) both events take place in the past, but the
reading of the sentence is not unequivocally perfective because it does not
specify whether, in fact, John ever arrived at home, or if Mary ever arrived at
work, just that the two were taking place simultaneously at some time in the past.
By contrast, the progressive in (6d) overtly indicates that the walking eventuality
was in progress, either at the time of speech in the case of the present tense, or
prior to the time of speech in the case of the past tense, without indication that
the walking eventuality stopped by the time of speech. It is imperfective.

In sum, viewpoint aspect is related to perfectivity and imperfectivity and is
expressed through grammatical devices. Although VP telicity, viewpoint aspect,
and tense interact within a given sentence, they are independent from one
another. In other words, a telic VP does not preclude the use of a tense
morpheme that implies imperfectivity or the use of an aspect morpheme that
marks atelicity ( be + -ing). Telic VPs express eventualities that have inherent

endpoints, regardless of the verbal morphology that accompanies them and the

11

time at which the event takes place, as the examples in (7) below demonstrate.

Atelic VPs do not have inherent endpoints, mgardbss of the verbal morphology

that accompanies them, as the examples in (8) below demonstrate.

(7)

(8)

a. Logan is building a bridge/built a bridge.
(The inherent endpoint is when the bridge is completed).

b. Marly is sewing/sewed a dress.
(The inherent endpoint is when the dress is completed).

c. The dog is dying/died.
(The inherent endpoint is when the dog is dead).

a. John is walking in the park/John walked in the park.
(There is no logical endpoint).

b. Marty is swimming in the pool/swam in the pool.
(There is no logical endpoint).

c. The dog is eating/ate.
(There is no logical endpoint).

The examples above demonstrate that, although VP aspect and viewpoint

aspect interact with regard to interpretation of a sentence, they are independent

of one another. lmperfective morphology can occur in either a telic or atelic VP

and perfective morphology can occur in either a telic or atelic VP.

Telicity Ingredients

Verb type: Quantity-sensitive vs. quantity-insensitive verbs. Non-stative

verbs can be divided into two categories, quantity-sensitive verbs and quantity-

insensitive verbs. For quantity-sensitive verbs the amount of the object being

affected matters for the interpretation of the VP as telic or atelic, but for quantity-

12

insensitive verbs information about the quantity of the object does not matter. For
example, the verb build is a quantity-sensitive verb because information about
the object is relevant to determine an activity reading (atelic) or an event (telic)
reading. For example, the sentence He built houses for years is an acceptable
atelic sentence but the sentence He built two houses for years, is not. The latter
requires some semantic coercion in order to get a reasonable interpretation. The
verb push is an example of a quantity-insensitive verb, in that the amount of
pushing does not matter, once any pushing has begun, pushing has taken place.
For example, He pushed the cart/two carts would be judged as true whether the
cart/two carts were pushed two centimeters or two kilometers, as long as there
was a cart in the context and it was pushed. Whether the object is carts, a cart,
the carts, or two carts has no bearing on the interpretation of telicity with push-
type verbs. In sum, quantity-insensitive verbs are not sensitive to the quantity
information in the object DP, while quantity-sensitive verbs are very sensitive to
the quantity information in the object DP.

Quantized and non-quantized DPs. Deten'niner phrases may be divided
into two main types: quantized and non-quantized. A quantized DP provides
information about the size of the set that is being referred to or specifies the
particular set of objects that is being referred to. For example, the DP, two toys,
indicates the quantity of toys being referred to. When no quantity information is
present, the noun phrase is said to be non-quantized. For example, the DP, toys,
does not provide information about the amount of toys being described.

Potentially, the DP toys can be associated to an infinite quantity of toys. It simply

13

indicates more than one toy, or the concept of toys, as in Toys are expensive,
with no information about the size of the set (i.e., the number of toys or the
particular set of toys). Besides DPs that contain a cardinal number, other
examples of quantized DPs include those with a possessive pronoun or a definite
determiner such as in her toys or the toys. Although these DPs do not indicate a
specific number of toys, in order to interpret the definite determiner or the
possessive in an DP in argument position, there must be a specific set in the
discourse that is being referred to. In this case the quantity is the whole set, no
matter how many individual objects the set contains. Quantized DPs include
those with definite or indefinite deterrniners, cardinal numbers, or possessive
pronouns, among others. Non-quantized DPs include those with bare plurals or
mass nouns (e.g., cheese, milk, rice, etc.).

Resultative particles. The presence of a resultative particle verb in a VP
also contributes to its telicity. Although different authors may categorize particle
verbs differently, the categorizations tend to be similar and commonly distinguish
between three types. Dehé (2002) summarizes these as 1) compositional particle
verbs whose meanings are the combination of the meaning of the verb and the
meaning of the particle (e.g., take out, cany in); 2) idiomatic particle verbs that
are a complete unit, but whose meanings are not literally composed of the
meanings of the verb and the particle (e.g., tum down, put all“); and 3) aspectual
particle verbs in which the particle contributes an aspectual interpretation to the
verb (e.g., eat up, used up). Only compositional and aspectual particles are of

interest here. Both compositional and aspectual resultative particle verbs can

14

indicate a result state and therefore contribute to a telic interpretation. For
example, She carried out the suitcases contains a compositional particle verb
and indicates a result state in which the suitcases are in a new location outside of
the original location and She ate up the sandwich contains an aspectual particle
verb and indicates a result state in which the sandwich has been completely
consumed. Both resultative particle verbs cause the VP to have an
unambiguously telic meaning. However, there is an important difference as well.
In the compositional resultative particle verb, the particle is directional or goal
oriented and the lexical meaning of the verb itself is altered. In the aspectual
resultative particle verb, the particle verb unambiguously indicates telicity, but the

lexical meaning of the verb itself is not altered.

15

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.05.... :5 E3
2.. .26 :2...
9.3 rad a: Etna-o .66
32.830...
.05....83 3.5.3 33> 0.2.3... 33>
23:52 .2..an 268.38. 20:39:02
M — — L
. 322.3... 25.2..»
no bounces—v.32 an 39:25 $55.5 $55.5
F L H L
. W
7.5.3.9.. 3552 H 7.5.85... .seoL

 

 

#

_
H 8.8.3.9.. 226 5.2.2. ..> H

 

 

_

 

 

16

Figure 1. Schematic of basic telicity ingredients.

Theoretical Predictions for the Interpretation of Telicity

The basic ingredients of telicity are presented in Figure 1 above. The ways
in which these ingredients combine are essentially responsible for determining
whether a VP is telic or atelic. Although there may be differences in terminology,
the following predictions follow largely from the seminal work of Verkuyl (1972)
and its various implementations (Verkuyl, 1989, 1993), We (1989) and others.
The various combinations of quantity-sensitive vs. quantity-insensitive verb,
quantized vs. non-quantized DP and presence vs. absence of resultative particle
predict either a telic or atelic VP interpretation. The specific predictions are listed

in (9) below.

(9) a. Quantity-sensitive V (intransitive) —> Atelic VP

e.g., She ate.

b. Quantity-sensitive V + non-quantized object DP —- Atelic VP
e.g., She made birthday cards.

0. Quantity-sensitive V + quantized object DP -> Telic VP
e.g., She made the/two/her birthday cards.

d. Quantity-sensitive V + resultative particle + quantized object
DP—> Telic VP

e.g., She ate up the apples.

17

e. Quantity-insensitive V + resultative particle + quantized
object DP—-> Telic VP
e.g., She carried out the/her luggage.

f. Quantity-insensitive V + quantized object DP -> Atelic VP
e.g., She carried her luggage.

g. Quantity-insensitive V + non-quantized object DP —-> Atelic
VP

e.g., She carried luggage.

An Experimental Test of Telicity Interpretation

van Hout (1998) points out that identifying telic vs. atelic VPs in languages
such as English and Dutch presents a challenge for children because telicity is
not overtly indicated. Rather, due to the compositional nature of telicity, it must
be computed from the specific properties of the verb, the direct object DP, and
particles. In her study, she specifically examined the role that the stmcture of
direct object DPs and resultative particles have on adults’ and children’s
interpretation of telicity in English and Dutch in the context of the quantity-
sensitive verbs eat and drink. This study is particularly important because it
provided the first behavioral test of some of the theoretical predictions listed in (9)
above. In other words, it evaluated whether or not participants actually

interpreted the telicity of VPs as theoretically predicted.

18

The participants in the study were adults, and 3-, 4-, and 5-year-old
children who were monolingual speakers of either Dutch or English. The
research participants were presented with a picture scene in which there were
two distinct characters and the experimenter provided a narrative story to explain
the scene. One character engaged in an eating or drinking event until it was
completed, (i.e., culminated in a logical endpoint). The other character engaged
in the same event but stopped before it culminated in its logical endpoint. For
example, a scene was presented in which a red mouse ate an entire piece of
cheese and a white mouse ate a few bites of a piece of cheese. In the case of
the character that completed the event, the end of the accompanying narrative
emphasized the completion: “Look, here he’s eating. The red mouse likes his
cheese very much. You can see that here: his cheese is all gone” (van Hout,
1998, p. 401). In the case of the character that stopped before the event reached
its logical endpoint, the end of the accompanying narrative emphasized that the
lack of completion: “Look, there he ’s eating. He takes a couple of bites, but his
cheese is too big for him for now. He leaves a piece for later” (van Hout, 1998, p.
401).

Four different types of sentences were tested in a modified tnrth-value
judgment task: (a) intransitive (e.g., Did the white/red mouse eat?); (b) transitive
with a bare object DP (e.g., Did the white/red mouse eat cheese?); (c) transitive
with a quantized direct object, indicated by a possessive pronoun in the object
DP (e.g., Did the red/White mouse eat his cheese ?); and (d) a resultative particle

verb (e.g. Did the white/red mouse eat up his cheese ?). These conditions

19

essentially correspond to the conditions listed in (9a) through (9d) above. The
specific predictions of van Hout (1998) were: (a) resultative particles would be
understood as telicity markers early on; (b) whether or not an object is present
will not be recognized as a (a)telicity marker early on; (c) initially the
(un)boundedness of the VP (i.e., a quantized or non-quantized DP), is not
understood as a (a)telicity marker. The last of van Hout’s (1998) predictions
implies that adults will understand the boundedness of the VP (i.e., quantity-
sensitive verb + quantized DP), as a telicity marker.
Sentence types (a) and (b) were atelic and their results were collapsed into one
group called atelics because the participants’ responses in these conditions were
essentially identical. The results were analyzed in terms of the proportion of
participant answers that indicated a telic interpretation of the question. For the
convenience of discussion, van Hout’s (1998) results are reproduced in Tables 1
and 2 below. The numbers indicate the proportion of respondents in each group
that had a telic interpretation of the VP in the target question. That is, it is the
proportion of the respondents who, consistent with the theory, answered no to
the target questions for the character who did not completely consume the
object. Given the theoretical predictions, the proportions in the Atelies columns
were expected to be low and the proportions in the His/her object and Particle up
columns were expected to be high.

For both English-speaking and Dutch-speaking adults, atelics were rarely
interpreted as telic. For the particle sentences, the proportion of telic responses

by the English-speaking adults was .81 and the proportion of telic responses by

20

Table 1: Results for English-Speaking Children and Adults: Mean Proportion of

Telic Answers as a Function of Sentence Type

 

 

Group Atelics His/her object Particle up
3-year-olds .37 .45 .66
4-year-olds .44 .56 .62
5-year-olds .38 .56 .91
Adults .03 .25 .81

 

Note. From “On the role of direct objects and particles in learning telicity in Dutch and English,” by
A. van Hout, 1998, Proceedings of the Annual Boston University Conference on Lanuguage

Development, 22, p. 405. Copyright 1998 by Angeliek van Hout. Reprinted with permission.

Table 2: Results for Dutch-speaking Children and Adults: Mean Proportion of

Telic Answers as a Function of Sentence Type

 

 

Group Atelics His/her object Particle up
3-year-olds .20 .17 .50
4-year-olds .35 .50 .87
5-year-olds .38 .47 .90
Adults .14 .78 1.00

 

Note. From “On the role of direct objects and particles in Ieaming telicity in Dutch and English,” by
A. van Hout, 1998, Proceedings of the Annual Boston University Conference on Lanuguage

Development, 22, p. 404. Copyright 1998 by Angeliek van Hout. Reprinted with permission.

21

the Dutch-speaking adults was 1.0. Those results were largely expected. The
results for the possessive DP sentences, which were used to test the influence of
quantized NPs on telicity, did not meet the expectation for adults (i.e., telic
response to questions with quantity-sensitive verb and quantized object DP),
particularly for English-speaking adults. The proportion of telic responses by the
English-speaking adults in the quantized DP sentences was only .25 and the
proportion for the Dutch-speaking adults was .78.

Because the primary interest here is the English-speakers’ results, the
Dutch results will not be discussed in detail. For the English-speaking children,
the 3- and 5-year-olds demonstrated a significant effect of sentence type, with
the particle condition being distinguished from the atelic sentences even by the
3-year-olds, whereas the 4-year-olds seemed to treat all of the sentence types
equally. The 5-year-old children appeared to be particularly adept at identifying
the particle as a marker of a telic VP.

Statistically, none of the child groups distinguished the his/her object
sentences from the atelic sentences. Although the adults only interpreted the
his/her sentences as telic a quarter of the time, this was significantly different
from their interpretations of the atelic sentences. However, it is interesting to note
that the lack of difference in the children’s interpretation of the atelic and his/her
object sentences appears to not only be due to a relatively low proportion of telic
interpretations of his/her object sentences, but also due to relatively high
proportion of telic interpretations of the atelic sentences, as compared to the

adults. In contrast, the adults had a very low proportion of telic interpretations of

22

atelic sentences. Furthermore, it is interesting to note that the children at all ages
responded more closely to the prediction for the his/her object condition than did
the adults.

The results for the English-speakers in van Hout’s (1998) study are
somewhat difficult to interpret, especially since the adults did not clearly interpret
VPs containing the verbs eat and drink and a quantized NP as being telic,
contrary to theoretical predictions (cf. Verkuyl, 1993), and the children did not
demonstrate a clear developmental pattern in this condition. There are several
methodological and design issues that may have contributed to these results.

One potential issue is the use of static pictures supplemented with a
narrative description of the actions. First, using static pictures requires the
participants to fill in and imagine the entirety of the event themselves and not all
participants may imagine the event progressing in the exact same way. Second,
there seems to be an assumption that the participants will correctly draw the
conclusion that the picture they are looking at for the atelic event type depicts the
end of the event that they are asked to judge. Third, the narrative may be
distracting to subjects, making them think about the story that is being told about
the circumstances of the event (e.g., “He takes a couple of bites, but his cheese
is too big for him for now”) and not necessarily about the event itself. Lastly, the
narrative for the atelic event concludes with a line such as “He leaves a piece for
later,” which may be confusing to participants and they may answer the

questions with regard to what was likely to occur later. These issues may be

23

controlled for by using videotaped stimuli, eliminating the need for narrative
stories as the progression of the eventuality would be clear from the video.

Another issue is that possessive pronouns such as his and her may not be
representative of quantizers as a whole. The possessive pronoun does not
ensure a count reading of the determiner phrase and can allow for a mass
reading of it. If one has a mass reading of the determiner phrase, it becomes
atelic because it is non-quantized. For example, if two farmers, one male and
one female, produce cheese, one can say Her cheese is better than his cheese
and be referring to the cheese produced by each farmer in general, not some
specific chunk of cheese that can be counted. The same can be said for a noun
that is usually a count noun. If two bakers make cookies, one can say, His
cookies are better than her cookies. In this example, cookies are referred to as a
mass, and not a specific set of cookies. Therefore, the determiner phrases
composed of a possessive pronoun and a noun are not necessarily quantized.
Furthermore, questions such as “Did he eat his cheese?” can easily be
interpreted as asking “Did the mouse eat his own cheese as opposed to the other
mouse’s cheese?” If this is the way a participant interprets the question, then yes
is a reasonable answer in both the telic and atelic event conditions. This may
mask the participants’ true interpretations of the VP.

Finally, the study only examined two exemplars of one verb-type. The
compositionality of aspect includes the interaction between the verb and the rest
of the VP. Not all verb types would necessarily be expected to interact with a

quantized NP in the same manner. Furthermore, cross-linguistically, the verbs

24

eat and drink behave in peculiar ways (Zribi-Hertz, 2003) which may limit the
degree to which we would be able to generalize the results obtained with these
verbs.

van Hout’s (1998) study was particularly important for several reasons.
First, it indicated that the general predictions for telic/atelic VPs may not be fine-
grained enough to account for adults” true interpretations. Second, it introduced
important questions that need to be addressed in order to further understand
children’s development of the interpretation of telicity. Given that the children did
not interpret telicity in her study in the same manner as adults, it leads to yet
another question; if children have difficulty calculating telicity, how is it that they
appear to be matching the past tense and progressive morphemes to telic and
atelic descriptions, respectively, in the studies that examine early child
morpheme production (e.g., Antinucci & Miller, 1976; Bloom, Lifter 8 Hafitz,
1980; Bronckart & Sinclair; Harner, 1981; Johnson & Fey, 2006; Shirai &
Andersen, 1995; Wagner, 2001; Weist, Wysocka, Witkowska-Stadnik,
Buczowska & Konienczna, 1984)? (See page 1). Third, it established a useful
framework for examining the interpretation of telicity experimentally. The studies
presented in this dissertation build on this foundation by altering the methodology
somewhat and by expanding some of the linguistic variables into more specific

categories.

25

CHAPTER 2

Linguistic Variables, Hypotheses and Predictions

As discussed in chapter 1, the meaning of a sentence is more than the
sum of its parts. In order to interpret a sentence, we not only need to know the
meaning of its parts, but also how they interact with each other. Full aspectual
interpretation of a sentence is determined by multiple semantic, syntactic and
pragmatic properties. Understanding how people interpret and indicate aspectual
information requires the examination of several semantic and syntactic properties
including the lexical properties of verbs, verb-particle combinations, the internal
structure of the internal arguments, and whether VPs include perfective or
imperfective morphology. The focus of this project was limited to the examination
of adults’ and children’s abilities to integrate the lexical properties of verbs and
verb particle constructions with internal verb argument properties (i.e., object DP)
in order to interpret the aspectual properties of the VP.

This project was comprised of two studies, one with adult participants and
one with child participants at 4 age levels. They addressed adults’ and children’s
abilities to integrate semantic and syntactic information in the VP in order to

obtain telic or atelic interpretations of VPs.

Linguistic Variables

Because the telicity of a VP is dependent on a variety of factors, it is

possible that different factors will influence interpretation at various stages of

26

linguistic development. The variables used here are representative of the
ingredients involved in calculating telicity and include verb-type, determiner-type
in the direct object DP, and the presence of a resultative particle. These have
been selected to address specific questions and to provide a finer-grained

analysis of telicity.

Verb-type

Quantity-insensitive verbs are expected to yield atelic interpretations of a
VP regardless of whether the object DP is quantized or non-quantized (in the
absence of certain types of prepositional phrases), because the VP is simply
insensitive to amount. On the other hand, quantity-sensitive verbs (in the
absence of certain types of prepositional phrases) are expected to yield a telic
interpretation with a quantized object DP and an atelic interpretation with a non-
quantized object DP. Note that various other combinations were outlined in (9)
above.

Many examples of telicity and the compositionality of telicity employ the
verbs eat and drink. This is reasonable because eat and drink are common verbs
cross-linguistically, which allows for comparisons, and they are quantity-sensitive
while also syntactically flexible. For example, these verbs can be transitive or
intransitive; they can combine with non-quantized DPs for atelic readings or
quantized DPs for telic readings; and they can take an aspectual particle such as
up to obtain a telic reading. This flexibility of eat and drink, make them ideal for

studying telicity across different syntactic conditions. However, it is challenging to

27

identify other verbs with the same range and type of flexibility. It is possible that
this flexibility is an indication that verbs like eat and drink are different from other
quantity-sensitive verbs. Zribi-Hertz (2003) notes that crosslinguistically, the
consumption verbs eat and drink behave in peculiar ways. Consider other
common quantity-sensitive verbs such as build and fix. Like eat and drink, build
and fix can yield telic and atelic VPs when followed by a quantized and non-
quantized DP, respectively, as in (10a) through (10d). Unlike eat and drink, they
cannot be used in an intransitive frame, as seen in (11a) and (11b). Furthermore,
when build or fix are combined with the particle up, the particle does not function
as a resultative particle as it does with eat and drink. The sentence in (12a) is
atelic. There is no logical endpoint, because the particle up with the verb build is
an idiomatic verb-particle construction that means to make something bigger or
stronger. Similarly, the sentence in (12b) does not mean that John finished fixing

the room, it means that he made the room nicer.

(10). a. John built the house. (telic)
b. John built houses. (atelic)
c. John fixed my car. (telic)

d. John fixed cars. (atelic)

(11) a. *John built.

b. *John fixed.

28

(12). a. John built up the wall.

b. John fixed up the room.

This does not mean that all other quantity-sensitive verbs necessarily
behave like build and fix. Verbs can be categorized in several different ways. The
point is that it is difficult to identify other quantity-sensitive verbs that behave the
same way as eat and drink in terms of transitivity, quantized/non—quantized DPs,
and telic readings with the particle up. This suggests that eat and drink may have
some special characteristics not common to all quantity-sensitive verbs, which
may also be partially responsible for the unexpected adult results in van Hout’s
(1998) study.

Within the quantity-sensitive group, we can make the following distinction:
There are quantity-sensitive verbs that allow partitive readings of the DP in the
object position and those which cannot, as illustrated in (13) through (15) below.
The former will be referred to as eat-type verbs and the latter as build-type verbs.
As is seen in (13a) and (13b), eat and drink are quantity-sensitive verbs that can
have partitive interpretations meaning that Logan ate pieces from the cake or
took sips from the milk, without eating any specific amount of the cake or the
whole cake or drinking any specific amount of the milk or the whole gallon of
milk. To illustrate that the preposition of or from does not need to be present in
order to get such and interpretation, consider a situation for (13b), in which I take
the milk out of the refrigerator, open it and say “This smells funny. Do you think

the milk is spoiled?” and someone responds, “Well, Logan drank the milk a little

29

while ago and he feels fine.” Obviously, if there is still milk left that smells funny,
the sentence could not have meant that Logan drank the entire gallon of milk, but
that he drank from the milk. Furthermore, if the time adverbial test of telicity is
applied to (13a) and (13b) as in (14a) and (14b), we see that the adverbial for 2
hours yields an atelic interpretation of the sentence, despite the presence of a

quantized object DP.

(13) 3. Logan ate the cake can be interpreted as Logan ate
ome the cake.

b. Logan drank the milk can be interpreted as Logan
drank of/l’rom the milk.

(14) a. Logan ate from the cake for two hours/“in two hours.

b. Logan drank from the milk for two hours/*in two hours.

In contrast, the sentences in (153) and (15b) demonstrate that the partitive
versions of the sentences with build and fix are ungrammatical and do not mean
that partial building or partial fixing had taken place. The possible partitive
behavior of eat and drink that was demonstrated above, is one of the cross-
linguistic peculiarities of these consumption verbs, as noted by Zribi-Hertz
(2003). For this reason, eat and drink may not be the best representation of a
general category of quantity-sensitive verbs, although they may be very good

representations of partitive quantity-sensitive verbs.

30

(15) a. Logan built the house cannot be interpreted as *Logan built
of/I'mm the house.

b. Logan fixed the ladder cannot be interpreted as *Logan fixed
affirm the ladder.

The partitive interpretations that are possible with eat-type verbs and not
with the others, is what causes us to treat them differently. If listeners have
access to either a partitive or non-partitive interpretation for these verbs, we may
expect variability in adults’ responses to questions containing an eat-type verb
and a quantized object DP. It may yield a telic or an atelic interpretation. This
may help explain the adult’s results in van Hout’s (1998) study. The English-
speaking adults interpreted VPs such as eat his cheese as telic only 25% of the
time and the Dutch-speaking adults 78% of the time. The lower than expected
levels of telic interpretations may be accounted for by partitive interpretations of
eat and drink. There may be other verbs that will also allow partitive and non-
partitive interpretations. However, this study limits the investigation to quantity-
sensitive verbs subdivided into partitive eat-type verbs and non-partitive build-

type verbs, and quantity-insensitive verbs.

Resultative particles

Not all verbs easily become particle verbs. As seen earlier, not all
quantity-sensitive verbs can combine with a purely aspectual particle to obtain a
resultative meaning. Verbs such as build and fix, do not obtain a resultative

meaning with the particle up, whereas others, such as eat and drink do obtain a

31

resultative meaning with the up. See (12a) and (12b) above. For example,
although one can say build up and fix up, these particle verbs are idiomatic,
whereas eat up and drink up are aspectual and specifically force telic
interpretations. With eat-type verbs, the aspectual particle up unambiguously
indicates that all of the object DP must be affected by the verb. Why would we
need an aspectual particle if eat-type verbs and a quantized NP should give us a
telic interpretation in the first place? It seems reasonable that this is because of
the accessibility of partitive interpretations for eat-type verbs. For some
individuals, in certain circumstances, the presence of a quantized object DP after
an eat-type verb may adequately do the work of creating a telic interpretation. If,
however, eat and drink are interpreted as eat unspecified quantities from, and
drink unspecified quantities from, the presence of a quantized object DP after an
eat-type verb, may not necessarily force a telic interpretation of the VP. Consider
the sentence in (16a). It could be interpreted as telic, meaning that the entire
cupcake was consumed, as indicated by the acceptable addition of the adverbial
in 10 minutes. It could also be interpreted as atelic if eat is interpreted partitively,
as indicated by the acceptable addition of the adverbial for 10 minutes. In the
latter case, the sentence would be interpreted as meaning that Marly ate some
unspecified quantities from the cupcake over a 10 minute period. Now consider
(16b) with the resultative particle up and the adverbials in 10 minutes (which is
telic) and for 10 minutes (which is ungrammatical). The resultative particle up

appears to block the partitive/atelic interpretation of the VP or at least

32

disambiguate the interpretation, and force the interpretation that all of the object

DP must be affected for a sentence such as Marty ate up the cupcake to be true

(16) a. Marty ate the cupcake in 10 minutes/ for 10 minutes.

b. Marty ate up the cupcake in 10 minutes/“for 10 minutes.

Thus, the resultative particle up with eat-type verbs, may be an
unambiguous marker of telicity, but because telic interpretations of VPs with eat-
type verbs and quantized DPs can be obtained without it, it is not the only
element of the VP responsible for the telicity. In other words, the resultative
particle up is not responsible for creating the telic interpretation, rather, it may
ensure that the partitive and subsequently atelic interpretation is avoided.

In contrast to the resultative particle up with eat-type verbs, compositional
resultative particles with quantity-insensitive verbs appear to be completely
responsible for telic interpretations. There is a marked contrast between (17a)
and (17b). The first sentence is clearly atelic and the second is clearly telic.
There is little room for other interpretations. There are no differences between
the object DPs in the two sentences, and the only difference is the resultative
particle in (17b) which apparently does all of the work of indicating that the VP is
telic. Unlike the particle up above, the particle here creates the telic
interpretation. To be clear, particle verbs such as eat up and push over are not
equal counterparts to each other. In the case of eat up X, up indicates that there

is a logical endpoint to the eating of X. In the case of push overX, over does not

33

indicate that there is a logical endpoint to the pushing of X. Rather, over indicates

that there is a logical endpoint to the pushing over of X.

17. a) Rob pushed the box.

b) Rob pushed over the box.

In order to compare the contribution of the resultative particle up to telic
interpretations with partitive quantity-sensitive verbs and resultative particles
such as over and out with quantity-insensitive verbs, eat-type verbs (i.e., eat and
drink) will be compared to eat up-type verbs (i.e., eat up and drink up), and push-
type verbs (i.e., push and cany) will be compared to push-over type verbs (i.e.,

push over and carry out).

Quantized DPs: Definite Detenniner vs. Cardinal Number

There are various types of quantizers, such as definite and indefinite
deterrniners, possessive pronouns (in certain contexts), quantifiers, and cardinal
numbers. The present study will include the definite determiner the, and the
cardinal number two. Although DPs headed by a cardinal number are similar to
those headed by a definite determiner, in that they both provide quantity
information, there are important differences in their potential interpretations. A
cardinal number such as two asserts at least two and implies exactly two (see
Horn, 2004). For example, if I ask someone to give me three books and they give

me two books or four books then they have not done as I asked because I

specifically requested three books. If I ask someone to give me the books, they
may give me two, three or four books, depending on the number of relevant
books in the discourse context.

While the quantity of items associated with a definite description in English
is determined, it is determined in the context. The definite determiner picks out
the relevant maximal set in the discourse. Following Heim (1991), the definite
determiner is associated with an existence and uniqueness presupposition. This
is illustrated by Heim’s example in (18) below. The sentence in (18) presupposes
the existence of one unique cat in the domain of the discourse and is therefore
true only if there is exactly one cat and that cat is asleep. If there is exactly one
cat and it is not asleep, then (18) is false. If, however, there are no cats or if there
are two or more cats, then (18) cannot be evaluated as to whether or not it is true
according to the Fregean analysis of definite deterrniners with a singular noun
(as cited in Heim, 1991). This analysis is summarized in (19) below (Heim, 1991,

p. 9).

(18) The cat is asleep.

(19) A proposition in the form [the z] x is:

a) true at index i, if there is exactly one 2 at i , and it is x at i;

b) false, if there is exactly one 2 at i, and it is not x at i; and

c) truth-valueless at index i, if there isn’t exactly one 2 at i.

35

This analysis can be extended to plural definites as well, although in the
case of the plural as in (20), cats is the set of all cats, and the cats is the set of all
possible groupings of more than one cat. The tmth-value of the cats is evaluated

by (21) below (Heim, 1991 , p. 22).

(20) The cats are asleep.

(21) [the z] x is true if the extension of 2 contains the greatest element
and this greatest element is in the extension of x.

In other words, (20) is true if the cats, extends to, and therefore refers to all the
cats relevant to the discourse, and they are all asleep. In situations where the
DP, [the z], is tnith-valueless, as described in (19c) above, listeners can try to
find a way to be able to judge the sentence as true via a process of pragmatic
accommodation (Kadmon, 2001). This is because, depending on the context,
what counts as the domain of the discourse may shift. The crux of the above
discussion is that, although semantically the definite determiner picks out the
maximal set in the discourse that fits a description, listeners are quick to
reevaluate the context in such a that allows them to accept a non-maximal set as
maximal via a process of pragmatic accommodation. Thus, if I say lpicked up
the kids at the park, I do not really mean that I picked up all the kids that were at
the park. The listener will interpret this sentence as meaning that I picked up the
kids that I am responsible for picking up. This is due to a knowledge and/or

presupposition of to whom the kids refers. In other words, the definite determiner

36

is crucially dependent on the discourse context to be interpreted. Furthermore,
when someone says Mary built the houses, in order for this sentence to be true,
we need to see whether Mary built the houses referred to in the discourse or not.
This may depend on some shared knowledge or reference between the speaker
and the listener. The definite determiner picks a set of houses that is salient in
the discourse; and that may be a subset of the houses built.

In contrast to the definite determiner, a cardinal number asserts the
quantity, rather than presupposes that there is a set in the context that fits the
description of a particular DP. Therefore, it is not possible for one to accept the
sentence in (22a) when only two cats are asleep. Furthermore, if a scenario
included three sleeping cats, people are not likely to accept the sentence in (22b)
as an adequate description of it because, if three cats are asleep, although it is
also true that two cats are asleep, cardinal numbers in this context carry the
implicature of exactly three cats. (See Horn, 2004.) Therefore, object DPs with

cardinal numbers would not be expected to lend themselves to accommodation.

(22). a. Three cats are asleep.

b. Two cats are asleep.

In sum, the interpretation of a DP with a definite determiner is much more
dependent on the discourse conditions than other deterrniners, and what counts
as the relevant set in the domain of the discourse will vary from context to

context (see Heim, 1991 ). On the other hand, interpretation of the cardinal

37

number two as at least two does not depend (necessarily) on the context, and
does not vary. In other words, a definite DP may refer to two objects in one
context and three objects in another context; a DP with the cardinal number two
will always refer to a set of at least two objects. The comparison of definite
detenniners to cardinal numbers in the object DP could clarify the role of the
object DP’s syntactic properties on the interpretation of VP telicity, in the absence
of the potential effects of pragmatic accommodation.

There is a second motivation for the comparison of object DPs with
definite deterrniners to those with cardinal numbers that has to do with children’s
abilities in understanding these categories. It is well documented in the literature
that children as old as 6 do not yet have a fully developed determiner system and
have particular difficulties Ieaming all of the subtleties of the definite determiner
(Karrniloff-Smith, 1979; Maratsos, 1976; Munn, Miller and Schmitt, 2006; Perez-
Leroux, Munn, Schmitt and Delrish, 2004; and others). Although children use the
definite determiner early on, there is an extended period of development during
which the system is fine-tuned before it reflects the adult system. One factor that
impacts the difficulty of fully understanding the definite determiner is that its
interpretation is fundamentally discourse-based, as described above. The listener
must decide what constitutes the actual referent that is being talked about by the
speaker within the context. For children, the definite determiner appears to have
broader application than it does for adults (Kanniloff-Smith, 1979 and Perez-
Leroux, et al., 2004). On the other hand, the cardinal number specifies the

quantity of objects being referred to by the speaker, eliminating the need to make

38

a discourse-based decision. Children acquire the ability to count small quantities
using one-to-one correspondence by age 2 and perhaps even earlier (Gelman &
Gallistel, 1978). Wynn (1990) found that by about age 3‘/2 children learn the
cardinal word principle, which is that the last word used in a count indicates the
number of items present. Given the differences in children’s abilities with definite
detenniners and cardinal numbers, the present research also explores whether
children would be better able to use cardinal numbers to understand a DP as
quantized than the definite determiner. The cardinal number is predicted to have
a more effective quantizing role than the definite determiner.

It is important to note that the rationale for the cardinal number yielding
more telic interpretations than the definite number is slightly different for adults
than for children. In both cases, the definite determiner requires reference to the
discourse context. In the case of the adults, the rationale is that they may use
pragmatic accommodation to shift what constitutes the set referred to by the DP
with the definite determiner, and that cardinal numbers do not. In the case of
children, they are anticipated to be less proficient with the definite determiner
overall, specifically with identifying exactly what set the definite DP refers to;

whereas the cardinal number provides a specific and predetermined referent.

39

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

e E8
05 .66 2.5 .66 . E8 “.5 :2...
=2... .66
a: get—Sac ..u.c
r .
33>
...”_._._._._......._.. "an...
258381
— _ H
_
25.2.3... 3.22.5
.3 89:39.62 so 85.5.6 9.230 LEE-=0
. _ J .

 

 

A _

H 22.3.9... .2952 H 7.2.8.0515;

_
m .8235... 95.8.9: 28m 5.2.8. .5 H

 

 

 

 

 

 

 

 

Figure 2. Expanded schematic of basic telicity distinguishing between

partitive and non-partitive quantity sensitive verbs and between cardinal

numbers and definite determiners serving as quantizers.

4O

Specific Variables

Given the issues discussed above, the schematic of linguistic ingredients
that are relevant to telicity can be expanded to include a distinction between
partitive and non-partitive quantity-sensitive verbs and a distinction between
cardinal numbers and the definite determiner as quantizing detenniners, as seen
in Figure 2. The two major independent linguistic variables in the study are verb-
type and determiner-type. In order to investigate the potential impact of specific
verb-subtypes on the interpretation of telicity, the verb types employed in the
studies are quantity-sensitive verbs that do not allow partitive complements,
referred to as build-type verbs (i.e., build and fix); quantity-sensitive verbs that do
allow partitive complements, referred to as eat-type verbs (i.e., eat and drink);
and quantity-insensitive verbs, referred to as push-type verbs (i.e., push and
cany). In order to investigate the impact of resultative particles with partitive
quantity-sensitive verbs and quantity-insensitive verbs, the studies include these
verb types as well, referred to as eat up-type verbs (i.e., eat up and drink up) and
push over-type verbs (i.e., push over and cany out), respectively. In order to
investigate the potential impact of different quantizing detenniners, the
determiner types employed were the definite determiner the and the cardinal

number two.

41

Hypotheses and Predictions
This dissertation had two major aims: (1) To establish the adult levels of
performance in the interpretation of telicity in the contexts of particular verb-
typeslsub-types, object DP types (definite determiner vs. cardinal number), and
resultative particles; and (2) To investigate the development of the interpretation
of telicity as a function of verb-type, object DP determiner-type and resultative

particles by comparing the performance of 3-, 4-,5-, and 6-year-old children.

Task Description

The following brief description of the task is presented here in order to
provide a context for understanding the specific predictions tested. (See Methods
below for details.) The participants viewed short video segments, depicting actors
engaged in situations that either culminated to their logical endpoints, (i.e.,
culminating event; CE), or did not culminate to their logical endpoints, (i.e., non-
culminating event; NCE). NCEs were non-culminating because the actor in the
video discontinued the task prior to its completion. After each video segment, the
participants were asked to respond to a yes/no question such as: Did the man
eat the/two brownies? or Did the man eat up the/two brownies? The responses
following video segments portraying NCEs were of primary interest and relevant
for interpreting the results because these indicate whether the participant
interpreted the verbal stimulus sentence as telic or atelic. If a participant
responded no to a yes/no question following a NCE video segment it is taken to

mean that the VP in the questions was interpreted as a telic VP.

42

Study l: Adults

All of the hypotheses and predictions below refer to responses to yes/no
questions following non-culminating events (NCEs) presented visually. For clarity
and simplicity, the response conditions of verb-type + definite determiner and
verb-type + cardinal number will be referred to as verb-type + def. det. and verb-

type + card. num.

Part A: build-type vs. eat-type verbs.

Hypothesis 1: Not all VPs containing quantity-sensitive verbs and
quantized DPs will be interpreted as telic. Because build-type verbs
cannot be interpreted as partitive and eat-type verbs can be interpreted as
partitive, quantity-sensitive VPs with build-type verbs will be interpreted as
telic, while quantity-sensitive VPs with eat-type verbs may be interpreted

as telic or atelic.

Prediction 1 (Part A: Hyp. 1): Participants will respond no, more
often to questions in build-type + quantized DP conditions than they
will to those in eat-type + quantized DP conditions, regardless of

the determiner-type (i.e., definite determiner vs. cardinal number).

Hypothesis 2: Quantity-sensitive VPs containing DPs quantized by a

cardinal number (i.e., two), will be interpreted as telic, (with a no response)

43

more often than those containing DPs quantized by a definite determiner
(i.e., the) because the cardinal number makes the quantity explicit
whereas the definite determiner is discourse based and can be subject to

pragmatic accommodation.

Prediction 2 (Part A: Hyp 2): Participants will respond no to those
questions in verb + card. num. conditions more than to questions in

verb + def. det. conditions.

Hypothesis 3: Verb-type and determiner-type will interact to produce
different response patterns to questions about NCEs with regard to telicity.
VPs with build-type verbs, with a cardinal number in the object DP, will
induce the most telic interpretations followed by build-type verbs with a
definite determiner, eat-type verbs with a cardinal number, and finally, eat-
type verbs with a definite determiner. This hierarchy is predicted because
of the partitivity of eat-type verbs and pragmatic accommodation of the

definite determiner.

Prediction 3 (Part A: Hyp.3): Participants will respond no to
questions in the build-type + def. det. condition more often than

they will to questions in the eat-type + def. det. condition.

Prediction 4 (Part A: Hyp.3): Participants will respond no to
questions in the build-type + card. num. condition more often than

they will to questions in the eat-type + card. num. condition.

Prediction 5 (Part A: Hyp.3:): Participants will respond no to
questions in the build-type + card. num. condition more often than

they will in the build-type + def. det. condition.

Prediction 6 (Part A: Hyp.3 ): Participants will respond no to
questions in the eat-type + card. num. condition more often than

they will in the eat-type + def. det. condition.

Part B: Eat-type vs. eat up-type verbs and push-type vs. push over-type

verbs.

Hypothesis: With quantity-insensitive verbs, resultative particles are solely
responsible for the telic interpretation of the VP, whereas with quantity-
sensitive verbs that allow a partitive reading, the resultative particle up
serves to disambiguate the partitive vs. non-partitive readings of the verb,
resulting in an unambiguously telic interpretation of the VP. Thus, for
push-type verbs, only the resultative particle, and not the determiner,

contributes to a telic interpretation, while for eat-type verbs, the determiner

45

does contribute to the telicity, while the resultative particle serves to

eliminate the possibility of a partitive interpretation of the verb.

Prediction 1 (Part B): Participants will respond no to questions in
the eat-type + card. num. condition more often than in the eat-type
+ def. det. condition, but not at ceiling levels (NB. This prediction is
the same as prediction 5 for hypothesis 3 above and is repeated

here because of its relevance to this hypothesis).

Prediction 2 (Part B): Participants will respond no to questions in
the eat up-type + def. det. condition more often than in the eat-type

+ def. det. condition.

Prediction 3 (Part B): Participants will respond no to questions in
the eat up-type + card. num. condition more often than in the eat-

type + card. num. condition.

Prediction 4 (Part B): Participants will respond yes to questions in

the push-type + def. def. and push-type + card. num. conditions

equally, indicating an atelic interpretation in both conditions.

46

Prediction 5 (Part B): Participants will respond no to questions in
the push over-type + def. det. condition more often than in the

push-type + def. det. condition.

Prediction 6: Participants will respond no to questions in the push
over-type + card. num. condition more often than in the push-type +

card. num. condition.

Study ll: Children

Part A: Build-type vs. eat-type verbs. Most of the hypotheses and
predictions for Study lI-Part A are the same as those for the Study l-Part A.
Given the developmental dimension of the child study, the specific hypotheses

and predictions for children are re-stated, re-numbered and expanded here.

Hypothesis 1: Not all VPs containing quantity-sensitive verbs and
quantized UPS will be interpreted as telic. Because build-type verbs
cannot be interpreted as partitive and eat-type verbs can be interpreted as
partitive, quantity-sensitive VPs with build-type verbs will be interpreted as
telic, while quantity-sensitive VPs with eat-type verbs may be interpreted

as telic or atelic.

47

Prediction 1 (Hyp. 1 for Part A): Participants will respond no more
often to questions in the build-type +quantized DP conditions than
to those in the eat-type + quantized DP conditions, regardless of

the determiner-type (i.e., definite vs. cardinal).

Hypothesis 2: Quantity-sensitive VPs containing DPs quantized by a
cardinal number (i.e., two) will be interpreted as telic more often than
those containing DPs quantized by a definite determiner (i.e., the)
because children have difficulties with the definite determiner, which is
discourse-based. whereas cardinal numbers are explicit and not

discourse-based.

Prediction 2 (Hyp. 2 for Part A): Participants will respond no to
those questions in verb + card. num. conditions more often than to

questions in verb + def. det. conditions.

Hypothesis 3: Older children will demonstrate greater sensitivity to verb-

type and determiner-type with regard to telicity than younger children.

Prediction 3 (Hyp.3 for Part A): Older children will respond no to

questions about NCEs more often than younger children.

48

Hypothesis 4: Verb-type and determiner-type will interact to produce
different response patterns to questions about NCEs with regard to felicity.
VPs with build-type verbs, with a cardinal number in the object DP, will
induce the most telic interpretations followed by build-type verbs with a
definite determiner, eat-type verbs with a cardinal number, and finally, eat-
type verbs with a definite determiner. This hierarchy is predicted because
of the partitivity of eat-type verbs and children’s difficulties with the definite

determiner.

Prediction 4 (Hyp. 4 for Part A): At each age level, participants will
respond no to questions in the build-type + def. det. condition more

often than to questions in the eat-type + def. det. condition.

Prediction 5 (Hyp. 4 for Part A): At each age level, participants will
respond no to questions in the build-type + card. num. condition

more often than to questions in the eat-type + card. num. condition.

Prediction 6 (Hyp 4 for Part A): At each age level, participants will

respond no to questions in the build-type + card. num. condition

more often than in the build-type + def. det. condition.

49

Prediction 7 (Hyp. 4 for Part A): At each age level, participants will
respond no to questions in the eat-type + card. num. condition more

often than in the eat-type + def. det. condition.

Part B: Eat-type vs. eat up-type verbs and push-type vs. push over-type

verbs.

Hypothesis 1: Resultative particles entail an endpoint, which is compatible
with telic interpretations of VPs, whereas detenniners have multiple
functions and do not necessarily entail telic interpretations. Therefore,
when presented with eat-type VPs, children will be more sensitive to the
presence of resultative particles for the interpretation of VPs as telic, than
they will to the presence of a definite determiner or a cardinal number in
the object DP. This hypothesis is addressed by predictions 2 and 3 of

Hypothesis 2 below.

Hypothesis 2 and its predictions are essentially the same as the
hypothesis and predictions for the Adult Study B, but also include a provision to

examine the predictions at each age level.

Hypothesis 2: WIth quantity-insensitive verbs, resultative particles are

solely responsible for the telic interpretation of the VP, whereas with

quantity- sensitive verbs that allow partitive readings, the resultative

50

particle up serves to disambiguate the partitive vs. non-partitive readings
of the verb, resulting in an unambiguously telic interpretation of the VP.
Thus, for push-type verbs, only the resultative particle, and not the
determiner, contributes to a telic interpretation, while for eat-type verbs,
the determiner does contribute to the interpretation, while the resultative

particle eliminates the possibility of a partitive interpretation of the verb.

Prediction 1 (Hyp.2 for Part B): At each age level, the participants
will respond no to questions in the eat-type + card. num. condition
more often than in the eat-type + def. det. condition, but not at
ceiling levels (NB. This prediction is the same as prediction 6 in
Child Study A presented earlier; it is repeated here because of its

relevance to this hypothesis).

Prediction 2 (Hyp. 1 and 2 for Part B): At each age level, the
participants will respond no to questions in the eat-up type + def.

det. condition more often than in the eat-type + def. det. condition.

Prediction 3 (Hyp.1 and 2 for Part B): At each age level, the
participants will respond no to questions in the eat up-type + card.
num. condition more often than in the eat-type + card. num.

condition.

51

Prediction 4 (Hyp. 2 for Part B): At each age level, the participants
will respond no to questions in the push-type + def. det. and push-

type + card. num. conditions with similarly low frequencies.

Prediction 5 (Hyp 2 for Part B): Participants will respond no to
questions in the push over-type + def. det condition more often than

in the push-type + def. det condition.

Prediction 6 (Hyp. 2 for Part B): Participants will respond no to

questions in the push over-type + card. num. condition than in the

push-type + card. num. condition.

52

CHAPTER 3

Methods

This project was comprised of two studies. The first study examined the
interpretation of telicity by adults and the second one used similar materials and
methods to examine the interpretation of telicity by children. Prior to all pilot
studies and data collection this project was approved by the Michigan State
University Committee on Research Involving Human Subjects (UCRIHS,
currently referred to as the Institutional Review Board or IRB), as project number
04-825. Each of the adult participants signed an informed consent form. A parent
or legal guardian of each of the child participants signed an informed consent
form and all of the children for whom data was collected also gave their verbal

assent for participating in the research activities.

Study l: Adult Interpretations of Telic and Atelic VPs
Participants
The participants were 48 college-age monolingual speakers of
mainstream American English who were students in undergraduate introductory
speech-language pathology or linguistics courses and did not have any reported
history of speech, language, or hearing impairments. The students were
compensated for their participation with extra-credit points in their course. Thirty-

eight of the participants were female and 10 were male.

53

Materials

Linguistic stimuli. Three verb-types were examined in the study: a)
quantity-sensitive verbs, referred to as build-type verbs (build/fix); b) partitive
quantity-sensitive verbs, referred to as eat-type verbs (eat/drink); c) and quantity-
insensitive verbs, referred to as push-type verbs (push/cany). The partitive
quantity-sensitive verbs and the quantity-insensitive verbs were also examined
within a resultative particle verb condition and were referred to as eat up-type
verbs and push over-type verbs (eat up/drink up and push over/cany out),
respectively. The verbs selected for the study were chosen because they met the
semantic and syntactic requirements of the study, are short (1 or 2 syllables), are
common and familiar to both adults and young children, their meanings can
easily be portrayed in video, and they could be acted out relatively quickly and
unambiguously. All of the verbs selected had been identified as being understood
and/or used by over 75% of 30-month-old children (Dale & Fenson, 1996).

Two types of detenniners were examined in this study: the definite
determiner the and the cardinal number two. These were selected because the
definite determiner requires that the DPS referent be identified in the discourse
by the listener, whereas the cardinal number specifies an exact amount.

The target linguistic stimuli were presented to the participants in a
modified tnIth-value judgment task consisting of yes/no questions composed of a
particular verb-type and a particular determiner-type. Each specific verb and
determiner-type combination had two video segment types associated with it.

One was a non-culminating event (NCE) video and the other was a culminating

event (CE) video. All 20 NCE videos were shown to the participants and 10 of the
CE videos, which served as controls, were shown. By design, this resulted in half
of the target questions being asked twice, but in a different video event condition
each time. The questions took the form of: Did the man/Woman (verb) the/two
(objects). See Table 3 for a synopsis of all of the verb + object DP combinations
used.

Filler items were also included in order to decrease the participants’ ability
to discern the specific purposes of the task. The target-question-to- filler-question
ratio 1:3. The filler questions were either about the action portrayed in the video
segment or about the actors’ appearance. In sum, there were a total of 30
experimental questions (20 target questions and 10 controls) and 90 filler
questions, for a total of 120 questions. For a complete list of the target questions,
see Appendix A.

Visual Stimuli. The visual stimuli consisted of video segments in which
actors engaged in particular activities involving two identical objects. Each video
represented the VPs in the target linguistic stimuli or similar but unrelated VPs
(video segments that served as fillers). The use of video segments rather than
static pictures was particularly important for representing the actions associated
with verbs, thereby decreasing any speculation on the participants’ parts as to
exactly how the event proceeded. The video segments were not be accompanied

by any narrative as particular semantic or syntactic content of a narrative could

55

Table 3: Verbs and Object DPs Included in the Experimental Task, by Verb-type

and Detenniner-type

 

 

Verb-Type Verb Definite Detenniner Cardinal Number
Non-partitive build the houses two houses
quantity-sensitive fix the dolls two dolls
Partitive quantity- eat the brownies two brownies
sensitive drink the sodas two sodas

push the dogs two dogs
Quantity-insensitive

cany the bags two bags
Partit' e uant' -

w q 'ty eat up the brownies two brownies

sensitive with
drink up the sodas two sodas
resultative particle
uanti -insensitive
Q ty push over the dogs two dogs
with resultative
cany out the bags two bags

particle

Note. Although soda is generally considered to be a mass noun, pilot testing indicated that,

overall, speakers in Michigan accept soda as a count noun.

56

potentially alter participants perceptions of the events they observed. These
methodological considerations helped to ensure that the participants’ responses
to the experimental linguistic stimuli would be based only on their interpretation of
the scene and the semantic and syntactic characteristics of the experimental
linguistic stimuli.

There were two video segments created for each verb in each syntactic
condition. In the CE video segment, the actor engaged in an action until it
culminated in its logical endpoint with both objects (e.g. both houses were
completely built; both brownies were completely consumed). In the NOE video
segment, the actor engaged in the same action but completed it with only one
object. He/she began the action with the second object but abandoned it before it
culminated in its logical endpoint (e.g., one house was completely built and the
other was left with only 2 walls completed; one brownie was completely
consumed and the other had only one bite taken out of it).

Each video segment ended after the actor had left the scene and the
screen showed the result of the events. For example, the final scene for the
videos depicting ‘house building’ events was either two completed houses on a
table or one completed house and two connected walls and the remaining parts
of the second house. The final scene for the ‘brownie eating’ events was either
an empty plate or a plate with a partially eaten brownie on it. The final scene for
“dog pushing’ events was either two dogs laying on the table or one dog laying on

the table and one dog standing on the table. The description of a sample set of

57

video segments is presented in (14a) and (14 b) below. Videos segments that

were longer than 35 seconds were edited to less than 35 seconds in duration.

(14). a. Video Segment 1: Culminating Event
An actor is sitting at a table with two brownies on a plate in front of
him/her (opening). Slhe eats both brownies in their entirety
(middle). The actor gets up and leaves, leaving the empty plate on
the table (close).
b. Video Segment 2:
An actor is sitting at a table with two brownies on a plate in front of
him/her. (opening). Slhe eats one of the brownies in its entirety.
Slhe then eats one bite of the second brownie and puts the
remainder of it on the plate (middle). The actor gets up and leaves,
leaving the plate and the remainder of the second brownie on the
plate (close).

There were a total of 60 video segments, 20 target videos depicting NCE
events, 10 control videos depicting CE events, and 30 filler video segments
depicting similar actions. The presentation order of the videos was quasi-
randomized such that the participants never viewed the representations of the
same verb in the different conditions with fewer than two intervening video
segments, and they never viewed the same actor in two consecutive segments.
To further control for potential order effects there were 3 stimulus presentation

orders.

Procedures
General. The adult research participants were seen in small groups of no
more than 6 participants during a given session. The participants completed the

experimental task in one session. They were seated in a quiet and well-lit

58

classroom setting and viewed the videos on a large projection screen at the front
of the room. The videos were projected from a computer data display unit, using
Windows Media Player. The size of the video projection area was approximately
7 feet by 5 feet. The participants were seated directly in front of the screen, in
two rows of three student desks, at a distance between 10 to 16 feet from the
screen. Participants were asked to remove all items from their desks except a
pencil and a provided scantron response form on which they were to record their
responses.

Experimental task. The experimental task employed was yes/no question
version of the truth-value judgment (TVJ) task, in which the participants watched
video segments depicting either an NCE or a CE video segment. This task was
selected because, in general, the participants’ responses of yes or no would
indicate whether their interpretation of the VP was telic or atelic. Further, it can
be easily used with both adults and children, it has relatively low task demands,
and it is administered quickly. More detailed considerations for choosing this task
will be discussed in more detail below, with reference to Study II.

The participants viewed each video segment. At the end of each one, the
experimenter paused the presentation so that the projection on the screen was
the closing portion of the segment, (i.e., the actor was no longer present and the
scene portrayed the result of the event). After viewing each video segment,
participants were asked 2 yes/no questions about it. One of the questions was
about the scenario depicted (a target question/control or a filler), and one was

about the actors appearance (filler; e.g., Did the man wear a hat?) The order of

59

target/control and filler questions per video was counter-balanced. The
participants were instructed to decide whether their answer to the question was
yes or no, and to completely fill in the corresponding circle on their response
form. The top row of each column of the scantron response form indicated that
the number 1 was yes and number 2 was no. Additionally, 1 = YES and 2 = NO,
was written on the board as a reminder. After all of the participants responded,
the experimenter resumed the task. Prior to each new video segment, the screen
was solid green for two seconds in order to indicate that a new video was about
to begin. The instructions that were read to the participants by the experimenter

are in (15) below.

(15) Instructions for Adults

You will view short video segments of people acting
out simple tasks. Please watch each one carefully. Each
video segment is less than 35 seconds long. I will pause the
video after each segment and ask you two yes/no questions
about it. One of the yes/no questions will be about the
actor’s appearance and the other yes/no question will be
about what the actor did. If you think that the best answer is
yes, indicate yes by completely filling in the number 1 bubble
on your response form. If you think that the best answer is
no, then indicate no by completely filling in the number 2
bubble. Remember that 1 is yes and 2 is no.

If you are not sure how to best answer the questions,
please give the response that seems most natural to you. Do
not leave any items blank. Before each new video segment,
the screen will be green for 2 seconds.

Some of the items are very similar to each other but
are not necessarily identical, so please watch each of the
videos in its entirety.

60

Data Analysis

Preliminary analysis of the verb stimuli. The specific verbs categorized as
belonging to a verb-type, that is eat-type (eat, drink), build-type (build, fix), and
push-type (push, cany), were categorized as such based on their semantic and
syntactic similarities to each other. In addition to their semantic and syntactic
properties, the specific verbs were selected because of their familiarity to children
ages 3 and up (Dale & Fenson, 1996). However, in order to ensure that the verb
pairs could be categorized as a verb-type in the adult language, paired t-tests
were conducted comparing the adults’ responses to the NCE descriptions for
each pair of verbs within a type for each DP syntax condition. For example, the
number of no responses to Did X build the houses? were compared to the
number of no responses to Did X fix the dolls? This was repeated for each verb
pair within a verb type for each of the determiner conditions. There were no
significant differences between any of the verbs that were paired within a verb
type (p<.05; see Appendix B for additional information), which supported the
decision to conduct the subsequent analyses on verb types, rather than on
individual verbs.

Recording, coding, and scoring the responses. As noted previously,
the adult participants recorded their own responses by completing a
scantron form during the experimental task. The response forms were
scored by computer at the Michigan State University Scoring Office and
the data were imported into the SPSS 12.0 statistical package. The data

of interest, specifically the responses to video segments portraying non-

61

culminating events (NCE), were isolated. The data were re-coded so that
all yes responses were coded as 0 and all no responses were coded as 1.
Because there were two verb tokens within each verb-type, it was possible
that a participant could respond one way to one verb token and another way to
the other verb token, even though they were of the same verb-type. The use of
overall proportion scores across subjects would therefore violate the assumption
of independence of responses for the statistical tests used. Therefore, a count
score was calculated for each participant in each sentence condition. The count
scores were based on the number of no responses to the questions for the two
particular verb-type and determiner-type combinations for each sentence
condition. The possible count scores for each verb-type and determiner-type
combination were as follows: a) a score of O, which is 0 no responses; b) a score
of 1, which is 1 no response and 1 yes response; or c) a score of 2, which is 2 no
responses. This is illustrated in the following example. A participant who
responded yes to “Did the man eat the brownies?” and yes to “Did the man drink
the sodas?” will have a count of 0. A participant who responded yes to “Did the
man eat the brownies?” but no to “Did the man drink the sodas?” will have a
count of 1. A participant who responded no to “Did the man eat the brownies?”
and no to “Did the man drink the sodas?” will have a count of 2. The purpose of
using a count score was to control for the possibility that an individual may
respond differently to questions with different verbs within the same verb type.

Using the count score for each participant accounted for the possible variation

62

within each subject and therefore controlled against violating the independence
assumption for the statistical tests.

Specific analyses. The data that were relevant to the research
questions were responses to questions regarding the video segments that
depicted a non-culminating event (NCE). These events were considered
non-culminating because the person in the video has brought the action
on one object to its logical endpoint, but has not brought the action on the
second object to its logical endpoint. When participants responded no to
the target question, their interpretation of the question was taken to be
unambiguously telic and when they responded yes, their interpretation of
the question is taken to be atelic. Therefore, the statistical analyses were
executed on the number of no responses per relevant stimulus condition.
However, when interpreting the results, it is important to keep in mind that
in the case of partitive verbs, a yes response may be a telic response if
the participant interprets the verb, in the present case, eat or drink, as
taking a partitive object complement.

All of the analyses were conducted using repeated-measures analysis of
variance (ANOVA) on the mean counts of no responses across subjects. The
within-subjects factors were: (a) verb-type, and (b) determiner-type. An a priori
decision of a = .05, with two-tailed tests was used for all statistical tests, except
where otherwise noted.

In Part A, the responses to questions with build-type verbs were compared

to those with eat-type verbs, in both determiner-type conditions (i.e., definite

63

determiner and cardinal number). Part B consisted of 2 separate ANOVAs. In the
first, the responses to questions with eat-type verbs were compared to those with
eat up-type verbs, in both determiner-type conditions and in the second, the
responses to questions with push-type verbs were compared to those with push
over-type verbs, in both determiner-type conditions. Planned comparisons were
conducted to address the specific predictions listed above, and the alpha-levels

for those comparisons were adjusted accordingly.

Study ll: Children’s Interpretations of Telic and Atelic VPs

Participants

There were 80 typically developing children who participated in the study.
They were grouped into four groups on the basis of age: a group of 3-year-olds
(n = 15), ranging in age from 3:2 to 3:9, with a mean age of 3:6; a group of 4-
year—olds (n = 24), ranging in age from 3:10 to 4: 7, with a mean age of 4:4: a
group of 5-year-olds (n = 18), ranging in age from 4:11 to 5:9, with a mean age of
5:4: and a group of 6-year-olds (n = 23), ranging in age from 5:10 to 6:10, with a
mean age of 6:2. The children were recruited for participation in the study from
day care centers, preschools and elementary schools in the Greater Flint, and
Greater Lansing, Michigan areas.

All of the child participants were monolingual speakers of mainstream
American English. Children who were speakers of non-mainstream dialects of
English were excluded from the study due to the fact that some minority dialects

make distinctions of aspect and telicity that do not occur in mainstream American

English. For example, African-American English indicates aspect/telicity through
the copula and auxiliary system (e.g., Green, 2002). Different aspectual features
of this ethnic dialect would present a confound for the present research. Children
were identified as speakers of a non-mainstream dialect of American English
either by parent/guardian report or by the judgment of the primary investigator.
There were four children who were ethnically African American and for whom the
primary investigator was unable to make a judgment regarding their dialect
status. In these cases, the Diagnostic Evaluation of Language Variation
Screening Test (DELV — Screening Test; Seymour, Roeper and de Villiers, 2003)
was administered. These children were not included in the study because their
screening score indicated “Strong” or “Some Variation from Mainstream
American English.”

Background and case history information was obtained through parent
questionnaire and interview, when necessary. Based on pre-testing, parent
questionnaire, and/or parent interview, the child participants met the following
criteria: (a) no parental or teacher concerns regarding their development; (b)
passed a pure-tone hearing screening at 25 dB HL for the frequencies of 1000
Hz, 2000 Hz and 4000 Hz; (c) no history of hearing impairment, (d) did not have
otitis media at the time of the study; (e) passed a basic oral mechanism
screening; (f) passed a speech articulation screening: and (9) had no reported
history of neurological impairment, emotional problems, autism spectrum
disorders, or other developmental disabilities. The Pure-tone hearing screening

was conducted using play audiometry and either a DSP Puretone Audiometer by

65

Micro Audiometrics Corp. or a 120 Portable Audiometer by Beltone Special
Instruments. Further, all of the children demonstrated scores that were not more
than 1.00 SD below the mean on the Sentence Structure and Word Structure
subtests of the Clinical Evaluation of Language Fundamentals — Preschool
(CELF - P; Wiig, Secord, and Semel, 1992). For future matching purposes, most
of the children were administered the Peabody Picture Vocabulary Test - Third
Edition (PPVT-lll; Dunn and Dunn, 1997). All of those children who qualified for
the study and who were given the PPVT-lll scored no more than 1.00 SD below
the mean. The PPVT-Ill scores were not used in the inclusion criteria. For
detailed participant information see Appendix C. All pre-testing and data
collection sessions were conducted by a certified speech-language pathologist or
a trained speech-language pathology student under the supervision of a certified

speech-language pathologist.

Materials: Linguistic and Visual Stimuli

The linguistic and visual stimuli for the children were essentially the same
as those used for the adults in Study l, except that the children’s task was made
shorter by excluding most of the filler items and one-half of the CE videos and
their corresponding linguistic stimuli. Pilot testing indicated that longer versions of
the task, including those with filler videos and questions or only filler questions
were too taxing on young children’s attention skills.

The children’s linguistic and visual stimuli consisted of 4 pre-testltraining

items, 20 NCE video segments and corresponding questions, 10 CE video

66

segments and corresponding questions, and 6 filler videos and questions that
were included in order to check that the children were attending to the videos and
questions. The CE video segments and corresponding linguistic stimuli were
decreased in number but still included so that not all of the videos that the
children saw were of NCEs. As in the adult study there were 3 orders of

presentation used in order to control for possible order effects.

Procedures

General. The children participated in between 3 to 5 sessions of 20 to 45
minutes in length to complete the pre-testing and experimental tasks. Children
were given short breaks on an as-needed basis. Most children completed the
experimental task in 2 sessions. All of the experimental sessions took place at a
child-sized table in a room separate from other school activities. Measures were
taken to make sure that the room was as quiet as possible. All pre-testing took
place prior to the experimental task. This gave each child the opportunity to
become familiar with the experimenter before engaging in the experimental task.
The experimental task was administered via a laptop computer with either a 14 or
15 inch monitor using Microsoft WIndows Media Player. The child was seated
directly in front of the computer, approximately 24 inches away from the screen,
for the duration of the task. At the completion of each session, the children
selected a small prize to take with them. All of the experimental sessions were

audio recorded and approximately one-third of them were video recorded as well.

67

Pre-test training Items. Prior to the experimental task there was a training
period for the children during which they watched four training video segments
with content similar to that in the experimental task. After viewing each video,
they were asked a yes/no question (2 with yes answers and 2 with no answers.)
that was unrelated to the experimental items but that is of similar length and
form. The training items were administered to ensure that the children
understood the task and to reduce the likelihood that the children would be
influenced by the yes response bias (Fritzley & Lee, 2003; Wason, 1961) by
making it clear that both yes and no are acceptable responses. Once a child
responded correctly to 4 out of 4 training items, the experimental task was
presented. It a child did not immediately respond correctly to 4 out of 4 training
items, slhe told by the experimenter that yes and no are both acceptable
responses, the correct response(s) to the training question, as well as a models
of appropriate yes and no response were provided. The experimenter had
multiple yes/no questions that could be used for each training video segment if it
was necessary. If a child did not pass with 4 out of 4 conect training items after
presentation of instruction/models three times, the child did not continue in the
study. This training period also allowed the experimenter to confirm that each
child was able to respond to yes/no questions of an appropriate length and form
prior to the experimental task.

Experimental task. The experimental task that was employed was the
yes/no question version of the truth-value judgment (TVJ) task, in which the

participants watched video segments depicting either a CE or an NCE video

68

segment. One reason for having chosen this task is that it can yield information
regarding a child’s knowledge about complex syntactic structures without placing
high task demands on children (Gordon, 1996). This task has been used
successfully to test a wide range of syntactic knowledge in children (e.g., Avrutin
& Wexler, 1999; Crain, 1985, 1991). Other considerations for selecting the TVJ
task include: 1) unlike picture-pointing tasks, it does not require the child to
consciously match a particular grammatical form to a picture, 2) the response
required from the child is madepragmatically appropriate through careful
construction of the visual stimuli, 3) it is amenable to using video segments,
which can portray the dynamic course of events, 4) the response required from
the child is simple and, with only two choices, largely unambiguous, allowing
relatively young children to participate in the task, and 5) the simple response
also facilitates accuracy in data recording.

The children were told that they would watch movies with a puppet named
Henry the Hippo, but that Henry often fell asleep when he watched movies.
Therefore, he would need the child’s help to answer questions about the movies.
Given the fairly large number of experimental items, in order to maintain the
child’s attention to the task a sticker-game was devised to go along with the
experimental task. The questions were written in a small notebook, one per page.
The page was turned after the puppet was awakened, the question was read and
the child responded. If the child found a star sticker after turning the page, helshe
won a sticker to fill in items on a sheet of paper that was given to them. The

stickers were arranged in such a way that the filling up their sticker sheet

69

corresponded with completion of the task. Most children completed the
experimental task in two sessions. If a child did not need a break sooner, the task
was always stopped at the halfway point and continued during the next session.
Occasionally, if a child wanted to continue the task and time permitted, the child
would be given an extended break at the halfway point, such as playing a board
game or reading a book, and then continue the task during the same session.
The child viewed each video segment. At the end of each one, the
experimenter paused the presentation so that the screen displayed the closing
portion of the video segment, (i.e., the actor was no longer present and the scene
portrayed the result of the event). During the segment the experimenter made the
puppet fall asleep. After viewing each video segment, the child and/or the
experimenter woke up the puppet and the puppet read a question from the
notebook and the child responded. Throughout the task the child was frequently
reminded to answer yes or no to the puppet’s questions. After the child
responded, the experimenter resumed the task. Prior to each new video
segment, the screen was solid green for two seconds in order to indicate that a
new video was about to begin. The instructions that were read to the children by

the experimenter are in (16) below.

(16) Instructions for Children

You will watch some short movies on my computer with my
friend, Henry the Hippo. He really needs to watch these
movies, but he has a problem. Henry is always tired and he
falls asleep whenever he watches movies, so he is going to
need your help. If he falls asleep while you are watching a
movie, we’ll wake him up when it is over and he will ask you

70

a question about the movie. Watch the movies carefully and
after Henry asks you a question, you answer yes or no. He
will ask you a question about what the person did and you
will tell him yes or no. Can you do that? (Pause for
response/check for understanding).

We are going to play a game too. See all these
doghouseslkitty baskets? There aren’t any doggies/kitties
there. In the game you can win doggie/kitty stickers to put in
the dog houses/kitty baskets. Do you think you can win
enough stickers to fill up all of the houses/baskets? I think
you can. Here is how we play. I wrote down the questions for
Henry to ask you about the movies. If he misses a movie
because he fell asleep, he can read the questions from this
little book. After you answer the question, we will flip the
page and see if there is a star sticker on the back. If there is
a star there then you get a doggie/kitty sticker to put in one
of the doghouseslkitty baskets. When you fill up the all of the
doghouses/baskets, we will be all done. We will do some
today and some next time. Are you ready? (Pause for
response).

Okay, here we go. You and Henry will watch the movies. It
Henry falls asleep during the movie, he will have to ask you
a question about it when he wakes up. When he asks you a
question, you answer yes or no. Hmm, what are you going to
answer when Henry the Hippo asks you a question? (Pause
for response to check understanding). Okay good. He will
ask you a question about what the person did in the movie.
Before each new movie, the screen will be green. That
means get ready for a new movie. Are you ready to play?

Data Analysis

Recording, coding and scoring the responses. The experimenter

recorded the child’s oral response on a response form during the task,

whenever possible. When the response could not be fully written out

during the task, the experimenter checked the audio tape and recorded

the child’s response verbatim. After the session the experimenter recorded

the child’s responses on a scantron form.

71

Most of the time the children gave simple yes/no responses or
some or a response of a similar form such as yeah and nope. However,
some children sometimes provided qualified yes responses, such as yes,
but only one and yes, but not that one. Some responses were descriptive
responses that essentially indicated a no response, but did not include the
word no, such as one, only one, and not both of them. In order to preserve
as much information as possible in the coding of the data, the data were
initially coded using the following scheme: yes = 1, no = 2, qualified-yes =
3, and descriptive no = 4. Coding of responses as qualified-yes responses
was operationalized as a response in which the word yes is followed by
additional comments from the child that contains an element that
contradicts the yes response. All of the qualified-yes responses that were
produced during the task are listed in Appendix D.

The scantron forms were scored by computer at the Michigan State
University Scoring Office and the data were imported into the SPSS 12.0
statistical package. The data of interest, specifically the responses to
video segments portraying non-culminating events, were isolated. As in
the adult study, the data were re-coded so that all yes responses were
coded as 0 and all no responses were coded as 1. The responses
originally coded as 1 were recoded as 0 (yes). The responses originally
coded as 2, 3, or 4 were recoded as 1 (no). The descriptive no responses
(e.g., only one) essentially explained why the answer was not yes. Clearly

such responses could not be counted as yes responses. The qualified yes

72

responses, presented a problem in that it was not clear whether the child’s
underlying interpretation was yes or no. For the primary data analyses,
responses were coded as no because the qualifying phrases indicate that
the child was not satisfied with his/her yes response as adequately
answering the question and their qualifying comment contained a
contradiction to the word yes. However, it must be acknowledged that in
such cases the children were ambivalent about their response and
although yes may not have been an adequate response, neither was no. It
could be argued that the child was dissatisfied with both a yes or a no
response, but because they answered with a yes, their interpretation was
“closer to yes” and the qualified responses should have been counted as
yes responses. Therefore, the data were re-coded a second time with
original no and descriptive-no responses coded as 1 and yes and
qualified-yes responses coded as O. The same count scores were
calculated for the children as were calculated for the adults. All of the
subsequent analyses were conducted on the count scores.

Specific Analyses. The data that were relevant to the research
questions were responses to questions regarding the video segments that
depicted a non-culminating event (NCE). These events are considered
non-culminating because the person in the video has brought the action
on one object to its logical endpoint, but has not brought the action on the
second object to its logical endpoint. When children responded no to the

target question, their interpretation of the question was taken to be telic,

73

i.e., they interpreted the VP in the question as an event VP and when they
responded yes, their interpretation of the question was taken to be atelic,
i.e., they interpreted the VP in the question as an activity VP. Therefore,
the statistical analyses were executed on the number of no responses per
relevant stimulus condition.

All of the analyses were conducted using repeated-measures analysis of
variance (ANOVA) on the mean counts of no responses across subjects. The
within-subjects factors were verb-type and determiner-type. The between-
subjects factor was age group. An a priori decision of a = .05, with two-tailed
tests, was used for all statistical tests, except where otherwise noted.

In Part A the responses to questions with build-type verbs were compared
to those with eat-type verbs, in both determiner-type conditions (i.e., definite
determiner and cardinal number). Part B consisted of 2 separate ANOVAs. In the
first, the responses to questions with eat-type verbs were compared to those with
eat up-type verbs, in both determiner-type conditions and in the second, the
responses to questions with push-type verbs were compared to those with push
over-type verbs, in both determiner-type conditions. Planned comparisons were
conducted to address the specific predictions listed above, and the alpha-levels
for these comparisons were adjusted accordingly. Two additional sets of
analyses were conducted that were not originally planned. First, in order to
compare all three verb types in the non-resultative particle conditions, a mixed
model 4 (age) x 3 (verb-type) x 2 (determiner-type) mixed model ANOVA was

conducted. Secondly, because arguments can be made for both methods of

74

coding qualified-yes responses, all of the primary ANOVAs were conducted a
second time using the re-coding scheme in which all original no and descriptive-
no responses were coded as 1 and yes and qualified-yes responses were coded

as 0.

Reliability

Inter-judge reliability. The level of inter-judge reliability was determined by
having a second trained individual, who was a graduate student or senior
undergraduate student in speech-language pathology, listen to the audiotapes of
the experimental sessions for two participants in each group and transcribe and
score their responses. The two judges’ scores were then compared item by item.
The inter-judge reliability was 97% agreement.

Test-retest reliability. The level of test-retest reliability was determined by
having 2 children in each age group repeat the experimental task within 2 weeks
of completing the task the first time. Their responses to the questions were
compared item by item. For some of the children, determining agreement was
somewhat complicated by cases of the presence of qualified-yes response at
one time and a clear yes or no response at another time. The same
operationalization was used for determining agreement as was used in the
primary data coding. A score of 1 counted as a yes response and a score of 2, 3,
or 4 was counted as a no response. Therefore, if a child responded to a question
as no on the first occasion and as yes, but only one on the second occasion,

then the responses were counted as being in agreement because both

75

responses would have been recoded as a 1 for the primary data analysis. On the
other hand, if a child responded to a question as yes on first occasion and as
yes, but only one on the second occasion, then the responses were counted as
being in disagreement because the first would have been coded as a 0 for
primary data analysis and the second would have been coded as a 1. Another
example is that if a child answered no on the first occasion and yes, but only one
on the second occasion, it was counted as an agreement. This resulted in a more
conservative estimate of test-retest reliability than the alternative. The item by
item test-retest reliability was calculated at 88% agreement, demonstrating that

the children’s responses were rather stable from one time to the next.

76

CHAPTER 4

Results

Study l: Adults’ Interpretation of Telicity
Part A: Verb-type and Detenniner-type

Study IA compared the roles of two quantity-sensitive verb subtypes, (i.e.,
build-type and eat-type verbs) and determiner-type (i.e., definite determiner and
cardinal number) on the interpretation of telicity. Eat-type verbs can be
interpreted partitively, whereas build-type verbs cannot. Definite detenniners
refer to a specific set in the discourse, but what constitutes a set may be
determined pragmatically, whereas cardinal numbers require the counting of the
number of items referred to by the DP.

The first analysis addressed whether the verb-type and the detenniner-
type influenced the participants’ interpretation of the telicity of the VPs. The data
were analyzed using a repeated-measures 2 (verb-type) X 2 (determiner-type)
analysis of variance (ANOVA). The within-subjects factors were verb-type (i.e.,
build-type and eat-type vers) and determiner-type (i.e., definite determiner and
cardinal number). Table 4 presents the means and standard deviations per
condition for Study IA and Table 5 presents the statistical summary of the results.

An alpha level of .05 was used for all statistical tests, unless otherwise
indicated. Partial Eta squared (n92) is presented as a measure of effect size7. As
shown in Table 5, there was a significant main effect for verb-type, F(1, 47) =

70.68, p < .001. "p2 = .60. This result supports hypothesis 1 because, as

 

7 0,,2 describes the amount of variance accounted for in the sample by a particular independent
variable. It should be noted that r),,2 values for the independent variables may not sum to 1.0.

77

predicted, questions with build-type verbs were interpreted as telic more often
than those with eat-type verbs, as seen in Figure 3. The main effect of
determiner-type was also significant, F(1, 47) = 18.65, p < .001, m,2 = .28. This
result supports hypothesis 2 because, as predicted, the quantity-sensitive VPs
containing DPs quantized by two, were be interpreted as telic more often than
those containing DPs quantized by the, as seen in Figure 4. There was also a
significant interaction between verb-type and determiner-type, F(1, 47) = 14.84,
p<.001, npz =.24 (see Figure 5). In questions containing build-type verbs, the
determiner-type did not have significant impact on the interpretation, but with eat-
type verbs, the cardinal number condition yielded more telic interpretations than

the definite determiner condition.

Table 4: Means and Standard Deviations of the Counts of Telic Interpretations

per VP Condition for Build-type and Eat-type Verbs for Adults

 

 

VP Condition Mean (SD)
Build-type verb + definite determiner 1.94 (.24)
Build-type verb+ cardinal number 2.00 (.00)
Eat-type verb+ definite determiner .85 (.85)
Eat-type verb+ cardinal number 1.42 (.82)

78

Table 5: Summary of ANOVA Results for Adults: Verb-type by Detenniner-type

for Eat-Type and Build-Type Conditions

 

 

2
Source F Hypothesis Error P "D Observed
df df Power
Verb-type 70.68 1 47 <.001 .60 1.00
Determiner-type 18.65 1 47 <.001 .28 .99
Verb-type x Detenniner-type 14.84 1 47 <.001 .24 .96
I__ _ -7- -m# fir i-b-
2

_\
“—5
(”01

0.75

Mean Count of Tellc
Interpretations
.0
U! .3

0.25 '
0 -

 

 

 

 

build-type eat-type

L_._.,- ,7, L

 

Figure 3. Mean counts of telic interpretations by verb-type for adults in build-type

and eat-type conditions.

79

N

1.75
1.5
1.25

 

0.75
0.5
0.25

Interpretations

 

Mean Count of Tellc I

 

O

 

def. det card. num. I

Figure 4. Mean counts of telic interpretations by determiner-type for adults in

build-type and eat-type conditions.

Hypothesis 3 stated that verb-type and determiner-type would interact to
produce different response patterns to questions about non-culminating events
(NCEs) with regard to telicity. It was hypothesized that VPs with build-type verbs
and a cardinal number in the object DP would induce the most telic
interpretations, followed by build-type verbs and a definite determiner, eat-type
verbs and a cardinal number and finally, eat-type verbs and a definite determiner.
A series of planned comparisons for particular interactions was employed to
evaluate this hypothesis and its predictions. The adjusted alpha level for these
comparisons is a = .012. (See Figure 5.) Predictions 1 and 2 respectively, were
that participants would respond no to questions in the build-type + def. det.
condition more often than to questions in the eat-type +def. det. condition and
that participants would respond no to questions in the build-type + card. num.

condition more often than to questions in the eat-type +card. num. condition.

Participants interpreted questions in the build-type + def. det. condition as telic
significantly more often than those in the eat-type + def. det. condition F(1, 47) =
83.61, p < .001, 0,,“ = .64. They also interpreted questions in the build-type +
card. num. condition as telic significantly more often than those in the eat-type +
card. num. condition F(1, 47) = 24.24, p < .001, np2= .34. (See Figure 5.)
Predictions 3 and 4 respectively, were that participants would respond no to
questions in the build-type + card. num. condition more often than in the build-
type + def. det. condition, and that they would respond no to questions in the eat-
type + card. num. condition more often than in the eat-type + def. det. condition.
Participants did not interpret questions in the build-type + card. num. condition as
telic more often than build-type + def. det. condition, F(1,47) = 3.133, p = .083.
Participants interpreted questions in the eat-type + card. num. condition as telic
more often than in the eat-type + def. det. condition, F(1,47) = 17.929, p < .001,
n,2 = .28.

In summary, all of the hypotheses and predictions for Study IA were
supported by the results with the exception of the prediction that build-type +
card. num. condition would yield more telic interpretations than the build-type +
def. det condition. The lack of statistical significance here appears to be due to
ceiling levels of telic interpretations for both of the build-type questions,
suggesting that when the verb is a quantity-sensitive verb that does not allow
partitive readings, the telic interpretation is so strong that any possible

contribution of the determiner may be masked.

81

 

Mean Cound of Tellc
Interpretations

build—type+ eat-type+ build-type+ eat-type+ I
the the two two

Lgi.77.i

 

Figure 5. Mean counts of telic interpretations for verb-type by determiner-type for

adults in eat-type and build-type conditions.

Part B: Resultative Particles

Study lB examined the role of resultative particles on the interpretation of
telicity with quantity-sensitive verbs that allow partitive interpretations and with
quantity-insensitive verbs. In brief, the hypothesis for Study 1B, states that for
push-type verbs, only the resultative particle, and not the determiner, contributes
to a telic interpretation, while for eat-type verbs, the determiner does contribute to
the interpretation, and the resultative particle serves to eliminate the possibility of
a partitive interpretation of the verb. (See Table 6 for the means and standard
deviations per condition.) The data were analyzed using two repeated-measures
2 (verb-type) X 2 (determiner-type) ANOVAs. The data for eat-type vs. eat up-
type and push-type vs. push over-type verbs were analyzed separately because

the relevant comparisons are between each basic verb-type and its resultative

82

particle counterpart in the different determiner-type conditions. In both analyses
the within subjects factors were verb-type (eat-type vs. eat up-type and push-
type vs. push-over type, respectively) and determiner-type (definite determiner
vs. cardinal number). Planned comparisons were performed for each analyses to
test the specific predictions for eat-type vs. eat up-type questions and push-type
vs. push over-type questions. The adjusted alpha level for these comparisons is

a = .016.

Table 6: Means and Standard Deviations of the Counts of Telic Interpretations

per VP Condition for Eat-type, Eat up-type, Push-type and Push over-type Verbs

 

 

for Adults

VP condition Mean (SD)
Eat-type verb+ definite determiner .85 (.85)
Eat-type verb + cardinal number 1.42 (.82)
Eat up-type verb + definite determiner 1.96 (.20)
Eat up-type verb + cardinal number 1.96 (.20)
Push-type verb + definite determiner .06 (.24)
Push-type verb + cardinal number .15 (.35)
Push up-type verb + definite determiner 1.79 (.41)
Push over-type verb + cardinal number 1.88 (.33)

83

Eat-type vs. eat up-type. The comparison of conditions containing eat-type
and eat up-type verbs demonstrated a significant main effect for verb-type,
F(1,47) = 73.65, p < .001, npz = .61. (See Table 7 below for a summary of the
results). Questions with eat up-type verbs were interpreted as telic more often
than those with eat-type verbs (see Figure 6). There was also a significant main
effect for determiner-type, F(1,47) = 17.07, p < .001 ,. hp? = .27. Questions with
cardinal numbers were interpreted as telic more often than those with the definite
determiner (see Figure 7). A statistically significant interaction between verb-type
and determiner-type was also found, F(1,47) = 17.07, p < .001, hp“? = .27. The
interaction between verb-type and determiner-type type is accounted for by eat-
type verbs and the determiner-type condition. VPs containing eat-type verbs +
card. num. were interpreted as telic more often than VPs containing eat-type
verbs + def. det., whereas there was no statistically significant difference
between eat up-type verbs + def. det. and eat up-type verbs + card. num. (See
Figure 8.) The values for these two conditions were identical, which is likely due
to the very near-ceiling level of telic interpretation of eat up VPs, regardless of

the determiner type.

84

Table 7: Summary of ANOVA Results for Adults: Verb-type by Detenniner-Type

for Eat-Type and Eat up-Type Conditions

 

 

Source F Hypothesis Error P "92 Observed
df df Power
Verb-type 73.65 1 47 <.001 .61 1.00
Detenniner-type 17.07 1 47 <.001 .27 .98
Verb-type x Detenniner-type 17.07 1 47 <.001 .27 .98

—‘j
l
l
I
l
|
I

Mean Count of Telic Interpretations

 

 

eat-type

 

eat up-type

 

Figure 6. Mean counts of telic interpretations for verb-type for adults in eat-type

and eat up-type conditions.

85

Mean Count of Telic
Interpretations

 

 

[ def. det. card. num.

Figure 7. Mean count of telic interpretations by determiner-type for adults in eat-

type and eat up-type conditions.

 

l
I
l
l

 

 

 

Mean Count of Tellc Interpretations

eat-type + def. det. eat-type + card. eat up—type + def. eat up-type + card.
num. det. num.

 

 

T
l

 

 

Figure 8. Mean counts of telic interpretations by verb-type and determiner-type

for adults in eat-type and eat up-type conditions.

86

Three of the predictions are relevant to exploring the role of the resultative
particle with quantity-sensitive verbs that can have a partitive interpretation. See
Figure 8 for a graphic display of the means across the relevant experimental
conditions. Prediction 1 was that participants would respond no to questions in
the eat-type verb + card. num. condition more often than in the eat-type verb
+def. det. condition, but not at ceiling levels. This prediction, which was also in
Study IA was supported with a significantly more questions in the eat-type verb +
card. num. condition interpreted as telic than in the eat-type + def. det. condition,

Prediction 2 was that participants would respond no to questions in the eat
up-type verb + def. det. condition more often than in the eat-type verb+ def. det.
condition. The results support this prediction. Participants interpreted questions
in the eat-type + card. num. condition as telic significantly more often than in the
eat-type + def. det. condition, F(1,47) = 84.68, p < .001, :73 = .64.

Prediction 3 was that participants would respond no to questions in the eat
up-type verb + card. num. condition more often than in the eat-type verb + card.
num. condition. The results support this prediction. Participants interpreted
questions in the eat up-type verb + card. num. condition as telic significantly
more often than in the eat-type verb + card. num. condition, F(1,47) = 22.13, p <
.001, 17,2: .32.

Push-type vs. Push over-type. The second analysis in Study IB was the
comparison between conditions containing push-type and push over-type verbs
in the two determiner-type conditions. (See Table 8 for a summary of the results.)

There was a main effect for verb-type F(1, 47) = 845.39, p <.001. "p2 =.95.

87

Questions with push-type verbs were nearly always interpreted as atelic, and
questions with push over-type verbs were nearly always interpreted as telic (see
Figure 9). There was also a main effect for determiner-type, F(1, 47) = 7.23, p =
.01, mg2 =.13. Questions with the cardinal number two were interpreted as telic
more often than those with the definite determiner the (see Figure 10). However,
the low effect size and the co-occurrence of the detenniners in questions that
were othenlvise overwhelmingly interpreted as telic on the one hand and atelic on
the other, because of the verb-types, suggest that this result may be spurious.
(See Figures 9 and 10 below.) There was no interaction effect between verb-type
and determiner-type. The verb-type appears to be essentially responsible for the

interpretation of telicity in this case (see Figure 11).

Table 8: Summary of ANOVA Results for Adults: Push-Type vs. Push over-Type

 

 

2
Source F Hypothesis Error P ’79 - Observed
df at Power
Verb-type 845.39 1 47 <.001 .95 1.00
Detenniner-type 7.23 1 47 .010 .13 .75
Verb—type x Detenniner-type .00 1 47 1.00 .00 .05

88

N

1.75

.3
01

1.25

.0 .0
NPN
01010:

 

O

 

Mean Count of Tellc Interpretations};

 

push-type push over-type

i
I
l
|
i
i
i
I

Figure 9. Mean counts of telic interpretations by verb-type for adults in push-type

and push over-type conditions.

 

 

1.75 w —
1.5 4
1.25

 

 

 

 

i

 

 

0.75 -

 

 

 

0.25 1» —

 

 

 

   

 

def. det. card. num.

 

i Mean Count of Tellc Interpretations

 

i
r

 

Figure 10. Mean counts of telic interpretations by determiner-type for adults in

push-type and push over-type conditions.

89

I
I
I
I
I
I
I
I
I

N I

 

1.75 _..*____,-_,- —— _.--__. .__.._'-___ -___.

—l
'01
l
|
I
I
I
I
I
I
|

1.25 —- ——~ -—A—- ,, #_ __,,___,_,,__g . ._ ,_

 

0.75 —~»—- -—-- ____, . _ .-__.____ _

I
I
I
I

0.25

 

 

 

 

Mean Count of Telic Interpretations I
o .s
' i
I
I

push-type + def. push-type + card. push over-type + push over- +
det. num. def. det. card. num.

 

 

_ W, -_- __. ___,. ., ___ ______I

Figure 11. Mean counts of telic interpretations by verb-type and determiner-type

for adults in push-type and push over-type conditions.

Taken together, predictions 4, 5, and 6 explore the role of the resultative
particle with quantity-insensitive verbs and its independence in establishing a
telic interpretation regardless of the type of determiner. (Refer to Figure 11 above
to graphically compare the conditions.) Planned comparisons were used to test
these predictions. Prediction 4 was that participants would respond yes to
questions in the push-type + def. det. and push-type + card. num. conditions
equally, indicating an atelic interpretation in both conditions. The result supports
this prediction as the difference between responses to push-type + def. det. and
push-type + card. mm, were non-significant. F(1, 47) = 4.27, p = .04, n92 =.08.
Prediction 5 was that participants will respond no to questions in the push over-
type + def. det. condition more often than in the push-type + def. det. condition.

The result supports this prediction as the difference between the two was

90

statistically significant, F(1, 47) = 711.61, p < .001, ",2 =.94. Prediction 6 was that
participants would respond no to questions in the push over-type + card. num.
condition more often than in the push-type + card. num. condition. As predicted,
the difference between the two conditions was statistically significant, F( 1, 47) =
500.44, p < .001, up? =.94.

The purpose of Study 18 was to see whether resultative particles would
behave differently in VPs with partitive quantity—sensitive verbs than in VPs with
quantity-insensitive verbs. When comparing the differences, especially the
degree of differences in telic interpretations, between eat-type verbs with and
without a particle and push-type verbs with and without a particle, we see that
there are qualitative differences. (See Figure 12 which combines the information
from Figures 8 and 11). For the comparison of eat-type vs. eat up-type verbs we
see that the mean number of telic interpretations increases with the card. num.
as compared to the def. det. condition and increases to near-ceiling levels with
the addition of the resultative particle. Although there are differences, the
differences are graded. For the comparison of push-type vs. push over type
verbs we see that the mean number of telic interpretations are at the extremes,
based primarily on the presence or absence of the resultative particle, as

expected.

91

 

 

 

 

 

 

 

 

 

Mean Count of Telic Interpretations

eat- eat— eat up- eat up- push- push- push push
type + type t type + type + type + type + over- over-
def. card. def. card. def. card. type + type +
det. num. det num. det. num. def. card.

det. num.

 

 

I. __ _ _ __.. _____ __ . .____._ ‘____4______u____________ __1_____.___________.._.___.

Figure 12. Mean counts of telic interpretations by verb-type and determiner-type

for adults in eat/eat up-type and push] push over-type conditions.

Study ll: Children’s Interpretation of Telicity

Part A: Verb-type and Detenniner-type

Study "A compared the roles of 2 quantity-sensitive verb subtypes, one
that allows partitive interpretations (eat-type) and one that does not (build-type)
and determiner-type (definite determiner and cardinal number) on the
interpretation of telicity by children ages 3, 4, 5 and 6. The data were analyzed
using a repeated-measures 4 (age) x 2 (verb-type) x 2 (determiner-type) mixed-
model ANOVA. The between-subjects factor was age. The within-subjects
factors were verb-type (i.e., build-type and eat-type verbs) and determiner-type

(i.e., def. det. and card. num.). See Table 9 for the means and standard

92

deviations per condition for build- and eat-type verbs and Table 10 for a

statistical summary of the results.

Table 9: Means and Standard Deviations of the Counts of Telic Interpretations

per VP Condition for Build-type and Eat-type Verbs for Children

 

Mean (SD)

 

VP Condition 3—year-olds 4-year-olds 5-year-olds B-year-olds

 

Build-type+definite determiner 0.93 (.70) 1.29 (.75) 1.33 (.77) 1.35 (.71)

Build-type+cardina| number 0.73 (.96) 1.42 (.77) 1.44 (.70) 1.52 (.73)

Eat.type+definite determiner 0.33 (.62) 0.92 (.88) 1.33 (.84) 1.04 (.71)

Eat-type+cardinal number 1(.85) 1.42(.78) 1.5(.62) 1.57 (.66)

93

Table 10. Summary of ANOVA Results for Children: Eat-Type vs. Build-Type

 

 

Source F Hypothesis Error P ”92 Observed
df df Power
Age 4.34 3 76 .007 .15 .85
Verb-type 2.18 1 76 .114 .03 .31
Verb-type x Age .39 3 76 .762 .02 .12
Detenniner-type 18.21 1 76 .000 .19 .99
Detenniner-type x Age .61 3 76 .614 .02 .17
Verb-type x Detenniner-type 18.53 1 76 .000 .20 .99
2.6 3 76 .059 .09 .62

Verb-type x Detenniner-type x Age

There was no main effect for verb-type, F(1, 76) = 2.18, p =.14, 11,,2 = .03.
This result does not support hypothesis 1. It was predicted that questions with
build-type verbs would be interpreted as telic more often than those with eat-type
verbs. The children did not make a statistically significant distinction between
build-type verbs and eat-type verbs with respect to telicity (see Figure 13). There
was a main effect for determiner-type, F(1, 76) = 18.21, p<.001, r792 = .19. This
result supports hypothesis 2 because, as predicted, the quantity-sensitive VPs
containing DPs quantized by a the cardinal number were interpreted as telic
more often than those containing DPs quantized by the definite determiner, as
seen in Figure 14. There was also a statistically significant interaction between
verb-type and determiner-type, F(1, 76) = 18.53, p < .001, m,‘2 = .20. In questions
containing build-type verbs, the determiner-type did not have significant impact
on the interpretation, but with eat-type verbs, the cardinal number yielded more
telic interpretations than in the definite determiner condition (see Figure 15).
There was also a between-subjects effect for age, F(3, 76) = 4.339, p = .007, m,2
= .15. A post hoc Bonferroni analysis indicated that the 3-year-olds had
statistically significantly fewer telic interpretations than the 5—year-olds (p = .012)
and the 6-year-olds (p = .012). This supports hypothesis 3 which predicted that
older children would respond no to questions about NCEs more often than

younger children.

95

 

 

 

 

 

 

2 1.75
'22 1.5~
”631.25
53 1~
°E
380.7%—
g5 0.5
o.25~

 

 

 

O
I

 

 

 

build-type eat-type

Figure 13. Mean count of telic interpretations by verb-type for children in build-

type and eat-type conditions.

I
I

 

 

 

N

 

 

I
I
I
I
I

15 ._- *— ‘ T

 

 

.A
N
01

0.75 «
0.5 I
0.25 I -

 

 

 

 

 

 

 

O

 

card. num.

Mean Count of Telic lnterpretationsI

 

 

 

 

Figure 14. Mean count of telic interpretations by determiner-type for children in

build-type and eat-type conditions.

 

ii,~_ _. _ fl_%a. _,,- , . T

2
I 1.75 ,
‘ 1.5 1 _ ,, i 7,,
I 125 3 I03 year olds
I , £34 year olds
a 5 year olds
° 5 I ;!§X¢?L°!9:I

MON
01

O
01

 

 

\ ” \ §
build-type + build-type + eat-type+ def. eat-type +
def. det. card. num. det. card. num.

Mean Count of Tellc Interpretations
o _s

 

,__24 __,_>.,.

 

Figure 15. Mean count of telic interpretations by verb-type and determiner-type

for children in build-type and eat-type conditions by age group.

Hypothesis 4 states that verb-type and determiner-type will interact to
produce different response patterns to questions about NCEs with regard to
telicity. VPs with build-type verbs with a cardinal number in the object DP will
induce the most telic interpretations followed by: build-type verbs with a definite
determiner, eat-type verbs with a cardinal number, and finally, eat-type verbs
with a definite determiner. In order to test this hypothesis a series of planned
comparisons were conducted for the specific predictions for each age group. The
adjusted alpha level for each age group is .013. (See Figure.15.)

Prediction 4 was that at each age level, participants would respond no to
questions in the build-type + def. det. condition more often than to questions in

the eat-type + def. det. condition. The results for each of the age groups were as

97

follows: non-significant for 6—year-olds, F(1, 22) = 2.48, p = .13, npz = .10; non-
significant for 5-year-olds, F(1, 17) = .00, p = 1.00, rip2 = .00; non-significant for 4-
year—olds F(1, 23) = 4.97, p = .04, m,2 = .18; and significant for 3-year-olds, F(1,
14) = 9.95, p = .007, up? = .42. The prediction was supported only for the
youngest children.

Prediction 5 was that at each age level, participants would respond no to
questions in the build-type + card. num. condition more often than to questions in
the eat-type + card. num. condition. The results for each of the age groups were
as follows: non-significant for 6-year-olds, F(1, 22) = .14, p = .71, npz = .01; non-
significant for 5-year-olds, F(1, 17) = .14, p = .72, m,2 = .01; non-significant for 4-
year-olds F(1, 23) = .00, p = 1.00, n92 = .00; and non-significant for 3-year-olds,
F(1, 14) = 1.67, p = .22, m} = .11. The prediction was not supported any of the
age groups.

Prediction 6 was that at each age level, participants will respond no to
questions in the build-type + card. num. condition more often than in the build-
type + def. det. condition. The results for each of the age groups were as follows:
non-significant for 6-year-olds, F(1, 22) = 1.65, p = .21, rip2 = .07; non-significant
for 5-year—olds, F(1, 17) = 1.00, p = .33, n92 = .06; non-significant for 4—year-olds
F(1, 23) = .68, p = .48, lip2 = .03; and non-significant for 3-year-olds, F(1, 14) =
.81, p = .38, m,2 = .06. The prediction was not supported for any of the age
groups.

Prediction 7 was that at each age level, participants would respond no to

questions in the eat-type + card. num. condition more often than in the eat-type +

98

def. det. condition. The results for each of the age groups were as follows:
significant for 6-year-olds, F(1, 22) = 14.14, p = .001, npz = .39; non-significant for
5-year-olds, F(1, 17) = 1.00, p = .33, m,2 = .06; significant for 4-year-olds F(1, 23)
= 13.80, p = .001, r792 = .38; and significant for 3-year-olds, F(1, 14) = 17.50, p =
.001, 17,} = .56. The prediction was supported for all of the age groups except the
5—year-olds.

Overall, the analyses addressing hypothesis 4 found that for 3-, 4-, and 6-
year—olds, VPs with eat-type verbs + card. num. were interpreted as telic more
often than VPs with eat-type verbs + def. det. and for 3-year-olds, VPs with build-
type verbs + def. det. were interpreted as telic more often than VPs with eat-type
verbs + def. det.

Part B: Resultative Particles

Study IIB examined the role of resultative particles on the interpretation of
telicity with quantity-sensitive verbs that allow partitive interpretations (eat— and
eat up-type verbs) and with quantity-insensitive verbs (push— and push over-type
verbs) in children ages 3, 4, 5, and 6. Hypothesis 1 for Study B essentially states
that for push-type verbs, only the resultative particle, and not the determiner,
contributes to a telic interpretation, while for eat-type verbs, the determiner does
contribute to the interpretation, and the resultative particle serves to eliminate the
possibility of a partitive interpretation of the verb. (See Table 11 for the means
and standard deviations per condition for each age group.) The data were

analyzed using two repeated-measures 4 (age) X 2 (verb-type) X 2 (determiner-

99

type) mixed-model ANOVAs. The data for eat-type vs. eat up-type and push-type
vs. push over-type were analyzed separately because the relevant comparisons
are between each basic verb-type and its resultative particle counterpart in the
different determiner-type conditions. In both analyses the between subjects factor
was age and the within subjects factors were verb-type (eat-type vs. eat up-type
and push-type vs. push-over type, respectively) and determiner-type (definite
determiner vs. cardinal number). Planned comparisons were performed for each
analysis to test the specific predictions for eat-type vs. eat up-type questions and
push-type vs. push over-type questions. For the eat-type vs. eat up-type
conditions, the adjusted alpha level for multiple tests per age group is .01 and for
the push-type vs. push up-type analysis it is .02 (.016 unrounded).

Eat-type vs. eat up-type. The results for the analysis comparing the
conditions containing eat-type and eat up-type verbs are summarized in Table 12
below. There was a main effect for verb-type, F(1,76) = 10.33, p = .002, rip2 ==
.12. Questions with eat up-type verbs were interpreted as telic more often than
those with eat-type verbs (see Figure 16). There was also a main effect for
determiner-type, F(1,76) = 40.03, p<.001,. 0.32 = .35. Questions with cardinal
numbers were interpreted as telic more often than those with the definite
determiner (see Figure 17). There was also a between-subjects effect for age,

F(3, 76) = 6.34, p=.001. A post hoc Bonferroni analysis indicated that the

100

Table 11: Means and Standard Deviations of the Counts of Telic Interpretations

per VP Condition for Eat-type, Eat up—type, Push-type and Push over-type Verbs

 

 

 

for Children
Means (SD)
VP Condition 3-year-oids 4-year-oids 5-year-olds 6-year—olds

Eat-type verb +
definite determiner 0.33 (.62) .91 (.88) 1.33 (.84) 1.04 (.71)
Eat-type verb +
cardinal number 1 (.85) 1.42 (.78) 1.5 (.62) 1.57 (.66)
Eat up-type verb +
definite determiner .53 (.83) 1.29 (.86) 1.44 (.78) 1.65 (.71)
Eat up—type verb +
cardinal number 1.00 (.85) 1.54 (.66) 1.67 (.69) 1.74 (.62)
Push-type verb +
definite determiner .53 (.74) .79 (.78) .83 (.79) .70 (.63)
Push-type verb +
cardinal number 1.13 (.83) 1.21 (.88) 1.00 (.84) 1.13 (.87)
Push up-type verb +
definite determine, -80 (.86) 1.38 (.71) 1.22 (.73) 1.39 (.78)
Push over-type verb +

.80 (.86) 1.33 (.82) 1.39 (.70) 1.52 (.73)

cardinal number

101

responses of 3-year-olds were significantly different than those of the 4-year-olds

(p = .03), 5-year-olds (p = .002) and 6-year-olds (p = .001). There was no

statistically significant interaction between verb-type and determiner-type or

between age and the within subjects factors.

Table 12: Summary of ANOVA Results for Children in Eat-type and Eat up-type

 

 

Conditions
Hypothesis Error 2 Observed
Source F df df '79 Power

Age 6.34 3 76 .001 .20 .96
Verb-type 10.33 1 76 .002 .12 .89
Verb-type x Age .93 3 76 .432 .04 .25
Detenniner-type 40.03 1 76 .000 .35 1 .0
Determiner-type x Age 1.60 3 76 .196 .06 .41
Verb-type x Detenniner-

3.43 1 76 .068 .04 .44
type
Verb-type x Detenniner-

.85 3 76 .473 .03 .23

type X Age

102

I
I
l
I
I
I
I
I
I

 

—‘.. ________,

 

 

 

 

 

 

 

IMean Count of Tellc Interpretationst

 

[m

Figure 16. Mean counts of telic interpretations by verb-type for children in eat-

type and eat up-type conditions.

I
I
I
I

 

N

 

1,75 «I _--_-_,____,V . . __

 

 

.3
0'!
I
I

 

 

1.25 +9 __- w

 

0.75 fl-

.0
at
?

I

 

0.25 ~—-

 

 

 

 

O

 

 

def. det.

 

'Mean Count of Tellc Interpretations

 

I
I
I
I
I
I
I
I
I

Figure 17. Mean counts of telic interpretations by determiner-type for children in

eat-type and eat up-type conditions.

103

Three of the predictions are relevant to exploring the role of the resultative

particle with quantity-sensitive verbs that can have a partitive interpretation. See

Figure 18 for a graphic display of the means across the experimental conditions

by age. Planned comparisons were used to evaluate the predictions. The

adjusted alpha level for multiple tests per age group is .01 (Three tests for

Hypothesis 1 and one for Hypothesis 2).

 

Mean Count of Tellc Interpretations ,

0.75
0.5

0.25 .

 

x‘

 

eat-type+ def.
det.

   

  
         
     

///

/
2f

eat-type +
card. num.

I IIZI 3 year oldsI

I I4 year olds ;
; E 5 year olds

: I l 6.192095

‘ I

eat up-type + eat up-type +
def. det. card. num.

 

Figure 18. Mean count of telic interpretations by verb-type and determiner-type

for children in eat-type and eat up-type conditions by age group.

Prediction 1 was that at each age level, participants would respond no to

questions in the eat-type verb + card. num. condition more often than in the eat-

type verb + def. det. condition, but not at ceiling levels. This prediction was also

in Study A above and it is repeated here because of its relevance to evaluating

the relative contributions of the determiners as compared to the resultative

104

particle. The results for each of the age groups were as follows: significant for 6-
year-olds, F(1, 22) = 14.14, p = .001, n92 = .39; non-significant for 5-year-olds,
F(1, 17) = 1.00, p = .33, n,2 = .06; significant for 4-year—oids F(1, 23) = 13.80, p =
.001, hp? = .38; and significant for 3-year-olds, F(1, 14) = 17.50, p = .001, hp? =
.56. The prediction was supported for all of the age groups except the 5—year-
olds.

Prediction 2 was that at each age level, participants would respond no to
questions in the eat up-type verb + def. det. condition more often than in the eat-
type verb + def. det. condition. The results for each of the age groups were as
follows: significant for 6—year-olds, F(1, 22) = 16.33, p = .001, npz = .43; non-
significant for 5-year-olds, F(1, 17) = .21, p = .65, 1192 = .012; significant for 4-
year-olds F(1, 23) = 8.07, p = .009, a: = .26; and non-significant for 3-year-oids,
F(1, 14) = .81, p = .38, m,2 = .06. The prediction was supported for 6-year-olds
and 4-year-olds, but not for the 3-year-olds and 5-year-olds.

Prediction 3 was that at each age level, participants would respond no to
questions in the eat up-type verb + card. num. condition more often than in the
eat-type verb + card. num. condition. The results for each of the age groups were
as follows: non-significant for 6-year-oids, F(1, 22) = 1.65, p = .213, m} = .07;
non-significant for 5-year-olds, F(1, 17) = 1.00, p = .33, rip2 = .06; non-significant
for 4-year-olds F(1, 23) = .52, p = .48, rip2 = .02; and non-significant for 3—year-
olds, F(1, 14) = .00, p = 1.00., r7,,,2 = .00. The prediction was not supported for any

of the age groups.

105

These two last predictions also address the second hypothesis for Study
"3, which was that children would be more sensitive to the presence of
resultative particles for the interpretation of VPs as telic, than they would to the
presence of a cardinal number or definite determiner in the object DP. The
results for predictions 2 and 3 above indicate that this is not the case. Although
the 4- and 6—year olds had more telic interpretations in the particle condition than
in the def. det. condition alone, none of the groups had an advantage in the
resultative particle condition over its corresponding card. num. condition.
Although the difference was non-significant for the 3-year-olds is it noteworthy
that the numerical difference in the means was in the opposite direction than
predicted. Overall, the results indicate that the resultative particle may have an
advantage over the definite determiner alone for obtaining telic responses; there
is no evidence that it has an advantage over the cardinal number. The resultative
particle did not appear to be an earlier or better indicator of telicity than the
cardinal number, although it may be better than the def. det.

Push-type vs. push over-type. The second analysis in Study B was the
comparison between push-type and push over-type verbs in the two determiner-
type conditions. The results for this analysis are in Table 13 below. There was a
main effect for verb-type F(1,76) = 28.48, p <.001, rjp2 =.27. Questions with push
over-type verbs were interpreted as telic significantly more often than questions
with push-type verbs (see Figure 19). There was also a main effect for
determiner-type, F(1, 76) = 14.48, p < .001, npz =.16. Questions with the cardinal

number two were interpreted as telic more often than those with the definite

106

Table 13: Summary of ANOVA Results for Children for Push-type vs. Push over-

 

 

type Verb conditions
Hypothesis Error 2 Observed

Source F df df p ”p Power
Age 1.174 3 76 .325 .044 .304
Verb—type 28.484 1 76 .000 .273 1.00
Verb-type x Age 3.873 3 76 .012 .133 .805
Detenniner-type 14.475 1 76 .000 .160 .964
Detenniner-type x Age .288 3 76 .834 .011 .103
verb'type x Detem‘imr' 11.648 1 76 .001 .133 .921
type
ve'b'type x Determine" 1.5 3 76 .221 .056 .381
type x Age

107

 

1.75 -0 770* ---~ ______._- ,-,#-_~M#_ _._-,. -...-___ ___
1.5 1
1.25 ~

 

 

0.75 ..-- -
0.5 -
0.25 1

 

 

 

 

 

  

 

Mean Count of Tellc Interpretations

 

 

push over-type

I
I
I
I
I
I

Figure 19: Mean count of telic interpretations by verb-type for children in push-

type and push over-type conditions.

 

 

 

I
I

N

 

1.75 - U-A_-__. a-

 

 

.

 

1.25 ~

 

 

 

.0
.ON
0101

 

 

 

 

 

def. def. card. num.

 

I Mean Count of Telic Interpretations!

 

Figure 20. Mean counts of telic interpretations by determiner-type for children in

push-type and push over-type conditions.

108

determiner the. (See Figure 20.) There was interaction effect between verb-type
and determiner-type, F(1, 76), p = .001, rho2 = .13. For questions with push-type
verbs, there are more telic interpretations in the card. num. condition than in the
def. det. condition, whereas for questions with push-over type verbs, the number
of telic interpretations are similar regardless of the determiner-type condition.
This result is particularly interesting because few telic interpretations were
expected for push-type questions regardless of the determiner-type, but this
expectation was not realized (see Figure 21). For further exploration of this result
see the discussion. There was no between-subjects effect for age, but there was
and interaction effect for verb-type x age, F(3, 76) = 3.873, p = .012, n92 = .13.
Children ages 4, 5, and 6 had more telic interpretations for all questions with
push over-type verbs than for questions with push-type verbs, whereas 3-year-
old children did not.

Predictions 4, 5, and 6 of Hypothesis 1 are relevant to exploring the role of
the resultative particle with quantity-insensitive verbs and its independence in
establishing a telic interpretation regardless of the type of determiner. (See
Figure 21 .) Planned comparisons were used to test these predictions. Recall that

the adjusted alpha level for multiple tests was a = .016.

109

 

.....
"n

3Eriolds—I
I4 year olds
‘ E 5 year olds
L Ixearlidi

 

     

push-type + push-type + push over— push over-
def. det. card. num. type + def. type + card.
det. num.

Mean Count of Telic Interpretations

 

Figure 21. Mean count of telic interpretations by verb-type and determiner-type

for children in push-type and push over-type conditions by age group.

Prediction 4 was that at each age level, participants would respond yes to
questions in the push—type + def. det. and push-type + card. num. conditions
equally, indicating an atelic interpretation in both conditions. The results for each
of the age groups were as follows: non-significant for 6-year-olds, F( 1, 22) =
4.87, p = .04, m,2 = .18; non-significant for 5-year-olds, F(1,17) = 1.89, p = .19,
:73 = .10; significant for 4-year-olds, F(1,23) = 4.17, p = .015, npz = .23;
significant for 3—year-olds, F(1, 14) = 13.50, p = .003, I)“,2 = .49. in the Adult Study
part B, the results for this prediction were easy to understand, because there
were very few telic interpretations in both of these two conditions. However, that

is not the case in the children’s data. The statistical analyses indicate that for 3-

110

and 4-year-olds there were significant differences between the push-type + def.
det. and the push + card. num. conditions. The push-type + def. det. condition
yielded more atelic responses than the push-type + card. num. responses. The
analyses indicate that for the 5- and 6-year-olds there were not statistically
significant differences between the two conditions. However, it does not address
the issue of whether the responses indicated telic or atelic interpretations. in fact,
for the 5— and 6-year-old, no more than half of the responses indicated an atelic
interpretation in either determiner-type condition. (See Figure 21 .)

Prediction 5 was that at each age level, participants will respond no to
questions in the push over-type + def. det. condition more often than in the push-
type + def. det. condition. Although all groups responded to the push over-type +
def. det. condition more often, the difference only reached significance for the 4
and 6-year-olds, given the adjusted alpha level. The results for each of the age
groups were as follows: significant for 6—year-olds, F(1, 22) = 22.52, p < .001, np‘z
= .51; non-significant for 5-year-olds, F(1,17) = 4.50, p = .05, m,2 = .21; significant
for 4-year-olds, F(1,23) = 32.20, p < .001, up2 = .58; non-significant for 3-year-
olds, F(1, 14) = 5.09, p = .04, of = .27. The lack of significance for the 3- and 5-
year olds seems to be due to the surprisingly relatively high levels of telic
responses to the push-type + def. det. condition. The prediction was supported
for the 4- and the 6-year-old groups.

Prediction 6 was that at each age level, the participants would respond no
to questions in the push over-type + card. num. condition more often than in the

push-type + card. num. condition. The results for each of the age groups were as

111

follows: significant for 6-year-olds, F(1, 22) = 8.17, p < .009, npz = .27; significant
for 5-year-olds, F(1, 17) = 7.37, p = .015, m,2 = .30; non-significant for 4-year-
olds, F(1, 23) = .52, p < .48, m,2 = .02; and non-significant for 3-year—olds, F(1,
14) = 3.18, p = .10, 17‘,2 = .19. In terms of the mean number of telic
interpretations, the 3-year-olds, answered no more often in the push-type + card.
num. condition than in the push over-type + card. num. In general the prediction
was supported for the 5- and the 6-year-old groups, but not for the 3- and 4-year-

old age groups.

Follow-up Data Analyses
Build-type vs. push-type verbs.

In the adult studies, it is obvious from comparing the mean counts of telic
interpretations for the quantity-sensitive build-type verbs (Figure 3) and those for
the quantity insensitive push-type verbs (Figure 9), that the adult participants
treated these two verb-types very differently with respect to telicity. VPs with
build-type verbs received telic interpretations and push-type verbs received atelic
interpretations. In the children’s studies, this is not obvious (see Figures 13 and
19). In order to determine whether the children treated the three main verb types
as different types, a post hoc 3 (verb-type) x 2 (determiner-type) x 4 (age) mixed-
model ANOVA was performed on the stimuli containing build-type, eat-type and

push—type verbs. (See Table 14 for a statistical summary of the results.)

112

Table 14: Summary of Results for Children for Build-type vs. Eat—type vs. Push-

 

 

type Verb Conditions.
Hypothesis Error 2 Observed

Source F df df p ”P Power
Age 2.43 3 76 .072 .089 .585
Verb-type 10.918 2 75 .000 .225 .989
Verb-type x Age 1.666 6 152 .133 .062 .622
Detenniner-type 39.793 1 76 .000 .344 1 .000
Detenniner-type x Age 1.164 3 76 .329 .044 .301
ve'b‘type x Detem‘iner' 9.168 2 75 .000 .196 .972
time
ve'b'type x Determine" 1.264 6 152 .277 .048 .486
type x Age

113

There was a main effect for verb-type, F(2, 75) = 10.918, p < .001, n,,2 =

.23. Earlier analysis indicated no difference for children between build-type verbs
and eat-type verbs. Comparison between build-type and push-type verbs
indicated a significant difference between these two verb-types, F(1,76) = 22.05,
p < .001, rip2 = .23. Comparison between eat-type and push-type verbs also
indicated a significant difference, F(1, 76) = 7.076, p = .01, rip2 = .085. Thus, as a
group, the children did not make a distinction between the two quantity-sensitive
verb types, but did make a distinction between the quantity-insensitive verb-type

and the two quantity-sensitive verb types. See Figure 22.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

 

L build-type push-type

Figure 22. Mean counts of telic interpretations by verb-type for children in build-

type, eat-type, and push-type conditions.

114

Analyses of Recoded Data

As indicated in the methods section, the primary data analyses above
were performed on the data coded with the following data scheme: Responses of
yes were counted as atelic interpretations, responses of no were counted as telic
interpretations, qualified yes responses, such as yes, but only one and
descriptive responses such as only one or not that one were also counted as telic
interpretations of the questions (i.e., no), for reasons outlined previously.
However, it must be acknowledged that a response such as “yes, but only one,”
is not unequivocal. One could argue that the interpretation of a given question is
something between a yes and a no. A participant’s choice to say yes, although
qualified, may indicate that they are ambivalent, but that their interpretation is
more atelic than telic. Therefore, the data were recoded so that that all yes and
qualified yes, but only one types of responses were counted as atelic and all no
and only one types of responses were counted as telic. Using the recoded data,
the three age x verb-type x determiner-type mixed-model ANOVAs were
repeated. There were no differences with respect to whether a factor or
interaction was statistically significant or non-significant, with one exception. With
the recoded data, the ANOVA examining stimuli with eat-type verbs vs. eat up-
type verbs indicated a statistically significant 3-way verb-type x determiner-type x
age interaction, F(3, 76), = 2.99, p = .036, rho2 = .11. The 3-year-old group did not
demonstrate any differences. This interaction appears to be due to the 3-year old
group having more atelic interpretations for the definite determiner conditions

than the other age groups and unlike other age groups, not exhibiting differences

115

between eat-type and eat up—type verbs across the same determiner-type
conditions. Overall, however, the two coding schemes had very similar results.
Therefore, the results presented above are not simply due to the specific coding

scheme that was used.

116

CHAPTER 5

Discussion

Study l: Adults’ Interpretation of Telicity

The major aim of the Adult Study l was to examine the roles of particular
verb types (i.e., non-partitive quantity-sensitive, partitive quantity-sensitive and
quantity-insensitive), particular detenniners in the object DP (i.e., definite
determiner and cardinal number), and the presence of resultative particles on the
interpretation of telicity by adults. Overall, the results indicate that all of these
factors play a role in the interpretation of telicity to varying degrees and that verb-
type and determiner-type interact with each other to produce particular

interpretations of sentences with regard to telicity.

Part A: Verb-type and Detenniner-type

Study IA focused on the comparison of sentences that contained either of
two types of quantity-sensitive verbs. The critical difference between the two verb
types was that eat-type verbs allow a partitive interpretation of the
object/complement, whereas build-type verbs do not. For example, one can eat
of/l'mm the cake but cannot build of/I'mm the house. Partial eating of a cake could
satisfy the description He ate the cake. On the other hand, partialhnitial steps in
the building of a house cannot satisfy the description He built the house.

The first hypothesis was that VPs containing build-type verbs and

quantized object DPs require a telic interpretation, whereas VPs containing eat-

117

type verbs and quantized object DPs allow for either a telic or an atelic
interpretation. The hypothesis was strongly supported by the results. Nearly all of
the adult responses to questions with build-type verbs indicated telic
interpretations, while for eat-type verbs only about half of the responses indicated
telic interpretations.

The findings reported above also serves as a potential explanation for the
results in van Hout’s (1998) study. In her experiment, adults were expected to
interpret questions with the quantity-sensitive verbs eat and drink and quantized
DPs as telic, but they only did so 25% of the time in English and 78% of the time
in Dutch. It is possible that certain task variables such as static pictures, the use
of the possessive pronoun as a quantizer, or the presentation of a narrative story
may also have had undue influence on the results in the earlier study (van Hout,
1998). Given the present results and the proposed explanation, one may
conclude that participants’ reduced level of interpretation of VPs with eat-type
verbs does not indicate a lack of recognition of eat and drink as being quantity-
sensitive. Rather, it may indicate that many adults interpret VPs with eat-type
verbs as partitive and which allows them to interpret the VP eat the/two brownies
as eat from the/two brownies, an Activity VP which is therefore atelic even when
followed by a quantized DP.

There is another way in which one might mnsider partitivity influencing the
interpretation of the questions. It could be that partitivity allows an implicit some
of reading, as in eat (some of) the brownies. In this case, the partitive would

actually yield a telic interpretation. Here, a yes response to the relevant target

118

questions would indicate a telic interpretation (eating of some of the brownies
culminated in a logical endpoint). However, recall that for the adults, the
presence of the resultative particle up in questions such as Did the woman eat up
the/two brownies? resulted in ceiling levels of telic interpretations. The resultative
particle up blocks the atelic interpretation (e.g., *She ate up from the cake) but
does not block the telic interpretation (e.g., She ate up some of the cake). If the
partitive interpretation was due to an implicit some of the presence of up would
not be expected to result in such a large increase in telic interpretations.
Therefore, I the former explanation is preferred.

The second hypothesis dealt with the relative contribution of two target
detenniners, the definite determiner the and the cardinal number two, to the
interpretation of telicity. Because both of these detenniners serve to quantize the
object DP, theoretically, they should both equally result in telic interpretations
when they follow a quantity-sensitive verb. The hypothesis that the definite
determiner would have less influence than a cardinal number on creating a telic
interpretation is based on the premise that the definite determiner, and not the
cardinal number, would allow for pragmatic accommodation of the sentence as
true (i.e., allow a yes response). Although the cardinal and definite behave
differently, the differences are predictable. In other words, in the quantity-
sensitive VPs following NCE videos, the definite determiner may allow a different
construction of the relevant set and therefore allow adults to conclude that the
event has culminated with respect to a portion of the set and consider this portion

to be the relevant set. This flexibility in the construction of the set is not available

119

for the cardinal number. The cardinal number two requires a counting/matching
of at least two objects. The number two asserts that the relevant DP object
involves at least two of a kind. In contrast, the referent of a definite determiner is
not fixed; it is dependent on the discourse. In sentences with a definite
determiner in the object DP, the listener must decide on what constitutes the set
being referred to in the discourse context. The listener has some latitude in
deciding what constitutes the discourse-relevant set.

In order to pragmatically accommodate a stimulus sentence as being true
in the discourse context, a listener may decide that the set, the Xs, can be
satisfied by any number of objects present to which some threshold amount of
the verb applies, without requiring that the action be completed with all of the
objects in order to be true. Thus, in the cases such as build the houses or eat the
brownies, where a quantity-sensitive verb is followed by a definite object DP, a
listener can decide that the sentence is true if the action is completed on both
objects or that it is true if the action involves a certain threshold amount of each
object. For example, one may decide that that biting off a piece of a cookie,
chewing it, and swallowing it, may be enough to allow the phrase, eat the cookie
to be true, whereas just biting into it may not be. Another individual may require
that at least half of the cookie be consumed to allow it to be true.

The hypothesis regarding the definite determiner and cardinal number was
supported. Overall, there were significantly more telic responses to questions
with the cardinal number in the DP than those with the definite determiner. it was

predicted that this would be the case for questions with both verb types, but to a

120

greater extent with eat-type verbs, because atelic interpretations could also be
obtained due to the partitivity of eat-type verbs. However, this was not the case
with both verb types. When the interaction between verb-type and detenniner-
type is examined closely (see Figure 5), we see that it was only evident in the
case of eat—type verbs. The lack of an observable determiner effect with build-
type verbs appears to be due to the strong telic interpretation that is elicited by
the quantity-sensitive and non-partitive characteristics of build-type verbs. The
number of telic responses in both build-type verb conditions was at or very near
ceiling level showing that adults did use the set as provided in the video.

With eat-type verbs, both determiner conditions present listeners with an
option to interpret the questions as atelic due to the option of interpreting eat-type
verbs as partitive. F urtherrnore, even if the verb is not interpreted as partitive,
listeners can answer yes (with a telic interpretation) by pragmatically
accommodating DPs with definite detenniners as referring to whatever portion of
the object was acted upon. On the other hand, if the verb is not interpreted as
partitive, the cardinal number specifies the amount, i.e., two, and accommodation
is not possible.

Hypothesis 3 specifically addressed the potential interactions between
verb-type and determiner-type by proposing a hierarchy of telic responses based
on combinations of verb-type and determiner-type. It was hypothesized that in
response to yes/no questions following non-culminating events, VPs containing
build-type verbs with a cardinal number would induce the most telic responses,

followed by build-type verbs with a definite determiner, eat-type verbs with a

121

cardinal number, and finally eat-type verbs with a definite determiner. With the
exception of VPs containing build-type verbs with a cardinal number and those
with a definite determiner inducing equally high levels of telic interpretations, the
hierarchy was evidenced in the results.

Overall, Study 1A demonstrated that subtypes of quantity-sensitive verbs
and different detenniners have an impact on the interpretation of VPs as telic or
atelic and that the relative impact of each could be predicted based on the

semantic, syntactic and pragmatic characteristics of the verbs and detenniners.

Part B: Resultative Particles

Study IB focused on the role of resultative particles on the interpretation of
telicity in VPs with quantity-sensitive verbs that allow partitive interpretations (i.e.,
eat-type) and those with quantity-insensitive verbs (i.e., push-type). It was
hypothesized that for push-type verbs, the presence of a resultative particle
would result in a telic interpretation, whereas for eat-type verbs the presence of a
resultative particle would confirm a telic interpretation by disambiguating between
an otherwise possible telic or atelic interpretation. For push-type verbs, which are
quantity-insensitive, the presence of a resultative particle as in push over the X,
creates a telic interpretation and also adds another layer of semantic meaning to
the VP. The specific type of determiner that may be present does not influence
the interpretation of telicity. For VPs containing eat-type verbs, the role of the
resultative particle is not to create a telic interpretation, but to eliminate the

possibility of a partitive, and therefore atelic, interpretation of the VP. Meanwhile,

122

the presence of a determiner that quantizes the DP will contribute to a telic
interpretation. In other words, for quantity-insensitive verbs, it is the resultative
particle alone that creates a telic VP and for partitive quantity-sensitive verbs the
determiner in the object DP makes a contribution to a telic interpretation of the
VP, while the resultative particle serves to overtly mark the VP as telic and
thereby remove the possibility of a partitive/atelic interpretation.

The comparison between eat-type and push-type verbs with particles was
done indirectly because opposite responses are expected for eat-type and push-
type verbs, which would create a confound in direct comparison. Responses to
stimuli with push-type verbs in the various conditions were evaluated separately
from those with eat-type verbs in the various conditions and those results will be
qualitatively examined here.

For eat-type verbs, there was a significant difference between the def. det.
and card. num. conditions, with the cardinal number making a greater
contribution than the definite to the number of telic interpretations. The addition of
the resultative particle up significantly increased the mean number of telic
interpretations in both the def. det. and the card. num. conditions. These results
demonstrate that the determiner-type made a relative contribution to the
interpretation of the questions as telic, but that the resultative particle up, made
an additional contribution, which judging from the near-ceiling level of telic
interpretations, essentially eliminated atelic interpretations (see Table 6 and

Figure 8).

123

For push-type verbs, there was no significant difference between telic
interpretations in the def. det. and card. num. conditions, suggesting that the
determiner type in object DPs of quantity-insensitive verbs does not contribute to
the interpretation of telicity, as predicted. In fact, the overall levels of telic
interpretations were very low (see Table 6). Given the semantics of push-type
verbs, specifically that they are quantity-insensitive, this result was expected. In
both determiner conditions, the addition of the resultative particle over or out
significantly increased the mean number of telic interpretations in both the def.
det. and the card. num. conditions. As illustrated by Figure 11, the difference in
telic interpretations between the non-particle and particle conditions is drastic for
push-type verbs; the telic interpretation appears to be wholly accounted for by
the resultative particle (occurring with a quantity-insensitive verb). By contrast,
the differences between the non-particle and particle conditions with eat-type
verbs, is more gradual (see Figure 8). The quantity-sensitive semantics of eat-
type verbs makes a contribution when combined with a quantized object DP,
presumably when the verb is interpreted non-paritively. The determiner type
makes an additional contribution, with object DPs containing the cardinal number
resulting in more telic responses. Finally, the contribution of the resultative
particle up appears to be to establish the VP as telic, without ambiguity. When
the resultative particle is present, adults seem to ignore the determiner.

The finding that resultative particles function differently indicates that the
resultative particle up with eat-type verbs functions to clarify the meaning of a VP

whose telicity is potentially ambiguous, while resultative particles with quantity-

124

insensitive verbs create telicity and add substantial semantic meaning (overt
change of position/state of the object). It also serves to support the premise that
build-type and eat-type verbs, although both quantity-sensitive, have different
characteristics that impact telicity. If eat-type verbs functioned the same way as
build-type verbs, then we would expect that the telic effect of the particle up
would be present for both verb types, but this is not the case. When the particle
up occurs with build-type verbs, as in build up or fix up, it does not function as a
resultative particle. Build up essentially means to make something larger and fix
up means make something nicer. In these cases, the particle up is not
resultative. Furthermore, if build-type and eat-type verbs did behave identically,
and an eat-type verb with a quantized object DP always resulted in a telic
interpretation, we would have to attribute the occurrence of the parb’cle up in
utterances such as, George, eat up your broccoli, to some other purpose. What
this other purpose could be is not clear. In other words, the partitive characteristic
of eat-type verbs, and the subsequent possibility of atelic interpretations of
sentences such as Did he eat the/two brownies, appears to be the reason why
the resultative particle is needed when a speaker’s intended description of an
event specifically includes, and focuses on, the endpoint of that event. With build-
type verbs there is no partitivity. Essentially, there is no reason for the listener to
think that the description may not include the endpoint, and therefore there is no
need for a resultative particle to establish telicity unambiguously; the verb and

the syntax of the object DP have already done so.

125

The compositional particles as in push over the/two dogs and carry out
the/two bags do not seem to have a primary function of indicating an endpoint,
although they do so obligatorily without influence of the determiner-type in the
object DP. These particle verbs provide additional semantic meaning by
indicating a location or goal, and their telicity may be considered a by-product of
the location or goal. When eat and eat up are compared, there is no substantial
difference in meaning except the overt indication of completion. Eat and eat up
are both quantity-sensitive and the particle up does not appear to add any new
dimension of meaning. On the other hand, when push and push over are
compared, there is a substantial difference in meaning. First, push is quantity-
insensitive, while push over is quantity-sensitive. Second, over adds the
semantic dimension of a goal. In the case of verbs such as push/push over, the
addition of the particle seems to create a new verb; in a way that for eat/eat up it
perhaps does not.

The findings of Study I lead to the conclusion that understanding the
telicity calculus for English-speaking adults requires consideration of subtypes of
quantity-sensitive verbs, subtypes of detenniners and the roles of different types
of resultative particles. The present study examined and found differences
between partitive and non-partitive quantity-sensitive verbs and between the
definite determiner and cardinal numbers that have not been previously formally
examined. It also found that resultative particles play a different role with different

verb types.

126

Study ll: Children’s Interpretation of Telicity
The major aim of Study 2 was to investigate the development of the
interpretation of telicity as a function of: a) particular verb types (i.e., non-partitive
quantity-sensitive, partitive quantity-sensitive and quantity-insensitive), b)
particular detenniners in the object DP (i.e., definite determiner and cardinal
number), and c) the presence of resultative particles, using adult performance

levels of the adults in Study l as a reference.

Part A: Verb-type and Detenniner-type

Study IIA examined children’s responses to stimuli containing eat-type and
build-type verbs in the two determiner-type conditions (i.e., definite determiner
and cardinal number) across 3-, 4-, 5- and 6—year-old age groups. There was a
main effect of age. The responses of the 3-year-old group were quantitatively
different than those of the 5- and 6-year old groups. Specifically, the 3-year-old
responses indicated fewer overall telic interpretations of the target stimuli than
the oldest two groups. The 4-year-old group did not differ from either the 3 year-
olds or the 5- and 6-year olds. in general, there was a developmental trajectory
that emerged, with the 4-year-olds generally appealing to be at an intermediate
stage. However, the results also indicated that the oldest children in the study
had not yet reached adult levels of performance, at least with respect to verb-
type and resultative particles. Although the children’s and the adults’ results were

not statistically compared due to differences in group-size and in the procedures,

127

their results have been placed side by side and clustered by age group in Figures
23 through 25 to facilitate the discussion.

Build-type vs. eat-type verbs. Although numerically the responses to
questions with build-type verbs suggested more telic interpretations than to those
with eat-type verbs (see Figure 13), the difference was not statistically significant.
This suggests that, across the four age groups, there was no distinction between
the two verb-types (but see below for comparisons within each age group). For
both verb-types just over half of the responses indicated a telic interpretation.
Therefore, the hypothesis that build-type verbs would yield more telic responses
was not supported for the children as a single group.

The difference between the adults’ and children’s results suggests that the
children have not yet acquired an adult level of understanding of the specific
verb-type characteristics of build- and eat-type verbs (see Figure 23). Some of
the children did not seem to appreciate that build-type verbs with quantized DPs
require that an endpoint be reached for the VP to be true, whereas essentially all
of the adults clearly did. Those children who did not have a telic interpretation for
such sentences appear to be treating the VPs build the/two houses as if it were
an activity VP, and any amount of building a house is good enough to say that
building a house took place. Alternatively, they just assume that the event will be
completed at some point in the future and that is good enough to consider it to
have happened.

Some of the children’s spontaneous comments suggested that they may

have been answering the questions with a focus on the verb alone, with little

128

influence from the DP on their interpretations. During data collection, some of the
child participants were noted to occasionally answer questions such as Did the
man build the houses? with responses such as “Yep, he was building”, or
questions such as Did the woman drink the sodas? with “Yeah, she drank.”
Although, such responses were identified across age groups, they seemed to be
more frequent among the younger children. Notably the youngest children also
had the fewest telic responses with either verb-type. (See Figure 23 for a
comparison between ages.)

The proposal that the children are initially treating build-type VPs as
activity VPs is reasonable, but tentative, as information about the interpretations
of individual children, and the pattern of responses within an individual child is
lost in the course of averaging across children. At all age levels, some children
indicated a telic interpretation for both tokens of a given verb type + determiner
type condition, others indicated an atelic interpretation for both tokens and others
still gave one of each, suggesting ambivalence. It seems clear that the children,
as a group, did not interpret the verb-types, especially the build-type verbs, in the

same manner as adults.

129

 

 

  

E build-type + def. det.

   
  
  
 

El build-type + card. num.
I eat-type+ def. det.

Ieat-type + card. num.

 

    

 

 

Mean Count of Telic Interpretations

3-year- 4-year- 5-year- 6-year- adults
olds olds olds olds

 

 

 

Figure 23. Mean counts of telic interpretations by verb-type and determiner-type
presented by age group for children and adults for build-type and eat-type

conditions“.

The effects of various combinations of verb-type and determiner—type were
analyzed by age group. Although there was no overall effect of verb-type across
the age groups, the 3—year-olds did exhibit a significant difference between the
build-type + def. det. condition and the eat-type + def. det. condition. The
difference for the 4-year olds approached, but did not reach statistical
significance. This result is somewhat odd because at first glance it implies that
the younger children make some distinction between the verb-types that is later

lost. However, when we carefully examine the results, we see that this is not

 

8 The data for children and adults are presented side-by—side here for qualitative
comparison. Statistical comparisons between the adults and children were not
conducted.

130

necessarily the case (see Figure 15 or 23). For the 3—year-olds, there is a
significant difference between build-type + def. det. and eat-type + def. det.,
where there are very few telic interpretations with eat-type. It appears that the 3-
year-olds treat VPs such as eat the/two brownies as activity VPs more than VPs
such as build the/two houses. It may be the case that this is related to transitivity.
Build-type verbs are transitive and require a direct object, whereas eat-type verbs
can be detransitivized. and consequently not require an object. The fact that eat-
type verbs can be detransitivized may make it easier for the 3-year-olds to
interpret questions with eat or drink as being just about the verb and not
specifically about the direct object”.

Another reason that we must be careful not to interpret the verb-type
distinction made by the 3—year olds as the 3-year-olds having some knowledge of
verb-type distinction that the older children do not, is that theoretically, when
partitivity is not considered, both verb-types should yield telic interpretations
when followed by a quantized DP. In Figures 15 and 23 it is clear that the older
children demonstrate a better knowledge of both eat-type and build-type verbs as
it relates to telicity, as indicated by overall higher levels of telic interpretations in

the older age groups. Although the 3-year-oids have a statistically significant

 

9 This is somewhat consistent with Wagner’s (2006) finding that 2-year-olds have
a bias for interpreting transitive sentences as telic. In her study, 3-year-olds did
not demonstrate the bias. However, given task differences, verb differences and
the range of variation in language development, it is still possible that some of
the 3-year-0lds are exhibiting a transitivity bias in the case of build-type verbs
that is not present with the eat-type verbs, because eat-type verbs can be
detransitivized.

131

difference between build- and eat-type verbs, it seems to be qualitatively different
from the build- and eat-type difference exhibited by the adults.

Push-type verbs. It is fairly reasonable to have a lack of overall differences
between build- and eat-type verbs because they are both quantity-sensitive and
are both capable, if not required, to allow telic interpretations when followed by a
quantized DP. In terms of telicity however, push-type verbs should yield an atelic
interpretation regardless of the composition of the object DP. For example,
following an NCE video, we would expect the answer to Did the woman push
the/two dogs? to be yes. Whether or not the dogs were pushed over, should not
matter, if any amount of pushing took place, the answer is expected to be yes.
There was not a formal hypothesis that this would be the case for children,
because the primary purpose of including push-type verbs in the study was to
examine the role of resultative particles. However, because the children’s results
for questions with push-type verbs indicated a relatively high frequency (nearly
half) of no responses, i.e., (telic interpretations) which was unexpected, a follow
up analysis comparing push-type to build- and eat-type verbs was conducted.

If children had some notion of the quantity—sensitive/quantity-insensitive
distinction, we would expect that there would be differences between push-type
verbs and both build- and eat-type verbs with regard to telicity, despite the
relatively frequent telic interpretations of questions with push-type verbs.
Statistically significant differences between push-type and the quantity-sensitive
verbs were found, indicating that at least some, if not most, of the children had an

appreciation of this distinction. This is important because it shows that children

132

are not treating all of the verbs as the same. Although, there were no significant
differences between the two quantity sensitive verb-types (i.e, build-type and eat-
type), there are differences between both quantity-sensitive verb-types and the
quantity-insensitive verbs. Thus, the children do seem to be aware of the broader
distinction between quantity-sensitive verbs and quantity-insensitive verbs,
although they are still refining their knowledge of verb-type characteristics.
However, the fact that approximately half of the responses to push-type
VPs were telic, calls for some explanation, especially since in the adult data, the
percentage of telic interpretations of push-type verbs was very small. As was
mentioned earlier, young children have been found to exhibit a telic bias for
transitive sentences (Wagner, 2006). Perhaps some of the children continue to
have a telic bias for transitive sentences. Despite the quantity-insensitive
characteristic of push-type verbs, the fact that the sentences are transitive may
serve as a more salient, albeit incorrect, cue for a telic interpretation.
Alternatively, some of the children may have treated push as push over and cany
as carry out. For adults, one could argue that push and push over are lexically
distinct from each other, but perhaps some of the children have not reached that
conclusion yet or the verb push is ambiguous for them. Another possibility is that
push and/or carry are not the words that some of the children would choose for
descriptions of the scenes that they viewed. Their no responses may have been
denials that pushing or canying took place. They may have characterized the
action they viewed for the verb push as shoving or sliding, etc. and for the verb

carry as taking.

133

Learning verbs. Verbs are difficult to Ieam. They are much more than
simply labels for actions. Consider what it means to know a verb. Consider the
verb build, in the sense of build a house (vs. build an argument). Semantically,
this verb refers to the physical process of constructing an object by putting pieces
or parts together. It is durative and quantity-sensitive. It is transitive and requires
an animate subject and a direct object. There are multiple pieces of semantic and
syntactic information that are needed to ‘know’ a verb. Knowing a verb also
means having an understanding of how it interacts with the rest of the VP and
sentence. All of the verbs selected for this task were identified as being
understood and/or used by over 75% of 30 month old children (Dale 8r Fenson,
1996). However, the depth of understanding of each of those verbs is unknown.
Given the amount of information that is involved in verb Ieaming, it is not
surprising that even by age 6 the children do not demonstrate fully adult-like
knowledge of the verb-types examined here. Despite their apparently incomplete
knowledge of all of the aspects of the verbs, it is important to observe that at
least for the quantity-sensitive verbs, the children are not qualitatively very
different from the adults. Although they may not have responded no to the target
questions with the same frequency as adults, it is noteworthy that the response
patterns across conditions for the 4-, 5-, and 6- year old children are remarkably
similar to that of adults (see Figure 23). The most notable difference between the
children and the adults is that the adults have very high levels of telic
interpretations of build-type verbs with quantized UPS and the children do not yet

reach such levels by age 6. This suggests that there is some bit of information

134

about this type of verb and/or its relationship to the DP that not all of the children
have acquired yet.

The literature on syntactic bootstrapping (e.g., Gleitrnan, 1990; Naigles,
1990) has demonstrated that children use information about the syntactic frames
in which verbs appear to learn aspects of their characteristics. In the case of
telicity, the specific meaning of a verb plays a crucial role in how it will be
interpreted within the syntax of the rest of the VP. There appears to be an
relationship through which the syntax of a VP informs the child’s knowledge of
the semantics of the VP and conversely, the semantics informs the child’s
knowledge and interpretation of the syntax of the VP (Pinker, 1989). Since
children at 6 years of age do not yet have the adult levels of performance, but
appear to be approaching adult levels, indicates that the acquisition of telicity, as
well as acquisition of verb knowledge in general, is a process that occurs over an
extended period of time.

Detenniner-types. It is well documented in the literature that children as
old as six do not yet have a fully developed determiner system and have
particular difficulties Ieaming all of the subtleties of the definite determiner
(Karmiloff-Smith, 1979; Maratsos, 1976; Munn, et al., 2006; Perez-Leroux, et al.,
2004). Although they acquire and use the definite determiner early on, there is an
extended period of development during which the system has to be fine-tuned
and reflects the adult system. One factor that impacts the difficulty of fully
understanding the definite determiner is that its interpretation is fundamentally

discourse-based. The listener must decide what constitutes the actual referent

135

that is being talked about by the speaker. For children, the definite determiner
appears to have a broader/different application than it does for adults (Kanniloff-
Smith, 1979 and Perez-Leroux, et al., 2004). On the other hand, the cardinal
number specifies the quantity of objects being referred to by the speaker,
eliminating the need to make a discourse-based decision in contexts such as the
one used in this experiment. Children acquire the ability to count/match small
quantities by age two and perhaps even earlier (Gelman & Gallistel, 1978). Given
the difference in children’s abilities with definite detenniners and cardinal
numbers, it was hypothesized that they would be better able to use cardinal
numbers to understand a DP as quantized than the definite determiner. To be
clear, the claim was not that children could not understand or use the definite,
just that the understanding of its multiple functions and discourse-based
interpretation has not reached the adult state. In other words, the cardinal
number would be more effective as a quantizer than the definite determiner. The
children in the study did indicate telic interpretations more often when the
cardinal number was used to quantize the DP than when the definite determiner
was used.

There was an interaction between verbtype and determiner-type. The
difference between the determiner types was only present for eat-type verbs (see
Figures 15). With the exception of the 5-year-old group, this interaction was
found at all age levels. The lack of a difference for the 5-year-olds was due to
their greater frequency of telic interpretations in the eat-type + def. det. condition,

than any of the other age groups.

136

The same kind of verb-type by determiner-type interaction was found for
adults. In the adult case, the lack of difference between the determiner types in
the build-type verb conditions could not be further examined due to the ceiling
effect. VPs with build-type verbs were interpreted as telic nearly 100% of the time
regardless of the determiner. Thus, the children’s interpretations of questions
with build-type verbs do not appear to have depended on the determiner-type in
the object NP. Rather, it seems to have rested on their knowledge, or perhaps
lack of knowledge, about the verb types/subtypes.

In order to address the hypothesis regarding the role of the resultative
particle with different verb types, the push-type + def. det. condition was
compared to the push-type + card. num. condition. It was anticipated that push-
type verbs would equally yield atelic interpretations in both the def. det. and card.
num. conditions. As discussed earlier, although there were significantly more
atelic than telic interpretations, the levels of telic interpretations were surprisingly
high overall, particularly by comparison to the adult results. However, there were
significant differences between the two determiner conditions in the number of
telic interpretations for the 3- and 4-year-olds. These younger age groups had a
significantly higher mean count of telic interpretations in the cardinal number
condition than in the definite determiner condition, suggesting that for them, the
cardinal number made some independent contribution to a telic interpretation.
More than half of the 3- and 4-year-old responses to push-type verbs + card.
num. (i.e., Did the man push two dogs and Did the woman cany two bags) were

telic. It seems that for at least some of the youngest children, despite the

137

presence of a quantity-insensitive verb, the presence of the cardinal number in
the object DP was a stronger, albeit incorrect indicator of a telic VP.

There are at least two ways to explain children's larger number of telic
interpretations with object DPs quantized by two than those quantized by the.
Both are based on children’s ability to count and their simultaneous limitations in
understanding the range of functions of the definite determiner. The pragmatic
accommodation explanation, which was used to account for the adult results,
may not be available for the children’s results. Accommodation of the definite
determiner is largely dependent on a complete understanding of the definite
determiner as defining a maximal set and the adults’ ability to construct a new
relevant set, different from that in the context. Children may have difficulties with
this aspect of the definite determiner to begin with (Karmiloff-Smith, 1979;
Maratsos, 1976). Unless they have an adult-like knowledge of the definite
determiner, they would be unlikely to engage in pragmatic accommodations in
the same manner as adults.

The first possibility is that children interpret the plural definite as a
reference to a kind (or category). For example, in a sentence such as The lion is
the king of the jungle, the DP, the lion, refers to a kind of animal and not to a
specific animal. Although the plural definite does not yield a kind reading for
adults, Perez-Leroux, et al. (2004) found that some children do interpret the
plural definite as referring to a kind. In the context of the present study, it is
possible that a kind interpretation of the object DP would result in an atelic

interpretation. For example, with a kind reading of the plural definite, it is feasible

138

that a child could interpret the question Did the man eat the brownies? as asking
Did the man do brownie eating? regardless of whether or not the man completely
consumed something that belongs to the category of brownie. Given the visual
stimuli that preceded the question, this interpretation would yield a yes response
which is atelic because the entirety of the brownies was not consumed. The
second possibility is that the cardinal number in questions such as Did the man
eat two brownies? would trigger counting and be interpreted as asking whether
or not the man consumed two separate objects that were both brownies. Given
the visual stimuli of NCE video segments, this interpretation would be more likely
to yield a no response, which is telic because the entirety of the brownies was
not consumed. Furthermore, children’s ability to understand and use cardinal
numbers to count wholes, may facilitate their use of cardinal numbers as an
indicator of telicity even, as we have seen, when the adults would not, as in the
case of push-type verbs.

The finding that the cardinal number facilitated telic interpretations by
children is also interesting in regard to a study by Wagner and Carey (2003). In
one portion of their study, they investigated children’s ability to use linguistic
information, in the form of telic vs. atelic sentences, to count events. The
participants in their study were 3-, 4-, and 5-year old children and a group of
adults. The participants viewed animated movies that depicted one or two
instances of an event with a clear goal, such as building a house. Each instance
of an event took place in two to three steps, with a temporal pause between the

steps. For example, in one of the movies a girl is shown painting two flowers.

139

Based on the task description, the girl in the movie presumably paints half of the
flower with one stroke, pauses, paints the rest of the flower with a second stroke,
and proceeds to the next flower in the same manner with a pause between
strokes. After each movie the children were asked how many times X happened.
The X referred either to a goal-oriented event or to temporally discrete process
actions. Using the flower-painting example, the verbal stimuli may have been
given a telic description as in How many times did the girl paint the flower? or it
may have been given an atelic description as in How many times did the girl
paint? The children counted goal-oriented events when the stimulus questions
were telic descriptions, approximately 60% of the time for all three age groups
(adults did so essentially all of the time) while they only did so approximately
20% of the time when the stimulus questions were atelic. Although the children’s
performance was significantly different than that of the adults, Wagner and Carey
(2003) concluded that the children’s decision of whether to count whole events or
temporal units was influenced by whether the linguistic description was telic or
atelic.

It is possible that, if children can use the cardinality of the object to
determine that they should count whole events, than it should also be possible
that they can use count information as an indicator of a telic description (in
combination with verb information). For the present study, this account would
mean that at least some children take the presence of the cardinal number two to
quantify the number of whole events that are to be counted, rather than just the

number of objects. With this understanding of the function of the cardinal

140

number, Did the man eat two brownies? would be construed as Were there two
unique events in which a brownie was eaten (in each). In response to NCE
videos, the answer would be no. In this case, the number of full events that can
be counted is important to the interpretation. For VPs with the definite
determiner, the question Did the man eat the brownies? would be construed as
Was there an event of brownie-eating? In response to NCE videos the answer
could be yes. . Thus, children’s broad view of the applicability of the definite
determiner may lead to more atelic responses. This is a plausible explanation but
further research is needed in order to determine whether listeners allow the
cardinal number to quantify the number of whole events or just the number of
objects.

Both accounts considered here are probable. The present experiments did
not allow for further examination of the difference between the two determiner
types in the eat-type verb conditions. What is rather clear however, is that the
cardinal number is significantly associated with telic interpretations. Consistent
with the findings of Wagner & Carey (2003), the present study suggests that
there is a link between the ability to count out discrete units of a stimulus, be they
objects or events, and telic linguistic descriptions. In other words, with quantity-
sensitive verbs, a discrete quantity facilitates a telic interpretation.

The preceding discussion focused on the difference in children’s
interpretations of telicity with the definite determiner vs. the cardinal number. One
of the striking things about the results however, was that although the children

interpreted VPs in the cardinal number condition as telic more often than in the

141

definite determiner condition, the older children did not have fewer telic
responses in the determiner condition than adults (based on visual inspection of
Figure 23). If there were any differences between the 4-, 5-, 6-year olds and the
adults, they would be in the direction opposite of what would have been
expected. The children had numerically higher rates of telic responses in the eat-

type + def. det. condition than the adults.

Part B: Resultative Particles

Study "3 examined how well a resultative particle serves as a cue for telic
interpretations with partitive quantity-sensitive eat-type verbs and quantity-
insensitive verbs. (T o facilitate reference to the data, Figure 24 below graphically
presents the results from the child study as well as the adult study. Note that
direct statistical comparisons between the child and adult results were not
conducted.) Overall, children had more telic interpretations of VPs in the
resultative particle conditions than in the non-particle conditions in the analyses
for both the eat— and push-type verbs. This result was expected because the
resultative particle should be an unambiguous indicator of a telic VP. Unlike verb-
type and determiner-type, the resultative particle does not have multiple
interpretations or functions. However, not all age groups exhibited a statistically
significant advantage for the resultative particle over the detenniner-only
conditions for telic interpretations. In van Hout’s (1998) study of the 3- and 5-
year-olds distinguished the particle sentences from the atelic sentences, and the-

5-year-olds had a high proportion of telic responses (.91) in the particle condition,

142

not unlike that of the 5- and 6- year olds in this study. The 3—year—olds in this

study had fewer telic interpretations of the particle up than in van Hout (1998).
Perhaps methodological differences may account for the different results. The
present study compared the particle conditions to the quantized DP conditions

and not to atelic conditions.

I

 

.5
N
U'IN

 
   
  
 
   

_\
0|

ligaziy’pe: @9177
Ieat-type + card. num.

._ IIeat up—type + def. det. I
Leatvup-typej card. num. I

.-
N
U1

 

0.75

.0
or

0.25

 

Mean Count of Tellc Interpretations I

O

3—year- 4—year— 5—year- 6-year- adults
olds olds olds olds

 

Figure 24. Mean counts of telic interpretations by verb-type and determiner-type
presented by age group for children and adults for eat-type and eat up-type

conditions.

In the comparison of the eat-type + def. det. condition and the eat up-type
+ def. det. condition, the differences were significant only in the 4- and the 6-
year—old groups. The lack of differences in the 3-and 5-year-old groups,

appeared to be for different reasons. As seen in Figure 24, the mean counts of

143

telic responses for 3-year-olds in both the eat-type + def. det. condition and the
eat up-type + def. det. cond. are quite low, .33 and .53 (on a scale of 2.0)
respectively, and the difference is non-significant. The 3-year-olds rarely used
the resultative particle up as an indicator of telicity with eat-type verbs. The
situation appears to be different for the 5-year-olds. Looking at Figure 24, one
can see that, although there is little difference between the two conditions for the
5-year-olds, the frequency of telic responses for both conditions is higher. The
resultative particle does not appear to be a better indicator of telicity for them. At
the same time, the definite determiner is not a particularly poor indicator either.
Interestingly, at least numerically, the 5-year-olds had the highest count of telic
interpretations in the eat-type + def. det. condition, even in informal comparison
to the adults.

The results for the comparison of the eat-type + card. num. condition and
the eat up-type + card. num. condition, were surprisingly non-significant for all of
the groups. Based on these results, it can be concluded that, with partitive
quantity-sensitive verbs, the cardinal number alone, is essentially as good of a
telicity marker for children as the resultative particle up. It may even be better for
the youngest of the children. Although it was not tested statistically, in the eat-
type + card. num. condition, the mean count of telic interpretations by 3-year-olds
was numerically higher than in the eat up-type + def. det. condition. The
youngest children seem to be able to make better use of the mrdinal number as

an indicator for a telic interpretation than the resultative particle up.

144

 

Ilfish-ty; + def. det. I
. Eipush-type + card. num.

push over-type + def. det. I
99390 QPEHE.L°§“L"PT; I

 

0.75

 

IOIIIIIIIIIIIIIIIIIIIIIIIIII
ilili!iliti!i!i!iIiIIIi!ililililrlililtliiili!i!i!i!i!t

W

'I:I'lll:l'I'l:l'I'I'l'l'l'l'l'l'l'l'l'
£23222

7:
g.
/
2
/
125'

i

._' l‘ _

ME.

I 3-year- 4-year— 5-year— 6—year- adults
olds olds olds olds

 

Mean Count of Telic Interpretations .

 

Figure 25. Mean counts of telic interpretations by verb-type and determiner-type
presented by age group for children and adults for push-type and push over-type

conditions.

When comparing quantity-insensitive verbs with and without the
resultative particle, the children had overall higher levels of telic interpretations in
the conditions with the resultative particle than in those without it, despite higher
than expected number of telic responses to the push-type verb conditions. (See
Figure 25.) However, this was not the case for all age groups. In the comparison
of the push-type + def. det. condition and the push over-type + def. det.
condition, the particles, over and out, did not significantly increase the number of
telic interpretations for the 3— and the 5- year-olds. In the comparison of the push-
type + card. num. condition and the push over—type + card. num. condition, the

differences were only significant for the oldest two groups. The youngest children

145

showed no significant advantage of the resultative particle over the cardinal
number two as an indicator of telicity.

It appears that although there are overall differences in the interpretation
of both partitive quantity-sensitive and quantity-insensitive verbs with and without
a resultative particle, the older children generally exhibited a better grasp of
these differences than the younger children. Perhaps the most interesting thing
discovered in the data concerning resultative particles was that for all age groups
with eat-type verbs and for the 3- and 4-year-olds with push-type verbs, the
cardinal number was at least just as strong of an indicator of a telic interpretation
as the resultative particles. (Compare the verb-type + cardinal number conditions
to the resultative particle verb-type + definite determiner conditions.) This is in
marked contrast to the adults, for whom, in both verb types, the resultative

particle served as a largely unambiguous indicator of telicity.

Indirect and Qualified Responses

During their introduction to the experimental task, children were told that
after each movie, the puppet would ask him/her a question about the movie and
that they should answer yes or no. Most of the time children followed the
instructions and gave yes or no responses. However, some of the children
sometimes gave responses that were less direct. Some children, rather than
saying no, gave indirect no responses that explained why the answer was not
yes. The most common forms of this type of response were “one”, “just one”, and

“only one." Table 15 below provides examples of the range of such responses.

146

These responses were counted as no responses because it was clear by their

explanation that they we denying that the events occurred as described in the

stimulus questions. In other words, they were saying no indirectly.

Table 15: Examples of Children’s Indirect No Responses

 

 

Child ID Age Question Response
Did the woman eat up the
33 5:2 Just one.
brownies?
12 3:8 Did the man fix the dolls? One.
12 3:8 Did the woman build two houses? One and a half of one.
He fixed this one, but he
37 3:9 Did the man fix the dolls? didn’t know how to fix her
legs.
136 5:4 Did the man fix two dolls? Almost.
A whole one and a half of
12 3:8 Did the man drink the sodas?
one.
58 3:2 Did the woman drink two sodas? Not that soda, this soda.
Did the woman eat up the She ate up one but not the
74 4:2
brownies? other one.
87 5:4 Did the man eat two brownies? Not both of them.
58 3:2 Did the woman cany the bags? Not that bag, the other bag.
89 4;1 Did the man carry out the two bags? He left one.

147

Some children gave qualified responses to questions. In such situations,
they answered yes, but followed up with a qualification for their response that
conflicted with the yes response. Examples of this type of response include “yes,
but only one though;" “yeah, but he not eat one brownie;” “yes, but one is still
left;" and “yes, well she pushed one down.” Table 16 below provides examples of
the range of such responses; a complete list of the children’s qualified responses
is in Appendix D. Such qualified responses were counted as no responses
because the qualifying phrases indicate that the child was not satisfied with
his/her yes response as adequately answering the question. In other words,
there is a contradiction between the qualifying phrase and yes, effectively
negating the yes response. It could be argued however, that in such cases the
child was ambivalent about the response and although yes may not have been
an adequate response, neither was no. Further, it could be argued that the child
was dissatisfied with both a yes or no response, but because they answered with
a yes, the qualified responses should have been counted as yes responses. Both
are valid arguments. As noted earlier, the data were reanalyzed with qualified
responses being counted as yes responses and there was very little difference in
the results. This makes it clear that the results were not simply a by-product of
the coding procedures.

Some children gave qualified responses rather consistently and others just
occasionally. The fact that there were qualified responses is very interesting.

Such responses seemed to indicate that at least for some of the children the

148

Table 16: Examples of Children's Qualified Responses

 

 

Child ID Age Question Response

3 6:0 Did the woman build the houses? Yes, but only one though.

12 3:8 Did the woman eat up the brownies? Yes, one.

13 4:3 Did the woman build the houses? Yeah, just one house again.
60 4:5 Did the woman build two houses? Yeah, didn’t finish the rest of it.
89 4:1 Did the man fix the dolls? Yep, but not that one.

10 4:2 Did the woman eat up the brownies? Yeah but he not eat one brownie.
14 3:5 Did the woman eat up two brownies? Yes, she didn’t eat the last one.
107 5:8 Did the man drink up two sodas? Yep but ones lefi.

87 5:4 Did the woman drink two sodas? Yeah, but not both of ‘am.

13 4:3 Did he woman push over the dogs? Yeah, she pushed over one dog.
10 4:2 Did the woman carry the bags? Yes one bag just not two bags.
80 4:7 Did the woman carry the bags? Yeah, but her cam'ed only one.

149

decision to answer yes or no was not always a simple one. It is possible that the
yes portion of their responses referred to whether or not the verb was
represented in the video and that the qualifying phrase referred to the amount of
the objects that was affected. This could be interpreted as meaning that the
children knew that there was something about the DP that did not make a yes
response adequate, but if they considered the question to be about the verb, they
were unwilling to say that the action represented by the verb did not occur by
answering no. It is possible that other children had inclinations to respond in this
manner as well, but that they simply made a choice between the two options
presented in order to comply with the instructions. The ambivalence about the
questions that is represented by the qualified responses may be an indication
that the children who responded in this manner are a point at which they are

discovering the contribution of a quantized DP, particle or verb-type to telicity.

Clinical Implications and Future Research

As discussed in the introduction, past research has indicated a
relationship between verb morphology and VP aspect in the language of young
typically developing (TD) children (e.g., Antinucci & Miller, 1976; Bloom, Lifter &
Hafitz, 1980; Hamer, 1981: Johnson & Fey, 2006; Shirai & Andersen, 1995:
Weist, Wysocka, Witkowska-Stadnik, Buczowska & Konienczna, 1984). In their
early used of verb morphology, children tend to use the progressive —ing
morpheme earlier and more frequently with activity VPs and the past tense

(regular -ed and irregular) with event VPs. The distinction between these two

150

types of VPs is that activity VPs are atelic descriptions of eventualities and event
VPs are telic descriptions of eventualities. Specifically, it has been shown that TD
children exhibit sensitivity to VP aspect in their early differential use of verb
morphology. It is well attested that during the preschool and early school-age
years children with SLI have significant difficulty with verbal morphology,
particularly verbal morphology that marks tense/agreement (e.g., Leonard,
Bortolini, Caselli, McGregor, & et al., 1992: Leonard, Eyer, Bedore, & Grela,
1997; Rice & Wexler, 1996; Rice, Wexler, & Cleave, 1995). A better
understanding of the relationship between verb morphology and VP aspect in
children with SLI may lead us to more focused intervention strategies, particularly
with the past tense, by guiding the selection of treatment stimuli.

If the distributional biases in child language are associated with the way
verbal morphology combines semantically with VPs in the most optimal way, then
we might expect that patterns of interpretation of VP aspect and the use of
tense/aspect morphemes by SLI children will be similar to that of TD children.
There is some evidence for this in that Leonard, Deevy, Miller, Charest, et al.,
(2003) have shown children with SLI to use the aspectual progressive -ing
morpheme in the same manner as TD children. In a previous study, Ogiela,
Casby and Schmitt, (2005) examined the relationship between verb morpheme
use and VP aspect in the Leonard corpus of SLI language transcripts in the
CHILDES database (MacWhinney, 2000). They found a similar distribution
pattern of verb morphology and VP aspect in children with SLI as others have

found with TD children. However, the SLI children were older than most of the TD

151

 

children in earlier studies. Conversely, using an elicitation probe task, Leonard,
Deevy, Kurtz, Krantz, et al., (in press) did not find similar patterns in children with
SLI compared to TD age-matched and TD MLU-matched children.

The relationship between VP aspect and the use of verb morphology
appears to implicitly rely on the ability to determine the telicity of a VP. It is
important to make it clear that the ability to calculate telicity is not being posited
as a requirement for the acquisition of verb morphology. As discussed earlier, VP
aspect and grammatical tense/aspect marking are related, but also function
independently. As evidenced by the present study, the wlculation of telicity itself,
which involves interactions between parts of the VP semantics and syntax, is a
rather formidable task that TD children have not mastered by age 6. It is possible
that children with SLI, in addition to difficulties with verb morphology, may
independently have difficulty with the calculation of telicity.

One might predict that because telicity is determined by an interaction
between the verb and the rest of the VP that children with SLI would have
difficultly with the interpretation of telicity, given that they tend to have difficulties
related to verbs, verb use, and VP development (e.g., Hadley, 1998; Kelly & Rice
1994; Rice & Bode, 1993; Watkins, Rice & Moltz, 1993). In a recent study,
Schulz and Wenzel (2005) examined the interpretation of telicity of VPs
comprised of eat-type verbs and object NPs quantized by a definite determiner
(i.e., eat and drink) in German-speaking adults, TD children 4- to 6-years old and
children with SLI. The methodology of the study was very similar to that of van

Hout (1998). They found no statistically significant differences between the

152

groups. Although statistical comparisons between the adults and children were
not completed for the present study, visual inspection on the results for eat-type
verbs in Figure 23 suggests that there may also not be much difference between
the 4-6-year-olds and the adults in English for eat-type verbs. In the present
study, excluding the 3-year-olds, approximately half of the responses in the eat-
type + def. det. condition were telic interpretations (mean counts ranging from .85
for the adults to 1.33 for 5—year-olds on a scale of 2.0). If there were significant
differences, they would be in the direction of the children having more telic
responses in the eat-type + def. det. condition than the adults. In Schulz 81
Wenzel’s (2005) study, the adults, TD children and SLI children had a telic
interpretations in the eat-type + clef. det. condition only 27%, 21%, and 19% of
the time, respectively. Additionally, there were substantially fewer atelic
responses by 4-6 year old children in the resultative particle condition (4%) than
in the present study, (mean counts of 0.71, 0.56, 0.26 on a scale of 2.0 for the 4-,
5-, and 6-, year olds, respectively). (See Figure 23.) Although direct comparisons
cannot be made, there seems to be a substantial difference between the findings
of the two studies, specifically pertaining to the overall frequency for atelic
responses for eat-type VPs and the greater number of atelic responses in the
particle condition in the present study. Language—specific differences between
German and English, such as the transparency of verb particle meanings, may
be one source of the discrepancies. Secondly, methodological differences such
as the use of pictures vs. video and the presence vs. absence of narrative stories

accompanying the visual stimuli may also play a role. Further, Schulz & Wenzel’s

153

(2005) study addressed only one verb type, the eat-type verbs, for which we find
variable results which are likely due to multiple permissible interpretations of eat
and drink (partitive or non-partitive). On these verbs, however, they found that
the SLI children performed similarly to the adults and TD children. Conducting
the present study with SLI children using the multiple verb types examined here
may add to our understanding of SLI children’s abilities to use and integrate

semantic and syntactic VP characteristics in interpreting telicity.

Conclusions

In sum, the present study identified a finer—grained homework within which
to examine the interpretation of telicity in adults and children. Further,
consideration of verb subtypes and the impact of different determiner types on
the calculation of telicity is necessary. Future research can further refine the
framework presented here. With regard to the development of telicity
interpretation, children appear to be making some distinctions between verb
subtypes, but these have not been mastered by age 6. Although the children did
benefit from resultative particles as a cue to telicity, for the children in this study,
their impact was not great as was expected. However, the cardinal number two
was found to be at least as good of a cue for telicity as the resultative particles,
even with verbs that were not expected to yield a telic interpretation. Using age
as a gross measure of development, there appear to be important changes in the
fourth year in the development of telicity, as 3-yeanolds were significantly poorer

at interpreting telic VPs than 5- and 6-year olds, while 4-year olds were not

154

different from either the younger or the older children. It is natural to expect that
older children will be more proficient than younger children in their language
skills. Therefore, age is a reasonable variable by which to group participants.
However, given the well-known variability in the rate of language acquisition
among TD children, such grouping can be somewhat artificial (although at times
necessary). During data collection it was clear that at all age levels there were
some children whose responses patterned closely with the predictions for adults,
and some whose responses were quite variable or perhaps would exhibited
some other pattern. Future analyses of the data that examines the patterns of
responses of individual children and sub-groups of children may yield further
information about the development of telicity.

Expansion of the present study to other verbs, verb-types and detenniner-
types will further add to our understanding of telicity and possibly allow for more
focused selection of materials and stimuli for examining interactions between
telicity and other areas such as verb morphology. Extension of this study to
children with SLI could potentially lead to a better understanding of their
difficulties with particular characteristics of verbs and verb phrases, specifically
as they may relate to telicity. Further, it may help us to better understand the
relationship or lack of relationship between VP aspect and verb morphology in

children with SLI.

155

APPENDIX A

Linguistic Stimuli

Table A1: Linguistic Stimuli that Served as Target Questions for Analyses

 

 

Item Number Question
1 Did the woman build the houses?
2 Did the woman build two houses?
3 Did the man fix the dolls?
4 Did the man fix two dolls?
5 Did the man eat the brownies?
6 Did the man eat two brownies?
7 Did the woman eat up the brownies?
8 Did the woman eat up two brownies?
9 Did the man drink the sodas?
10 Did the woman drink two sodas?
11 Did the woman drink up the sodas?
12 Did the man drink up two sodas?
13 Did the man push the dogs?
14 Did the man push two dogs?
15 Did the woman push over the dogs?

156

16

17

18

19

20

Did the woman push over two dogs?
Did the woman carry the bags?

Did the woman cany two bags?

Did the man carry out the bags?

Did the man carry out two bags?

157

APPENDIX 8

Results of t-tests Comparing Token Verbs of Each Verb-type

Table B1: Results of t—Tests Comparing Responses to Questions with the Token

Verbs of Each Verb-Type within Each Experimental Condition

 

 

Verb + Detenniner Comparison Pairs t cit p (2-tailed)
build the vs. fix the -1.77 46 .083
build two vs. fix two .00 46 1.0
eat the vs. drink the 1.40 46 .168
eat two vs. drink two 1.43 46 .160
push the vs. carry the -1.77 46 .083
push two vs. carry two .37 46 .710
eat up the vs. drink up the -1.43 46 .160
eat up two vs. drink up two 1.43 46 .160
push over the vs. carry out the -.33 46 .743
push over two vs. cany out two -1.66 46 .103

 

Note. Given the Bonferroni correction for multiple tests, a = .005; No significant

differences found between verb tokens within a verb-type.

158

APPENDIX C

Characteristics of Child Participants

Table C1: Test Scores and Demographic Characteristics of Child Participants

 

 

PPVT-III CELF-P Sent CELF-P Word
Standard Score Struc. Subtest Struc. Subtest
(Mean=100, Standard Score Standard Score
Child ID Age Sex Ethnicity SD=15) (Mean=10, SD=3) (Mean=10, SD=3)
52 3:2 M C 1 10 1 1 10
58 3:2 F C 102 14 12
16 3:4 F C 1 1 1 17 13
14 3:5 M C 82 1 1 10
12 3:5 M H 99 9 9
24 3:5 M C 96 10 9
53 3:6 F C 123 12 14
71 3:7 F C 1 17 15 13
34 3:7 M C, H 103 9 7
72 3:7 F AA, C 100 15 13
48 3:7 M C 1 32 16 13
80 3:7 F C 122 16 1 1

159

Table C1 (cont)

 

 

PPVT-III CELF-P Sent. CELF-P Word
Standard Score Struc. Subtest Struc. Subtest
(Mean=100, Standard Score Standard Score
Child ID Age Sex Ethnicity SD=15) (Mean=10, SD=3) (Mean=10, SD=3)

94 3:8 C 112 12 11
39 3:9 C 11 1 12 10
37 3:9 C, NA 108 15 1 1
47 3:9 C 117 13 12
88 3: 10 C 108 15 12
44 3:1 1 AA; C 109 19 15
90 3: 11 C 109 1 1 13
79 3:11 C 1 12 14 12
26 4:0 C 101 10 12
89 4:1 C 1 10 9 10
77 4:2 As, O, H 105 9 11
75 4:2 C 114 15 13
74 4:2 C 1 16 15 18
10 4:2 C, H 97 8 1O
31 4:2 C, NA 127 15 14

160

Table C1 (cont)

 

 

PPVT-Ill CELF-P Sent CELF-P Word
Standard Score Struc. Subtest Struc. Subtest
(Mean=100, Standard Score Standard Score
Child ID Age Sex Ethnicity SD=15) (Mean=10, SD=3) (Mean=10, SD=3)
29 4:2 C 1 14 1 1 11
15 4:2 C 118 10 12
99 4:3 C 127 13 14
98 4:3 C 13 11
13 4:3 C 1 18 9 7
30 4:3 C 109 13 9
8 4:4 C 107 9 1 1
32 4:4 C 107 10 13
60 4:5 C 103 1 1 13
18 4:5 C 108 11 12
28 4:6 C 101 18 15
69 4:6 C 92 14 9
9 4:6 C 120 14 15
95 5:0 C 97 7 7
106 5:1 C 1 13 13 12

161

Table C1 (cont)

 

 

PPVT-lll CELF-P Sent. CELF-P Word
Standard Score Struc. Subtest Struc. Subtest
(Mean=100, Standard Score Standard Score
Child ID Age Sex Ethnicity SD=15) (Mean=10, SD=3) (Mean=10, SD=3)
50 5:1 F AA, C 121 13 14
1 1 5:1 M C 127 11 14
1 52 F C 103 8 10
66 5:2 M C 107 1 1 9
33 5:2 M C 98 11 11
82 5:3 F C 128 15 12
78 5:3 M AA, C 1 12 11 12
87 5:4 M C 126 13 14
49 5:5 M C 113 13 14
104 5:7 M C 93 12 11
137 5:8 F C 108 8 15
107 5:8 F C 109 8 8
102 5:8 F C 108 10 11
131 5:9 F C 132 10 11
17 5:9 F C 1 14 13 16

162

Table C1 (cont)

 

 

PPVT-III CELF-P Sent. CELF-P Word
Standard Score Struc. Subtest Struc. Subtest
(Mean=100, Standard Score Standard Score
Child ID Age Sex Ethnicity SD=15) (Mean=10, SD=3) (Mean=10, SD=3)

132 5:10 C 103 10 10
2 5:10 C 110 14 11
136 5:11 C 128 8 15
130 5; 11 C 116 10 10
108 5:11 C 104 10 13
128 6:0 C 142 13 14
110 6:0 C 104 7 7
109 6:0 C 95 7 8
83 6:0 C 114 12 13
3 6:0 C 114 7 10
124 6:1 C 104 9 10
135 6:1 C 116 13 10
120 6:2 C 112 13 10
114 6:2 AA 96 7 8
101 6:2 C 114 7 9

163

Table Ci (cont)

 

 

PPVT-Ill CELF-P Sent. CELF-P Word
Standard Score Struc. Subtest Struc. Subtest
(Mean=100, Standard Score Standard Score
Child ID Age Sex Ethnicity SD=15) (Mean=10, SD=3) (Mean=10, SD=3)
125 6:2 M H 1 15 1 1 12
123 6:3 F C 129 9 12
127 6:4 M C 125 1 1 10
1 13 6:5 M C 98 13 9
121 6:5 F C 1 1 1 1 1 10
126 6:9 M C 108 10 1 1
103 6:10 F NA/C 114 10 9

 

Note. AA = African American; As = Asian: H = Hispanic: NA = Native American

164

 

APPENDIX D

Qualified Responses

Table D1: Children’s Qualified Responses

 

 

Child ID Age Item # Question Response

1 5:2 30 Did the man fix two dolls? Uh uh, one.

3 6:0 33 Did the woman build the Yes, but only one though.
houses?

3 6:0 20 Did the woman push over the Yes, only one though.
dogs?

8 4:4 33 Did the woman build the Yes but the blue one didn’t.
houses?

10 4:2 13 Did the woman eat up the Yeah but he not eat one brownie.
brownies?

10 4:2 25 Did the woman drink up the Uhhuh, just one soda.

sodas?

165

Table D1 (cont)

 

 

Child ID Age Item # Question Response

10 4:2 7 Did the woman carry the Yes one bag just not two bags.
bags?

12 3:8 9 Did the woman drink two Yes, one.
sodas?

12 3:8 13 Did the woman eat up the Yes, one.
brownies?

13 4:3 21 Did the woman build the Yeah, just one house again.
houses?

13 4:3 9 Did the man eat two Yeah, he just ate one brownie.
brownies?

13 4:3 8 Did he woman push over the Yeah, she pushed over one dog.
dogs?

14 3:5 14 Did the woman eat up two Yes, she didn’t eat the last one.

brownies?

166

Table D1 (cont)

 

 

Child ID Age Item # Question Response

14 3:5 33 Did the man eat two He’s not gonna eat that yeah.
brownies?

16 3:4 15 Did the man carry out the Yes but not that bag layinbg on
bags? the ground.

18 4:5 5 Did the man drink the sodas? Mm-hmm, only one.

18 4:5 9 Did the man eat two Urn huh yes but guess what he
brownies? only ate one half of it.
Did he woman push over the

18 4:5 8 Uh huh just one.
dogs?

24 3:5 25 Did the woman eat up the Yeah, there's one.
brownies?

30 4:3 13 Did the woman eat up the Yes, just one.
brownies?

33 5:2 27 Did the woman push over two Yes but she just knocked one dog
dogs? off.

167

Table D1 (cont)

 

 

Child ID Age Item # Question Response

33 5:2 32 Did the woman push over the Yes, just one dog.
dogs?

34 3:7 8 Did the man eat the Yeah, just one.
brownies?

34 3:7 13 Did the woman eat up the Yeah, just one.
brownies?

39 3:9 15 Did the woman build two Yeah, not the other one.
houses?

48 3:7 37 Did the man drink up two (nods head yes) 1 soda.
sodas?

52 3:2 35 Did the man fix the dolls? (nods yes) But he didn’t get the

toes on her.
53 3:6 8 Did he woman push over the No, yes, she pushed one down

dogs?

168

but not the other.

Table D1 (cont)

 

Child ID Age Item# Question Response

 

60 4:5 15 Did the woman build two Yeah, didn’t finish the rest of it.

houses?

72 3:7 14 Did the woman eat up two Yeah but she left one.

brownies?

 

72 3:7 37 Did the woman drink up the Yeah, one is left.

sodas?

74 4:2 8 Did he woman push over the Yes, well she pushed one down.
dogs?

80 4:7 35
Did the man fix the dolls? Yes, who well I mean one.

80 4:7 20 Did the man eat the Yeah but he eat one.
brownies?

80 4:7 25 Did the woman eat up the Yeah, her heat all of one.
brownies?

169

Table D1 (cont)

 

 

Child ID Age Item # Question Response

80 4:7 37 Did the woman drink up the Yes, only one.
sodas?

80 4:7 19 Did the woman any the Yeah , but her carried only one.
bags?

87 5:4 33 Did the woman drink two Yeah, but not both of ‘am.
sodas?

89 4:1 1 1
Did the man fix the dolls? Yep, but not that one.

90 3:11 9 Did the woman drink two Yes, only one of um.
sodas?

90 3:11 35
Did the man push the dogs? Yes only one of um.

90 3:11 36 Did the woman carry two Yes, only one of ‘em.
bags?

98 4:3 33 Did the woman build the Yep only one.

houses?

170

Table D1 (cont)

 

 

Child ID Age Item # Question Response
99 4:3 39 Did the woman build two Yeah, but one she couldn't finish:
houses? the white one she couldn’t finish.
104 5:7 17 Did the man drink the sodas? Yes, that one wasn't done (points
to bottle with soda in it).
106 5:1 35
Did the man fix the dolls? Yeah, but one is still broke.
106 5:1 39 Did the woman build two Yeah, but he drinked one and a
houses? half.
106 5:1 39 Did the woman build two
Yeah, but one is still half broke.
houses?
106 5:1 17 Did the man drink up two Yes, he drinked a half of one.
sodas?
106 5:1 20 Did the man eat the Yes, but he didn’t eat half, not all
brownies? of one.

171

Table D1 (cont)

 

 

Child ID Age Item # Question Response

106 5:1 21 Did the woman drink two Yes, but one she didn’t drunk all
sodas? of it.

106 5:1 25 Did the woman eat up the Yep but she only ate half of one,
brownies? she ate one and a half.

106 5:1 33 Did the man eat two Yep, but he ate one and a half.
brownies?

106 5:1 37 Did the woman drink up the Yeah, but she drinked up one and
sodas? a half.

106 5:1 19 Did the woman carry the Yes, but she left one in.
bags?

106 5:1 23 Did the man carry out two Yep, but he left one there.
bags?

Yep, but one was too tired to
107 5:8 27 Did the woman build two

houses?

172

build that house cause she just

built a tiny bit and left it for a ?

Table D1 (cont)

 

Child ID Age Item # Question Response
107 5:8 33 Did the woman build the Yep but she was too tired to build
houses? that one.
107 5:8 8 Did the man eat the Yes but there is just one left that
brownies? he didn’t eat yet.
107 5:8 9 Did the woman drink two Yes but one is still Iefi.
sodas?
107 5:8 17
Did the man drink the sodas? Yes, but one is left.
Did the man eat two Yes, but he just took one like of
107 5:8 21
brownies? that one and that one is left for
later.
107 5:8 32 Did the man drink up two Yep but ones left.
sodas?
107 5:8 11 Did the man carry out two Yes, but one is still there.
bags?

173

Table D1 (cont)

 

 

Child ID Age Item # Question Response
107 5:8 20 Did the woman push over the Yep, just one ‘cuz he wanted to
dogs? see and he wanted to take?
114 6:2 9 Did the woman build the
Yes, one and then 2 half houses.
houses?
120 6:2 8 Did the man eat the Yes, half of one.
brownies?
130 5:11 27 Did the woman build two Yes and no.
houses?
130 5:11 30 Yes and no.
Did the man fix two dolls?
130 5:11 33 Did the woman build the Yes and no.
houses?
130 5:11 25 Did the woman drink up the Yes and no.
sodas?

174

Table D1 (cont)

 

 

Child ID Age Item # Question Response
130 5:11 38 Did the woman eat up two Yes and no.
brownies?
Did the man carry out the
130 5; 11 39 Yes and no.

bags?

175

REFERENCES

Antinucci, F., & Miller, R. (1976). How children talk about what happened.
Joumal of Child Language, 3, 169-189.

Avrutin, 3., 8r Wexler, K. (1999). Children's knowledge of subjective clauses:
Obviation, binding, and reference. Language Acquisition: A Joumal of
Developmental Linguistics, 8, 69-102.

Bach, E. (1986). The algebra of events. Linguistics and Philosophy, 9, 5-16.

Bloom, L., Lifter, K., & Hafitz, J. (1980). Semantics of verbs and the development
of verb inflection in child language. Language, 56, 386-412.

Bohnemeyer, J. & Swift, M. (2004). Event realization and default aspect.
Linguistics and Philosophy, 27, 263-296.

Borer, H. (2005). Structuring sense: The normal course of events. Oxford; New
York Oxford University Press.

Bronckart, J. P., & Sinclair, H. (1973). Time, tense and aspect. Cognition Vol, 2,
107-1 30.

Comrie, B. (1976). Aspect. Cambridge, UK: Cambridge University Press.

Crain, S. (1991). Language acquisition in the absence of experience. Behavioral
and Brain Sciences, 14, 597-650.

Dale, P. S., & Fenson, L. (1996). Lexical development norms for young children.
Behavior Research Methods, Instruments, & Computers, 28, 125-127.

Dehe, N. (2002). Particle verbs in English: Syntax, information structure and
intonation. Amsterdam: Philadelphia: John Benjamins.

Dowty, D. (1979). Word meaning and Montague grammar. Dortrecht: Reidel.

Dunn, L. M., & Dunn, L. M. (1997). Peabody Picture Vocabulary Test - Third
Edition ( ed.). Circle Pines, MN: American Guidance Service.

Fritzley, V. H. and Lee, K. (2003). Do young children always say yes to yes-no
questions? A metadevelopmental study of the affirrnation bias. Child
Development, 74, 1297-1 313.

Gelman, R., & Gallistel, C. R. (1978). The child's understanding of number.
Oxford: Harvard University Press.

176

Gordon, P. (1996). The truth-value judgment task. In D. McDaniel & C. McKee
(Eds), Methods for assessing children ’9 syntax. (pp. 211-231).
Cambridge, MA, US: The MIT Press.

Green, L. (2002). African-American English: A linguistic introduction. Cambridge:
Cambridge University Press.

Hadley, P. A. (1998). Early verb-related vulnerability among children with specific
language impairment. Journal of Speech, Language, and Hearing
Research, 41, 1 384-1397.

Hamer, L. (1981). Children talk about the time and aspect of actions. Child
Development, 52, 498-506.

Heim, I. (1991). Artikel und definitheit IArticles and definiteness. (Read in English
Translation). In. A. V. Stechow & D. Wunderlich (Eds), Semantics: An
international handbook of contemporary research, Berlin: de Gruyter.

Horn, L. R. (2004). lmplicature. In L. R. Horn & G. Ward (Eds), The Handbook of
pragmatics (pp. 3-28). Malden, MA: Blackwell

Johnson, B.-W., & Fey, M.-E. (2006). Interaction of lexical and grammatical
aspect in toddlers' language. Journal of Child Language, 33, 419-435.

Kadmon, N. (2001). Formal pragmatics. Malden, MA: Blackwell Publishers, Inc.

Kanniloff-Smith, A. (1979). A functional approach to child language: A study of
detenniners and reference. Cambridge: Cambridge University Press.

Kelly, D.-J., & Rice, M.-L. (1994). Preferences for verb interpretation in children
with specific language impairment. Journal of Speech and Hearing
Research, 37, 182-1 92.

Klein, W. (1994). Time in language. London, England: Routledge.

Krifi<a, M. (1989). Nominal reference, temporal constitution and quantification in
event semantics. In R. Bartsch, J. van Benthem & P. van Emde Boas
(Eds), Semantics and contextual expression (pp. 75-115). Dortrecht:
Fons.

Leonard, L. B., Bortolini, U., Caselli, M. C., McGregor, K. K., & at al. (1992).
Morphological deficits in children with specific language impairment: The
status of features in the underlying grammar. Language Acquisition: A
Journal of Developmental Linguistics, 2, 151-179

177

Leonard, L., B., Deevy, P., Kurtz, R., Krantz, L., Owen, A., Polite, E., et al. (in
press). Lexical aspect and the use of verb morphology by children with
specific language impairment. Journal of Speech, Language, and Hearing
Research.

Leonard, L. B., Deevy, P., Miller, C., Charest, M., Kurtz, R., & Rauf, L. (2003).
The use of grammatical morphemes reflecting aspect and modality by
children with specific Iangauge impairment. Joumal of Child Language, 30,
769-795.

Leonard, L. B., Eyer, J. A., Bedore, L. M., 8r Grela, B. G. (1997). Three accounts
of the grammatical morpheme difficulties of English-speaking children with
specific language impairment. Journal of Speech, Language, and Hearing
Research, 40, 741-753.

Li, P., & Shirai, Y. (2000). The acquisition on lexical and grammatical aspect.
Berlin; New York: Mouton de Gruyter.

Maratsos, M. (1976). The use of definite and indefinite reference in young
children: An experimental study of semantic acquisition. Cambridge:
Cambridge University Press.

MacWhinney, B. (2000). The CHILDES Project: Tools for analyzing talk, Vol 2:
The database ( 3rd ed.). Mahwah, NJ, US: Lawrence Erlbaum Associates,
Publishers.

Munn, A., Miller, K., & Schmitt, C. (2006). Maximality and plurality in children's
interpretation of definites. Proceedings of the Annual Boston University
Conference on Language Development, 30, 377-387.

Ogiela, D. A., Schmitt, C., & Casby, M. W. (2005). Verb phrase telicity and verbal
morphology in children with specific language impairment. Proceedings of
the Annual Boston University Conference on Language Development, 29,
424-435.

Perez-Leroux, A. T., Munn, A., Schmitt, C., & Delrish, M. (2004). Learning
definite detenniners: genericity and definiteness in English and Spanish
[Electronic Version]. Boston University Conference on Language
Development Proceedings Supplement, 28, 1-12. Retrieved November 1,
2006 from http:/lwww.bu.edullinguistics/APPLIEDIBUCLDIsupp.html.

Pinker, S. (1989). Leamability and Cognition. Cambridge, MA: MIT Press.

Rice, M.-L., & Bode, J.-V. (1993). GAPS in the verb Iexicons of children with
specific language impairment. First Language, 13, 113-131.

178

Rice, M. L., 8 Wexler, K. (1996). Toward tense as a clinical marker of specific
language impairment in English-speaking children. Joumal of Speech and
Hearing Research, 39, 1239-1257.

Rice, M. L., Wexler, K., & Cleave, P. L. (1995). Specific language impairment as
a period of extended optional infinitive. Journal of Speech and Hearing
Research, 38, 850-863.

Schmitt, C. J. (1996). Aspect and the syntax of noun phrases. (Doctoral
Dissertation, University of Maryland at College Park, 1996) Dissertation
Abstracts lntemational: The Humanities and Social Sciences, 57, 4350-A.

Schulz, P., & Wenzel, R. (2005). Acquisition of compositional telicity by German-
speaking children with specific language impairment. In van Hout (chair),
crosslinguistic acquisition of telicity. Symposium conducted that the Xth
lntemational Congress for the Study of Child Language, Berlin, Germany.

Seymour, H. N., Roeper, T. W., & de Vllliers, J. (2003). Diagnostic Evaluation of
Language Variation - Screening Test. San Antonio, TX: The Psychological
Corporation.

Shirai, Y., & Andersen, R. (1995). The acquisition of tense-aspect morphology: A
prototype account. Language, 71, 743-762.

Smith, C. (1991). The parameter of aspect. Dord recht: Kluwer.

van Hout, A. (1998). On the role of direct objects and particles in Ieaming telicity
in Dutch and English. Proceedings of the Annual Boston University
Conference on Lanuguage Development, 22, 397-408.

Verkuyl, H. J. (1972). On the compositional nature of aspects. Dordrecht: Kluwer.

Verkuyl, H. J. (1989). Aspectual classes and aspectual composition. Linguistics
and Philosophy, 12, 39-94.

Verkuyl, H. J. (1993). A theory of aspectuality. Cambridge: Cambridge University
Press.

Wagner, L. (2001). Aspectual influences on early tense comprehension. Journal
of Child Language, 28, 661-681.

Wagner, L. (2006). Aspectual bootstrapping in language acquisition: Telicity and
transitivity. Language Learning and Development, 2, 51-76.

179

Wagner, L. & Carey, S. (2003). lndividuation of objects and events: A
developmental study, Cognition, 90, 163-191.

Wason, P. C. (1961 ). Response to affirrnative and negative binary statements.
British Journal of Psychology, 52, 133-142.

Watkins, R. V., Rice, M. L., & Moltz, C.-C. (1993). Verb use by language-
impaired and normally developing children. First Language, 13, 133-143.

Weist, R. M., Wysocka, U., Wrtkowska-Stadnil, K., Buczowska, E., & Konieczna,
E. (1984). The defective tense hypothesis: On the emergence of tense
and aspect in child Polish. Joumal of Child Language, 11, 347-374.

Wiig, E. H., Secord, W., & Semel, E. (1991). Clinical Evaluation of Language
Fundamentals - Preschool. San Antonio, TX: The Psychological
Corporation.

Wynn, K. (1990). Children's understanding of counting. Cognition, 36, 155-193.

Zribi-Herz, A. (2003). Pur une analyse unitaire de de partitif. In F. Corblin and L.

Kupfeman (Eds), Proceeding of the lndefinis et predication conference.
Paris: Presses de la Sorbonne.

180

   

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

...............
.....................