. ‘ , E‘ 4 , v
, 3...? tufiéwoflwrfi 2mg»? , “a.

“r .
Raw».

.3.
A.

s a... 3m.

,

35mm. .1 fiimflnfi .
fig.
.~ .V .
ax
mum, .

2...?

’A.

Jun .. m
mm... “m . I u » . .
we. 1 . . . U . .
‘ . . . . ‘ gut.
, . . . ‘ “I. . .1 O
A .u ‘ . . . keg...

( flu.
33”.“an
, .

, @i

t» . ~ 0

1 grin?!

.r... A. {In}:

:4. _
. .V . “.
ifiupwam.

‘ :43. h.
. mmmfi.
III I

n ahumnaunpn.

. kw.
.324

{655; R.
.l

. ‘nnu... .
1 .L.
: Wt?

.3.

 

 

 

 

THESIS

 

 

 

 

3
’°(
A Cl L'BRARY
Mlchigan State
'niversity
This is to certify that the

dissertation entitled

Sound Change in a Regional Variety of
Cuban Spanish

presented by

Gabriela G. Alfaraz

has been accepted towards fulfillment
of the requirements for

JED. degreein_Lj_nQ.u_l_S_tj_C5

 

 

 

 

 

Date December 14, 2000

MS U is an Affirmative Action/Equal Opportunity Institution O~ 1 2771

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

 

MOM? 1053

 

Z
O
<
e
A
. N
C)
C)
m

 

 

 

 

 

 

 

 

 

 

 

 

 

SOL'XD Cl

This
itegional
iiextema
ofi‘ne cur
that the 1]
preiiousi
5?iech of
secixpogi

ABSTRACT
SOUND CHANGE IN A REGIONAL VARIETY OF CUBAN SPANISH
By
Gabriela G. Alfaraz

This dissertation investigates the variation of (s), (r) and (l) in
a regional variety of Cuban Spanish, and examines the contribution
of external and internal factors to the variation. One of the objectives
of the current study is to examine generational changes. It is found
that the three variables have been involved in change. Variants
previously associated with marginalized groups have increased in the
speech of young college-educated adult speakers. It is proposed that
sociopolitical changes created conditions which motivated and
strengthened the diffusion of variants previously associated with
working class and rural speech. I argue that a new sociolinguistic
norm arose in Cuba as a result of sociopolitical changes. Conformity
with the new norm is reinforced by the fact that individuals must
show their allegiance and commitment to the revolution in order to
succeed in their studies and their work. It is suggested that working
class and rural features are associated with the new nationalism and
express integration and solidarity.

Another objective of this dissertation was to provide a
descriptive base for the internal linguistic constraints on the
variation of (s), (r) and (l) in this variety of Spanish, since it has not

been previously studied within this theoretical framework. An

imponantfini
uednunCISc
Monuicoutc
hcmrscontr
haninthe u
manhnguiSI
unauonin

ofCanbbear

important finding is that assimilation of the lateral and rhotic liquids
are distinct sociolinguistic processes, although they have the same
phonetic outcome. Another interesting finding is that linguistic
factors contribute more to variation in the word-internal environment
than in the word-final one, where social factors are more important
than linguistic ones. It is shown that the internal constraints on
variation in Cuban Spanish are different from those in other varieties

of Caribbean Spanish.

Dedica

Dedicated to my mother, Gladys Rojas, whose energy and love

for life have always inspired me.

iv

 

LIST OF HOT".
LIST OF TABL

lNiRODL'CTli
Research (
Spanish D
Origins of
Regional :
Histo:

Varia

(r) in

The
REgiona'
Cub;

Soc

TABLE OF CONTENTS

LIST OF FIGURES ........................................................... viii
LIST OF TABLES ................ . ............................................. ix
INTRODUCTION ................................................................ 1
Research Questions and Other Goals of this Dissertation ...... 1
Spanish Dialect Variation: Coda Consonants ...................... 3
Origins of Cuban Spanish ................................................ 7
Regional and Social Variation in the Caribbean .................. 9
Historical Background of /s/ and /r/ ............................ 9
History of Aspiration and Deletion of /s/ ......... 9

History of Lateralization and Aspiration of /r/. 11
Diffusion of Change: From Andalusia to

America .................................................... 11

Variation of /s/ in the Caribbean ................................ 12

(r) in Caribbean Varieties ......................................... 13
Lateralization and Aspiration ....................... 14

Other Regional Variants of (r) ...................... 15

The Status of Other Coda Consonants ......................... 16
Regional and Social Variation in Cuba ............................. 17
Cuban Regional Varieties ......................................... 17

The Role of (s) in Regional Variation in Cuba.21
The Role of (r) in Cuban Regional Variation ...21

Social Dialects ....................................................... 22

(3) Variation and Social Stratification in Cuba 22

Social Variation and (r) in Cuba .................... 23

Summary: Regional and Social Variants ...................... 24
Sociolinguistic Expectations of the Current Study .............. 25
METHODOLOGY ............................................................... 27
The Social Group Versus the Individual ............................ 27
Contextual Constraints on Linguistic Variation .................. 28
Internal Constraints on Variability ............................. 28
External Constraints on Variability ............................ 29
The Expression of Variable Rules: VARBRUL ................... 30
The Current Study: Method and Procedures ....................... 31
The Selection of Variants ......................................... 31
Internal Constraints on Variability ............................. 33
Sociolinguistic Constraints on Variability ................... 35
Data Collection ....................................................... 39
The Respondents ..................................................... 40
Region .................................................................. 43
Analysis ................................................................ 44

 

iHEVARlABl

Backgroun
Rfluhsan.
hind-'

“ind-

Summary

THEVARIAB
Pounoni
lnfini

Asunu1at
Back,
Resul

Cone
Lateraliz
Back
Resu

Con:
ASpuanc
Back
Resu
SUmp
The Basi<
Resul

THE VARIABLE (s) ........................................................... 46

Background ................................................................. 46
Results and Discussion .................................................. 47
Word-internal (s) .................................................... 47
Linguistic Factors ....................................... 47
Social Factors ............................................ 49
Word-final (s) ........................................................ 55
The Effects of Linguistic Factors on the

Variation of Word-final (s) .......................... 56

The Effects of Social Factors on the Variation
of Word-final (s) ........................................ 62
Inflectional (s): Linguistic Factors ................ 67
Summary Conclusions .................................................... 71
THE VARIABLE (r) ............................................................ 76
Position in the Word ..................................................... 77
Infinitive + Clitic Combinations ................................ 77

Distribution and Frequency of (r) Variants in
Word-internal, Word-final and Infinitive +

Clitic Environments .................................... 79
Assimilation ................................................................ 80
Background ............................................................ 81
Results and Discussion ............................................. 82
Linguistic Factors: Word-internal Position ...... 85

Linguistic Factors: Word-final Position: ......... 92

Social Factors and (r) Assimilation ................ 96

Conclusion ........................................................... 103
Lateralization ............................................................ 103
Background .......................................................... 103
Results and Discussion ........................................... 104
Following Environment ............................. 106

Preceding Sound ....................................... 110

Stress ..................................................... 113

Morpheme Type ........................................ 113

Social Factors .......................................... 114

Conclusion ........................................................... 123
Aspiration ................................................................. 124
Background .......................................................... 124
Results and Discussion ........................................... 125
Summary Conclusions ............................................ 132
The Basic Variant of (r) ............................................... 133
Results and Discussion ........................................... 134
Following Environment ............................. 135

Preceding Environment .............................. 138

Stress ..................................................... 141

Boundary Type ......................................... 141

Morpheme Type ........................................ 142

vi

Concl
Summary
FoHo‘

Piece
Stres
Socia
lHEVAthl

Results 3
Follc
Prec

Strei
Age.

Conclus

StuquY
BIBLIOGRA

Social Factors .......................................... 143

Conclusion ........................................................... 147
Summary Conclusions .................................................. 147
Following Phonological Environment ........................ 147
Word-internal (r) ...................................... 148

Word-final (r) .......................................... 150

Preceding Phonological Environment ........................ 152

Stress .................................................................. 153

Social Factors ....................................................... 154

THE VARIABLE (l) .......................................................... 155
Results and Discussion ................................................ 158
Following Environment .......................................... 159
Preceding Segment ................................................ 161

Stress .................................................................. 164

Age, Gender and Status .......................................... 164
Word-internal (l) ...................................... 164

Word-final (1) .......................................... 169

Conclusion ................................................................ 171
SUMMARY CONCLUSIONS .............................................. 174
BIBLIOGRAPHY .............................................................. 180

vii

Figure 1 Re
1986133

the curr

FizurH T
void-it

lirure.‘ I

“‘

Mid-it

turd-f

F321 r

assimi

01 the

LIST OF FIGURES

Figure 1. Regional varieties as classified prior to redefinition in
1986 by Choy Lopez ...................................................... 19

Figure 2. Reclassification of regional varieties by Choy Lopez (1986,
1994) .......................................................................... 20

Figure 3. The central dialect area with the area which is the object of
the current study marked with shading. ............................. 26

Figure 4. The contribution of gender and status to the application of
word-internal aspiration ................................................. 51

Figure 5. The contribution of status and origin to the application of
word-internal ............................................................... 53

Figure 6. Age and gender and the application of word-final (s) ..63

Figure 7. The contribution of gender and status to the application of
word—final (s) ............................................................... 65

Figure 8. Social variables and word-internal application of (r)
assimilation ................................................................. 97

Figure 9. Social variables and word-final application of (r)

lateralization .............................................................. 118
M Aspirated variant of (r): Gender, age and status ...... 131

Figure 11. Contributions of age, gender and status to the application
of the basic variant in word-internal position ................... 144

Figure 12. Social variables and word-internal application of lateral
assimilation ............................................................... 166

Figure 13. Probability weights for word-internal assimilation of (l)
and (r) ...................................................................... 168

Figure 14. The word-final application of assimilation according to
gender, status and age .................................................. 170

viii

 

 

Table 1 Liquid
um .

Tuhl VARE
Table 3 Varrar
Tahlei Xumh
TallcS The S
Table6 Educ
Table? Occu
Table 8 index
Tahle9 Resp
Table 10 We:
Talle 11 “'0
Table 12 W0

Table 13 We
Linguisu

Taile14 “C

Table 15 Eta
Table 16pm
““8 17 lnfi
Table 13 (r)

”we 30 Sig

We” Fell

LIST OF TABLES

Table 1 Liquid assimilation in a variety of Cuban Spanish Guitart

(1976) .......................................................................... 6
Table 2 VARBRUL Weight Values ....................................... 31
Table 3 Variants of (r), (l) and (s) ........................................ 33
Table 4 Number and Person Marking on Verbs ........................ 35
Table 5 The Social Variables ............................................... 36
Table 6 Education Index ..................................................... 38
Table 7 Occupational Index ................................................. 38
Table 8 Index of Social Rank ............................................... 39
Table 9 Respondents .......................................................... 42
Table 10 Word-internal Aspiration: Stress ............................. 49
Table 11 Word-internal Aspiration: Age ................................ 50
Table 12 Word-final (s): Linguistic Factors ........................... 57

Table 13 Word-final (s) in Plural and Verb Grammatical Morphemes:
Linguistic Factors Involved in the Application of Aspiration 68

Table 14 Word-final Inflectional (s): the Effects of Social Factors

on the Application of the Aspirated Variant ....................... 70
Table 15 Examples of the Variants of (r) ................................ 76
Table 16 Phonological Surface Structure of Spanish Clitics ....... 78

Table 17 Infinitive + Clitic as a Word-internal Environment for /r/79
Table 18 (r) Variants Word-internally and Word-finally ............ 80

Table 19 Significant Factor Groups for Word-internal Assimilation
.................................................................................. 82

Table 20 Significant Factor Groups for Word-final Assimilation84

Table 21 Following Environment: Word-internal Assimilation....86

ix

hMeEZThe

TaMeZi Per
and Age

Thh24Pm
hhefi “x
TaHeZb “in
Tdhlihb
T351638 ll'o
Tdhlan
Tflh3OWt
Tmh3ch
Table 32 \lt
hHeSSAg
Tnh34uj
Thh35uy
Tdh363i
ka37“-
Table 33 “-
Tab1839 L
Table 4,2. L

I
i

Tillie 41 ll

Table 4: I.

i
Tw
nine 44 F<

.‘
j 1

”bit 46 A.

Table 22 The Contribution of Voicing and Place to Assimilation 87

Table 23 Percent of Word-internal Assimilation by Following Sound
and Age ....................................................................... 88

Table 24 Preceding Sound: Word-internal Assimilation ............. 89
Table 25 Word-internal Assimilation by Preceding Sound and Age90
Table 26 Word-internal Assimilation and Stress ...................... 91
Table 27 Word-internal Assimilation: Syllable vs. Clitic Boundary92

Table 28 Word-final Assimilation and Following Environment ...93

Table 29 Preceding Sound and Word-final Assimilation ............ 95
Table 30 Word-final Assimilation: Stress ............................... 95
Table 31 Word-final Assimilation: Morpheme Status ................ 96
Table 32 Word-final Assimilation: Gender ............................ 101
Table 33 Age and Word-final Assimilation ........................... 102
Table 34 Word-final Assimilation: Status ............................. 102

Table 35 Word-internal Lateralization: Significant Factor Group5105
Table 36 Significant Factor Groups for Word-final Lateralization106
Table 37 Word-internal Lateralization: Following Environment 107

Table 38 Word-final Lateralization: Following Environment.... 109

Table 39 Lateralization: Preceding Sound ............................ 112
Table 40 Lateralization and Stress ...................................... 113
Table 41 Word-final Lateralization: Morpheme Status ............ 114
Table 42 Lateralization and Origin ...................................... 115
Table 43 Word-internal Lateralization: Gender and Age .......... 117

Table 44 Following Environment and Word-internal Aspiration 126
Table 45 Word-internal Aspiration: Preceding Segment .......... 128

Table 46 Aspiration and Stress ........................................... 129

Table 47 Wow
hHeiSAsfl
ka49 $gr
hkaO FoHt
hMeSlher
TaHeSZ Prec
ThhSSSue
Table 54 Wet
Table 55 We:
Thkibhbr

Table 57 F01

hHeSSWnr
Tdk59 Pre
Tukéo “0
hhe6lrhe

Table 62 Sir
find(1,‘

Lille 64 AS

Table 65 STI'Q

Table
Table
Table
Table
Table
Table
Table
Table
Table
Table

Table

Table
Table
Table

Table

47 Word-internal Aspiration: Syllable vs. Clitic Boundary129

48 Aspiration and Origin ........................................... 132
49 Significant Factor Groups for the Basic Variant ....... 135
50 Following Environment and Word-internal [r] ......... 136
51 Word-final [r]: Following Environment .................. 137
52 Preceding Sound and Application of the Basic Variant140
53 Stress and the Basic Variant .................................. 141
54 Word-internal [r]: Syllable vs. Clitic Boundary ....... 142
55 Word-final [r]: Morpheme Type ............................ 142
56 Word-final Basic Variant: Age, Gender and Status 146

57 Following Sounds Favoring (r) Variants Word-internally
............................................................................. 149

58 Word-final Variants of (r) and Following Environment15l

59 Preceding Segment in Word-internal Application of (r)152
60 Word-final Application of (r) and Preceding Sound .. 153
61 The Effects of Stress on Variation of (r) ................. 153

Table 62 Significant Factor Groups for Word-internal and Word-
final (1) ..................................................................... 158

Table 63 Application of Assimilation by Following Environment160

Table
Table

64 Assimilation of (1) and Preceding Sound ................. 163

65 Stress and the Application of Assimilation ............... 164

xi

This (ll:

 

change in a re,

examines the l

 

 

nnumgmsnc
research were
perceptions of
ontheisland (
uflesand nev
negatively on
Tamed quegrjg
theertenr anc
generation, ut-
Theugaung

Further
research 00 (
hdmminpz
briefPeriod '
fiandoned Si
Cuba‘ “'hlch
Dill, mUCh 0‘.

‘ l

kwhofemi

RESea
This st
are,” the r
Sykbh'fina

INTRODUCTION

This dissertation studies the degree and direction of sound
change in a regional variety of central Cuba. This research also
examines the impact of social mobility and social class leveling on
sociolinguistic stratification. The questions addressed in this
research were motivated by the results from a study on the
perceptions of Cubans in Miami toward the variety of Spanish spoken
on the island (Alfaraz, in press), which revealed that both long—term
exiles and newer arrivals tend to evaluate the Spanish spoken in Cuba
negatively on both status and solidarity dimensions. These findings
raised questions about linguistic change in Cuba, particularly about
the extent and types of changes that have occurred within the past
generation, which will be examined in this dissertation by
investigating sound changes in a regional Cuban variety.

Furthermore, the questions addressed in this study fill gaps in
research on Cuban Spanish in general, and its regional and social
dialects in particular. Such research was vigorously carried out for a
brief period in the late 1960’s and early 1970’s, but has been largely
abandoned since that time, except for some work by linguists in
Cuba, which is mostly unpublished. As Choy Lopez (1994) points
out, much of recent dialectological work has been discontinued as a

result of emigration and lack of resources.

Research Questions and Other Goals of this Dissertation

This study examines three sociolinguistic variables in order to
address the four research questions listed below. The role of
syllable-final (r), (l) and (s) in regional and social variation is

described in following sections of this chapter.

The first question concerns the extent of phonological change
in this variety. The objective is to determine the degree to which the
speech of younger and older Cubansl differs. A synchronic approach
is used to investigate this question: younger speakers are compared
to older ones in apparent time (Labov 1963, 1966/1982).2

The second research question addresses the direction of sound
change. This study seeks to identify the social and linguistic factors
involved in the diffusion of sound changes which have taken place
during the past thirty to forty years. Sound changes may be from
above or below (Labov, 1966/1982). Change from above tends to be
prescriptive and involves imposing changes which are considered
prestigious, or which serve a specific social purpose (i.e. an
ideological one, see Fairclough, 1989 and Kontra, 1992). Change
from above may also attempt to eradicate changes from below; in
other words, it aims to correct change from below, or those changes
that have entered the standard norm from groups of lower social
status.

Of interest to this study are the frequency and distribution of
previously stigmatized variants. Also relevant is whether the changes
were already in progress a generation ago, or whether they are new

ones initiated in the last twenty to thirty years. This question is of

 

lCuban refers to persons who acquired their native dialect in Cuba,
which does not include individuals who emigrated as children.
EMA-America; refers to individuals who acquired their native
dialect in the United States. Thus, yourgilubans here refers to new
arrivals in Miami who emigrated as young adults, and not to young
Cuban-Americans.

2I attempted a diachronic study with data from a large dialectology
study carried out in the 19603, but I did not have enough
demographic information to establish the necessary social variables.

I must.thank Terrell Morgan and Jorge Guitart for making the tape
recordings available to me.

particular int:

 

changes “ETC
during the pa:
question

The thi
sound change
upward socia
influenced ST

The th
Variation T‘
constraining
ei‘z'i'lTOTlmen

ln sur
””0 that
generation
Piogressed
let'eling ha

”Reunite 1

particular interest as it will shed light on whether phonological
changes were motivated by social changes that have taken place
during the past generation, which is addressed in the next research
question.

The third question focuses on the social determinants of the
sound changes and their diffusion. In particular, how have rapid
upward social mobility and the leveling of social class barriers
influenced sound changes and their spread?

The three previous questions focus on the social correlates of
variation. The fourth question is concerned with the internal factors
constraining variation. This study seeks to examine the linguistic
environments that favor application of the variants in question.

In summary the four research questions investigated here are:
(1) To what degree have sound changes progressed from one
generation to another; (2) In which direction have sound changes
progressed? (3) What effects have social mobility and social class
leveling had on sound changes and their diffusion? (4) What

linguistic factors correlate with variation?

Spanish Dialect Variation: Coda Consonants
Coda consonants tend to reflect dialect variation in Spanish.
The phonological processes that produce variation in syllable final
consonants are mainly reduction processes — lenition, assimilation
and deletion — which are thought to result from internal pressure
toward establishment of a CV syllable structure, and, ultimately, the
simplification and regularization of the phonological system (Choy

Lopez, 1986; Hammond, 1977; Poplack, 1981).

 

HOheie
been characte

(1989) points

    
   
 

It seems
not strut
and that
STTUCTUT.:
diachror.
80)3
It was only aft
lCiVCl'strucu
reduction bega
consonant :11”
only fire cons
Catalan (1939
“rim“ to a:

aeahening an

“lilies

is is most e.
that “ill uh;
some Varieti
Tromreduc1
and A“dalu

T 1
Tail {0 (‘qu

some “"211

,\

Ems and 1‘

11's

However, the historical development of Spanish has not always

been characterized by the reduction of coda consonants, as Catalan

(1989) points out:

It seems clear that old Spanish (as of the late XIV century) was
not structurally inclined to prefer the /CV/ syllable paradigm,
and that the tendency in modern Spanish cannot be considered a
structural characteristic that was consistently present in the

diachronic evolution of Spanish from Latin to the present. (p.
80)3

It was only after the XI century, when Spanish lost final vowels in
(C)VCV structures, resulting in (C)VC syllables, that coda consonant
reduction began. From around twenty-five possible consonants and
consonant cluster combinations at the end of the thirteenth century,
only five consonants appeared in the coda by the fourteenth century.
Catalan (1989) states that these five surviving coda consonants are in
“crisis” (p. 83) in many regional varieties of Spanish, with
Andalusian and Caribbean varieties displaying the most extensive
weakening and deletion (Catalan, 1989).

While some syllable-final variation can be explained as
weakening, not all of it can be attributed to reduction. Variation of
/s/ is most commonly considered a weakening process (5 -) h -) a)
that will ultimately eliminate /s/ from the phonological system of
some varieties. However, variation of /r/ does not always a result
from reduction. For instance, lateralization of /r/, found in Caribbean
and Andalusian varieties, does not arise from weakening and does not
lead to further weakening (Guitart, 1982; Poplack, 1981), although

some linguists claim otherwise (Alonso, 1961).

3This and many other of the following quotations are translations
from Spanish which I have made for this dissertation.

Assimil

 

consonant red
that in Cuban
general langu
since it is no
Assimilation
and effort th
pronunciatio
say.
al All
leaiin
There
the SU
onto 1
Elampies 0
Cuban Spar
will. are
Partial gen
seS-‘ttent T

CUM“ \‘ari

“hkh OCCY

Assimilation of /r/ in Puerto Rican Spanish is attributed to coda
consonant reduction (Poplack, 1981). However, Guitart (1982) argues
that in Cuban Spanish liquid assimilation cannot be explained as a
general language tendency to simplify syllable coda consonants,
since it is not a weakening process that minimizes articulatory effort.
Assimilation, Guitart points out, “can involve a degree of complexity
and effort that is equal to or greater than that of the original alveolar
pronunciation” (p. 65). He describes the process in the following

way.

a) All supraglottal features ofthe liquid are suppressed,
leaving only voicing which is realized as glottalization...b)
There is consonantal increase (the opposite of reduction), and
the supraglottal features of the second consonant are copied
onto the first; c) The articulation is fortis. (p. 66)

Examples of Guitart’s description of assimilation of the liquids in
Cuban Spanish, which is based on an urban middle class Havana
variety, are shown in Table 1. Assimilation may result in complete or
partial gemination, depending on the features of the following
segment. This description of liquid assimilation indicates that in the

Cuban variety this process is different from the complete gemination

 

which occurs in other varieties.

Table l

L Quid assii

 

T

hehpiEnteJ

[abbedto] 3

l
distal a;
55:4an Si

Table 1

Liquid assimilation in a variety of Cuba Spanish Guitart (1976)

 

[r]
[sebpisnte] serpiente 'serpent'
[afledto] Alberto 'Albert'

 

[pagke] 123—nuts 'park'
[baflero] barbegg 'barber'
[80140] some 'fat'

[pugga] was 'purge'
[seffino] ser fino 'be refined'
[hoh_he] Egg 'Jorge'
[semon] m 'sermon’

[sefijato] ser fiato 'be fiato'

[pednil] pernil 'pork'

[1]
[ebpwente] el puente 'the bridge'
[suedto] suelto 'loose'

[bagkon] balcon 'balcony'

 

[siflar] silbar 'whistle'

[cefla] celda 'cell'

 

[coggar] c_olg_a_lr 'hang'
[efieo] e1 feo 'the ugly one'

[eflefe] el jefe 'the boss'
[tamata] tal mata 'that tree'
[efi_fi_ame] el fiame 'the fiame'

[ed_nada] él nada 'he swims'

 

As for aspiration of /r/, Cedergren (1973) considered it a stage
in the weakening process, parallel to aspiration of /s/: r -) i -) h -)
a. After experimenting with statistical models of variable rule
ordering, however, Cedergren and Sankoff (1986) concluded that
aspiration of /r/ was an independent process not related to
spirantization or deletion of /r/, as suggested earlier by Cedergren.

In conclusion, the variation of /s/ is generally believed to be
motivated by reduction which simplifies and regularizes the
phonological system, but variation of /r/ is not always the product of
systemic reduction: lateralization, assimilation and aspiration of /r/
are separate processes unrelated to reduction caused by internal

pressure to establish open syllables.

Origins of Cuban Spanish

Consonant variation in Cuban Spanish, like other Caribbean
areas along the American trade-route, is believed to have originated
in the Andalusian variety. Phonological evidence (e.g. /6/ > /s/ and
/d2/ > /y/, /s/ aspiration and deletion, /r/ lateralization and
aspiration) suggests that the Andalusian variety was the principal
contributor to Caribbean dialect formation. Some linguists, however,
(i.e. Alonso, 1961) refute the monogenesis hypothesis, arguing that
some of these changes, lenition of the voiced coronal fricative, for
instance, appeared in America before they appeared in Spain.
Generally, however, it is accepted that many of the linguistic changes
that began in southern Peninsular varieties spread to the Caribbean
through intense and prolonged immigration. Moreover, not only were

Andalusians numerically predominant, during the eleventh, twelfth

 

and thirteenth
actively folio _

life of Spain.

  
  

Cadiz which \\
1181'. they part
(CatalaleSi'i
with large nuni
diffusion and \
hlontori
Cuban Spanisi
“WWI. sinc
iltm Andalus
immigration
consonantal .
Import
dfi'elopineni
dilitan i301
lililOlelOng
itiend “is
cImri‘bulit):

ia‘ttiaiTZati

and thirteenth centuries, the inhabitants of Caribbean port areas
actively followed and participated in the political and sociocultural
life of Spain, particularly of the Andalusian cities of Sevilla and
Cadiz which were focal areas of culture and civilization. In the same
way, they participated in the linguistic innovations of the time
(Catalan,1989). In Cuba, the prestige accorded Andalusia coupled
with large numbers of Andalusians created an ideal situation for the
diffusion and widespread acceptance of innovations from that region.

Montori (1916) described a second factor in the development of
Cuban Spanish: intense immigration from the Canary Islands.
However, since the Canary Islands had been colonized by immigrants
from Andalusia, the main contribution of the Canary Island
immigration was to reinforce previously existing patterns of
consonantal variation.

Importation of African slaves also contributed to the
development of Cuban Spanish. The contact of Spanish and west
African languages did not produce any widespread phonological
innovations (Garcia Gonzalez, 1980; Lipski, 1994). However, it did
extend existing processes well beyond their original linguistic
distribution. For instance, increased use of /s/ deletion and /r/
lateralization, sociolinguistic markers associated with low social
status, tends to occur more in the varieties of regions with larger
numbers of blacks (Megenney, 1989). The unique realization of
liquid assimilation in Cuban Spanish is attributed to influence from
the African substrate, which extended it well beyond its original “p
linguistic distribution (Garcia Gonzalez, 1980). In Andalusian

Spanish, and in other Caribbean varieties, /r/ assimilates to a

following :’l

  
    

following co"
lateral liquid
range of cons,
(Alonso. 196‘:

preconsonanta

 

Rei

it was pl
consonants su
occurred beret
southern Peni
smaller as a r
and r and 1's
reiional yari:
i'arieties of S
Cedelgren‘ 1
\VaiielO'Clar.
“0131353 195

Canary ISlan

Of The
E551 Variatit
.azrteenth ce

following /1/ or /n/, but in Cuban Spanish it assimilates to any
following consonant (Almendros, 1958; Guitart, 1976, 1980). The
lateral liquid, on the other hand, is reported to assimilate to a wider
range of consonants, including /r/, /e/, /t/, /d/, /dz/, /y/, and /ii/
(Alonso, 1961), but, again, in Cuban Spanish it assimilates in any

preconsonantal environment (Almendros, 1958; Guitart, 1976, 1980).

Regional and Social Variation in the Caribbean ’i”

It was previously noted that a very limited number of
consonants survived the severe reduction of coda consonants that
occurred between the late thirteenth and fourteenth centuries. In
southern Peninsular and Caribbean varieties, the number is even
smaller as a result of neutralizations. These remaining consonants,
and /r/ and /s/ in particular, are responsible for most social and
regional variation in Andalusian, Canary Island, and Caribbean
varieties of Spanish (see Alba, 1992, for Dominican Spanish;
Cedergren, 1973 for Panamanian Spanish; Choy Lopez, 1986 and
Vallejo-Clams, 1971 for Cuban Spanish; Poplack, 1981, and Lopez
Morales, 1987, 1992 for Puerto Rican Spanish; Catalan, 1989 for
Canary Island Spanish).

Historicfiflground of /s/fiand /r/

Of the four coda consonants, /r/ and /s/ have experienced the
most variation. Their variation is very old: /s/ dates to the tenth or
thirteenth centuries and /r/ to the sixteenth century.

History of Aspiraiion and Deletion of /s/
There is some disagreement over when /s/ aspiration and

deletion arose. Ariza (1994) states that deletion existed prior to the

 

fiheenth cent
inthetenth c:
eradicated lat
became wides

Whether
hanissue det

occurred first,

 

ofearly aspira
it was transfer
H.939) argues
no clear cases
that the fact it
ll lWIS not in i
however‘that
America from
li989) dlSniig
mum in ea:
lf dETet
higuists Clai
and deletion
CfiegoriCal d
irisuncena.1
weakening V

\be are Ten

in-
meg the int

Whether

eiiiain

it is

the1

fifteenth century in Andalusia. Evidence has been found for deletion
in the tenth century in Castille, where it is believed to have been
eradicated later through change from above, which explains why it
became widespread in Andalusia rather than Castille.

Whether aspiration occurred before deletion, or deletion first,
is an issue debated by two opposing camps: one claims aspiration
occurred first; the other that it came later. Frago (1983), a proponent
of early aspiration, sets its date around the thirteenth century, when
it was transferred to the Andalusian dialect from Arabic. Torreblanca
(1989) argues that aspiration occurred much later, claiming there are
no clear cases of aspiration before the eighteenth century. He argues
that the fact that not all regions of America aspirate is evidence that
it was not in use in sixteenth century Andalusia. Ariza (1994) claims,
however, that there are examples of aspiration from Andalusia and
America from the sixteenth and seventeenth centuries. Torreblanca
(1989) dismisses these as errors and sets aspiration in the eighteenth
century in eastern Andalusia.

If deletion did arise before aspiration, as some historical
linguists claim, then the widely accepted modern view of aspiration
and deletion as stages of a weakening process ultimately leading to
categorical deletion, may be questioned. Ariza (1994) points out that
it is uncertain “whether we are at a stage of intense articulatory
weakening, which favors the loss of /s/, or, on the contrary, whether
these are remnants of an old medieval aspiration” (p. 242). This
poses the interesting question of whether /s/ is involved in change or
whether it is a stable variable, in which case age-grading would

explain the fact that sociolinguistic studies commonly find that

10

  
    
  
 

younger spea
mediated by
History of La

The late
lateralization
was already it
T9751Aspira
linguists with
origins and d1
aSpiration is ;
thirteenth cer

ASpirat
0f D‘OTlgina
11989)‘ Who
Sou‘lhi‘vin Ext
£thgitout t
Colonizatioi
Kinetic“, “
cacti

f

Upated
Senna and

C035

Ia] area

younger speakers use more deletion. Increased deletion may be
motivated by both internal and external factors.
History of Lateralization and Aspiration of /r/ ‘\::\

The lateralization of /r/ is hundreds of years old. Evidence of
lateralization of /r/ from sixteenth century documents suggests that it
was already widespread in Andalusia by that time (Boyd-Bowman,
1975). Aspiration of /r/ is not widely discussed by historical
linguists with the exception of Catalan (1989) who claims that its
origins and diffusion parallel /s/ aspiration, suggesting that /r/
aspiration is as old as /s/ aspiration, which dates from around the
thirteenth century.
Diffusion of Chme: From Andalusiato Amerifl

Aspiration and deletion of /s/ and lateralization and aspiration
of /r/ originated in southern Spain, in Sevilla, according to Catalan
(1989), who notes that from there they spread to Malaga, Cordoba,
southern Extremadura and Cartagena, and later spread further
throughout the rest of Andalusia, Extremadura, Granada and Murcia.
Colonization spread these variants to the Canary Islands and the
Americas, where, as noted above, the areas along the trade-route
participated in linguistic changes from Spain because of their
frequent contact and dynamic involvement in the sociocultural life of
Sevilla and Cadiz. Lateralization of /r/ occurs in America along

coastal areas of the Pacific and the Caribbean.

11

    
  

.lspirati
and coastal a:
howeier. relic
instance. dele'.
nearly categor
19911, wherea.
nearly categori
Dominican. Pu
histories. incl

Giyen tl
years, it is no
throughout m
sharply strati
Alba (19921
l'alleio-Clar
Stratificatio
that the asp
educated m'

In Cu
“Ell lilOUc‘;
STECC‘TT.’ an

Variation of /s/ in the Caribbean

Aspiration and deletion of (s)4 is found throughout the Antilles
and coastal areas of the Caribbean. Constraints on variation differ,
however, reflecting the historical circumstances of each region. For
instance, deletion is widespread in Dominican Spanish, reaching
nearly categorical application in the lowest status group (Alba,
1992), whereas in Puerto Rican and Cuban Spanish, aspiration is
nearly categorical (Terrell, 1977b). The differences in the
Dominican, Puerto Rican and Cuban varieties reflect their individual
histories, including their degree of physical and cultural isolation.

Given that (s) variation has been widespread for hundreds of
years, it is not surprising that (s) functions as a social status marker
throughout most of the Caribbean. Cedergren (1973) found that (s)
sharply stratified higher and lower social status groups in Panama.
Alba (1992) found a similar pattern in the Dominican Republic.
Vallejo-Claros (1971), however, found gradient, rather than sharp,
stratification among Cuban Spanish speakers. Terrell (1979) noted
that the aspirated variant was “clearly the phonetic norm...for
educated middle class speakers” (p. 601) of Cuban Spanish.

In Cuban Spanish, the basic variant [3]5 is overtly prestigious
even though it has an extremely low rate of occurrence in vernacular
speech: around 18% word-finally and 2% word-internally for middle
class speakers (Terrell, 1979). Since it is prestigious, it tends to

appear more in formal styles, but its frequency is low, nonetheless.

 

4The use of parenthesis is standard sociolinguistic notation for
variables, and they will be used in place ofthe slashes used for
phonemes from this point on when discussing coda consonants to
indicate their status as sociolinguistic variables.

5Brackets are used for variants of sociolinguistic variables.

12

In reading. l
one-to-one t
theintense .
elementary
sound chan,
as the stant
Since
iatieties o
iatiants it
there is 111
both come.
tenants 1
social st;
lilba. 1I
Tattern t
In

be mOTe

thepri

In reading, however, [3] is more frequent, most likely because of the
one-to-one correspondence between letter and sound in Spanish and
the intense attention given to the use of [s] in reading style in
elementary education. While these efforts are unlikely to reverse a
sound change that is hundreds of years old, they have established [5]
as the standard for reading.

Since the basic variant [5] is used infrequently in some
varieties of Caribbean Spanish, aspiration and deletion are the
variants involved in social stratification. Aspiration is the norm, but
there is no social awareness of it, unlike [s] and [a], which are
both commented upon and openly subject to social evaluation. These
variants show typical sociolinguistic patterns: [h] increases as
social status increases, whereas [6] increases as status decreases
(Alba, 1992; Cedergren, 1973; Terrell, 1976, 1979), a common
pattern for stigmatized phonological variants.

In summary, the regional and social patterning of (3) tends to
be more or less the same throughout the Caribbean, with the
exception of advanced levels of deletion in the Dominican Republic.

(r) in Céaribbeafin Varieties

The varieties of the Antilles have been diverging from the
original Andalusian variety, and from each other, for hundreds of
years, as physical isolation, and historical and sociocultural
circumstances shaped a unique linguistic situation in each region.

The older variants of (r), lateralization and aspiration, brought
to the new world several hundred years ago, tend to occur throughout
the Caribbean region and show similar social patterning. In the three

Antillian varieties, there are also newer variants of (r) which are

13

uniquely ditt
regions.

Lateralizatior

 

The late
the Caribbean
coastal areas
Dominican R
and l'enezuel
Panama, and

The lat
of the Caribl
Cuba, it is It
in lower stat
which it i5 11

ValleTO-Clar

because U161
i3 Pnerm Ri
iodil‘idans
idionso‘ 19
not. the late
groups in al

The St
toregiOn P
on its 0%“
that lober s

soups l€n4

uniquely different from one another and widespread in their local
regions.
Lateralization and Aspiration

The lateralized and aspirated variants of (r) occur throughout
the Caribbean region. Lateralization is found in the Caribbean and
coastal areas of the Pacific. Aspiration is found in Cuba, the
Dominican Republic, Puerto Rico, the Caribbean coasts of Colombia
and Venezuela (although not in Caracas, according to Catalan, 1989),
Panama, and the Pacific coastal areas of Colombia and Ecuador.

The lateral variant is stigmatized in many, but not all, regions
of the Caribbean and it tends to correlate with lower social status. In
Cuba, it is reported not to have a very high rate of occurrence even
in lower status groups and rural speakers, which are the groups in
which it is most frequent, (Almendros, 1958; Garcia Gonzalez, 1980;
Vallejo-Claros, 1971). As for other regions of the Antilles, Alba
(1992) reports that it is not stigmatized in Dominican Spanish
because there is a vocalized variant which carries the social stigma.
In Puerto Rico, the lateral is widespread in the speech of educated
individuals and tends not to carry any negative social evaluation
(Alonso, 1961). Regardless of whether it is socially stigmatized or
not, the lateral tends to be used more frequently by lower status
groups in all regions (Poplack, 1981; Vallejo-Claros, 1971).

The social patterning of the aspirated variant varies according
to region. Poplack (1981) found that social factors had a small effect
on its occurrence in Puerto Rican Spanish. Cedergren (1973) found
that lower status males of rural origin in the youngest and oldest age

groups tended to use more aspiration. In Andalusia, Alonso (1961)

14

reports th
lateraliza
that it is
group. he
status an
stereotyq
Th
Antilles
except i
satiety,

final is

reports that aspiration correlates with urban origins and
lateralization with rural. In Dominican Spanish, Alba (1992) reports
that it is used only by speakers from the lowest socioeconomic status
group; he found that it was negatively evaluated by 93% of upper
status and 82% of lower status speakers. Lateralization is not a
stereotype in the Dominican variety.

Other Regional Variants of (r)

The development of (s) and (r) have been very different in the

 

 

Antilles. The processes affecting (s) have not changed dramatically,
except for the frequency of the deleted variant in the Dominican
variety; furthermore, no new socially significant variants of syllable-
final (5) have arisen. On the other hand, the evolution of (r) has been
marked by the appearance of regional variants unique to each region.
Some examples include a velar fricative common in Puerto Rico
(Lopez Morales, 1987), a vocalized variant (of both liquids) found in
parts of the Dominican Republic (Alba, 1992), and in Cuba, (r)
assimilated to a following consonant, resulting in partial or complete

gemination. On assimilation in Cuba, Almendros (1958) wrote:

The frequency and extension of this feature in the western
region are perhaps what most distinguishes Cuba within the
Caribbean region. Navarro Tomas states that ‘in Puerto Rico it
is infrequent, occurring mostly among blacks’ and before /m/,
/n/ and /l/, while in Cuba it occurs before any consonant. P.
Henriquez Urefia identifies it in Dominican Republic as only a
rare occurrence. (p. 149)

The variants of (r) which uniquely characterize the Puerto
Rican, Dominican and Cuban varieties are regional stereotypes, as
CVidenced by the fact that they are often used, along with the most
Salient suprasegmental features of these varieties, in imitations and

caricatures.

15

 

involved in r.

  
   

the other twt
more resistar
processes aft
in Cuban Spa
also occurs in
American yar
Spanish and i
assimilates p'
assimilates b

All [r]
01) Caribbean
elidertce for
Rmasand SC
Fflkfllggsa
TSUCqUenI hj
Since lhESe i
hate the lea:
lbtiaqeral O
iECOrpeCt (e

L .
“EqUentqy C

The Status of Other Coda Consonants

As noted earlier, (5) and (r) are the coda consonants most
involved in regional and social variation in Caribbean dialects. Of
the other two coda consonants, /l/ and /n/, the lateral liquid, although
more resistant to variation than /r/, is often involved in the same
processes affecting /r/: in Dominican Spanish it is vocalized [i], and
in Cuban Spanish, it assimilates. Although assimilation of the lateral
also occurs in Andalusian, Canary Island and many other Latin
American varieties, the difference between assimilation in Cuban
Spanish and in other varieties is that in the latter the lateral

assimilates primarily before /r/ and /n/, but in the former it

“ fi—h—

assimilates before any following consonant.

An [r] variant of the lateral is often reported in the literature
on Caribbean dialects, but there is a general lack of empirical
evidence for it (i.e. Alba, 1988, for Dominican Spanish; D’Introno,
Rojas and Sosa, 1980 quoted in Chela-Flores, 1986, for Caracas;
Fails, 1985 and Hammond, 1977, for Cuba). The [r] variant is an
infrequent hypercorrect form that occurs in the lowest status groups.
Since these are the groups who lateralize (r) most frequently and who
have the least access to the standard norm, the rhotic liquid replaces
the lateral one because speakers suspect that using the lateral is
incorrect (e.g. cerebro ‘brain’ and celebro ‘I celebrate’ are
frequently confused).

In general, except for vocalization in the Dominican variety,
and assimilation in the Cuban one, (1) does not appear to be involved

in any other socially significant variation.

16

 

Choy
new unpo
wnse'ter
Coba‘lp
Tanbbeat

a!

increnu

Pnortoi
U986.l
flownjq
pohncal
dnlectl
0Ul.teg
bbuhda
“ESTEIi
TESTEIY
bit the

the pp

6EST 0

Regional and Social Variation in Cuba
Choy Lopez (1986) states that coda consonant variation “is the
most important in the process of dialect formation—in the widest
sense: territorial, social and historical—of the Spanish spoken in
Cuba” (p. 412). As noted in the earlier discussion on variation in
Caribbean dialects, the variables (5) and (r) do most of the work of
differentiating regional and social dialects.
Cuban Regional Varieties
Prior to the reclassification of regional dialects in Choy Lopez
(1986, 1994), Cuba was believed to have three regional dialect areas,
shown in the map in Figure 1. Regional varieties reflected the ’
political boundaries of provinces6 because the exact location of major
dialect boundaries was unknown, but, as Almendros (1958) pointed
out, regional varieties were not neatly contained within political
boundaries. The western dialect covered the three provinces on the
western side of the island, Pinar del Rio, Havana, and Matanzas. The
eastern dialect was found in the large province of Oriente, situated
on the eastern side of the island. The central dialect region included
the provinces of Las Villas and Camagt’iey, located in the west and
east of the central area of the island. According to Almendros (1958),
the central region was a transition zone which had the phonological
features of the more conservative eastern dialect and the

suprasegmental features of the more innovative western one.

 

6The Cuban provinces discussed in older works, from west to cast,
are Pinar del Rio, La Habana, Matanzas, Las Villas, Camaguey, and
Oriente. In 1978, Cuba’s traditional provincial governments were
reorganized, and the provinces divided into smaller, more
manageable, areas. The map in Figure 2 shows these new political
divisions.

l7

These
11986. 199;.
consonantal

are shown tl

 

ofPinar del

part ofthis t
and Trinidac
central diale
part of the i
dissertation
Ciego de A
Tunas. prey
lianzanillo
eastern citi
Btracoa is

en ,
3.5 m‘

These regional dialect divisions were redefined by Choy Lopez
(1986, 1994), who identified five regional dialects based on the
consonantal differences of urban areas.7 These new dialect divisions
are shown the map in Figure 2. The western area includes the cities
of Pinar del Rio, Havana, and Matanzas, which had originally been
part of this dialect area, and it also includes the cities of Cienfuegos
and Trinidad,8 which had previously been considered part of the
central dialect region and which are physically in the south-central
part of the island. The central dialect region, the subject of this
dissertation, includes the cities of Santa Clara, Sancti Spiritus and
Ciego de Avila. The east-central area includes Camaguey and Las
Tunas, previously considered as the central dialect area, and Holguin,
Manzanillo, and Bayamo. The southeastern variety is found in the
eastern cities of Santiago de Cuba and Guantanamo. The city of
Baracoa is classified as a separate dialect area, which Choy Lopez

calls extremoriental, but called east in the map shown here.

 

7Choy Lopez does not suggest the location of major dialect
boundaries in relation to these urban areas.

8The inclusion of Trinidad in the western dialect region is based on
Goodgall de Pruna (1970a), who reported widespread liquid
assimilation in Trinidad, an unusual finding at the time for a city
located in central Cuba.

18

 

\e...s.-

~23; >25 .3 w:— E cognac?! 3 coin commas—o mm 333:; :Eomwom i; 9.: E

1.}13\«...)\(..~

.\

.\~. {N\. . .A : ..IIIWE)i¢/i\l$t. 5 is . ..

AmmmmmrLL-IHD o '3 0—.

uougm/ swag—am 25.55
./u/l..l)l.rl~I/\e
M,“ . 355...
first.»
{wk/«wad
It...
Em

 

19

. . . ...1.
”dag

wage/main {Kin}.

“new

Avoa_

“new "533%.

a

23: N83 2,6

.3 332.33 3:2on mo =230E3£oo¢ it; 8: E

 

Em E. 58m Jacaxvrss»

tastiest

20

 

dialect '
regiona
accordi.
the fact
one Ac
hequen
in the a
Baraco
Ciaros
than sp
howey
tastett
Horthw

TTEVio

The Role of (sLin Regional Vagamon in CM

In general, (8) plays a much lesser role than (r) in regional
dialect variation. The aspirated variant of (s) does not appear to show
regional variation, but the deleted variant is reported to vary
according to region, although perhaps this observation is based on
the fact that the deleted variant is more salient than the aspirated
one. According to Choy Lopez (1986), deletion occurs more
frequently in the eastern and southeastern varieties, in other words,
in the areas around the cities of Santiago de Cuba, Guantanamo, and
Baracoa. This observation contradicts earlier findings by Vallejo-
Claros (1971) that speakers from eastern Cuba tend to delete less
than speakers from the western or central regions. This difference,
however, may result from the fact that Vallejo-Claros grouped all
eastern speakers together, but deletion is less frequent in the
northwestern parts of this region, in Holguin and Las Tunas, areas
previously defined as belonging to the eastern region. If the
respondents in Vallejo-Claros were from this more conservative area,
then less deletion would have been found for the eastern variety. The
linguistic behavior of the rest of the island, the east-central, central
and western regions, is not reported as showing significant
differences in (s) variation (Choy Lopez, 1986; Vallejo-Claros,
1971).
Die Role ofir) in Cuban Regional Variation

The variable (r) plays an important role in regional dialect

variation in the Caribbean. In Cuba, too, (r) is an important variable
in regional dialects. Assimilation is reported to be the most salient

variant in Cuban regional dialect differentiation.

21

featur
categc
1958.
assimi
some .
the ce
occurt
more

tttqut

'tariat
accor

taster

R! .

boiler.

Liquid assimilation was historically considered a characteristic
feature of the western dialect region which reached nearly
categorical rates of application among certain speakers (Almendros,
1958; Guitart, 1976, 1982). In 1958, Almendros suggested that liquid
assimilation was moving eastward on the island, pointing out that
some cases of it could already be found in Santa Clara, the capital of
the central region. By 1986, Choy Lopez reported that assimilation
occurred in all regional varieties, although at varying rates. It was
more frequent in the western dialect than in the central one, but more
frequent in the central variety than in others to the east.

The lateralized variant apparently shows little regional
variation, as it tends to correlate mainly with status. Lateralization,
according to Choy Lopez (1986), is most frequent in parts of the
eastern region, but it is also common in the west and central dialect
areas. Vallejo-Clams (1971) found more lateralization for the
western dialect (13%), than the central (4.5%), or eastern (0%). As
with (s) deletion, Choy Lopez (1986) and Vallejo-Clams (1971)
report opposite findings for the eastern region, which, as suggested
earlier, may be because the latter’s respondents were from a more
conservative part of the western region, now classified as a separate
variety.

Social Dialects
is) Variation and Social Stratification in Cuba

(8) variation plays a role in social stratification throughout the
Caribbean. In general, deletion is stigmatized and more frequent in
lower status groups (Alba, 1992; Catalan, 1989; Lafford, 1986). In

Cuban, Puerto Rican and Dominican varieties, [h] occurs more than

22

 

is] in infe-

 

lllil) four:
with deletio
decreased. a

in Cut
associated w
aregional f
(Alfaraz, 1%
seem to be

The i:
will (All
lateral ten:
and lite pe
T970, T992
l'iTiCllesj t

[s] in informal speech, and in higher status groups. Vallejo-Claros
(1971) found that (s) showed more social than regional variation,
with deletion increasing and aspiration decreasing as status
decreased, and that it showed a pattern of gradient stratification.
Social Variation and (r) in Cuba

In Cuba the lateral variant is a negative social stereotype
associated with lower status and rural speech. Assimilation is mainly
a regional feature, but its social differentiation has been suggested
(Alfaraz, 1999; Guitart, 1976, 1980). For aspiration, there does not
seem to be any evidence of its social patterning.

The lateral variant was historically associated with lower status
speech (Almendros, 1958; Vallejo-Claros, 1971). In general, the
lateral tends to have a low rate of occurrence, usually between four
and five percent of all cases of syllable final /r/ (Lopez Morales,
1970, 1992; Terrell, 1976; Vallejo-Claros, 1971). In all regional
varieties, the frequency of the lateral variant increases as social
status decreases (Vallejo-Claros, 1971).

According to Garcia Gonzalez (1980), the lateral was becoming
more widely used in Cuba, even by educated speakers: “It is a very
frequent phenomenon in our popular speech...This usage has extended
to educated speech, although the normal pronunciation predominates”
(p. 118). Choy Lopez (1986), also reports use of the lateral in the
educated norm, noting that it was becoming more frequent in the
Speech of younger individuals, especially in the western region.

Almendros (1958) noted that assimilation occurred in all social
Classes in the western dialect region. Later studies also reported it in

all social status groups (Choy Lopez, 1986; Garcia Gonzalez, 1980).

23

 

r

Howeten L

 

mereasesir
and Guitart
behne cero
sutusspee
botless fre

thte
difierentia
asannlatic
handard n
nonmanda'
asTlllltue
toronals s
[pane] Q:
butthe st:

afldthe CC

Tire
and SOCia
llyhas a]
assimilati
are [he ba
is} are Th1

fi-

‘“6 To],

However, Lopez Morales (1992) claims that the rate of assimilation
increases in lower social status groups. Both Lopez Morales (1970)
and Guitart (1982) suggest that status correlates with assimilation
before certain consonants. Lopez Morales proposes that in lower
status speech assimilation tends to occur regularly before /t/ and /d/,
but less frequently preceding other consonants.

Guitart (1982) gives a more detailed description of social status
differentiation through assimilation. He proposes that the outcome of
assimilation preceding coronals and noncoronals distinguishes the
standard norm from other nonstandard dialects. Both standard and
nonstandard varieties have complete gemination before noncoronals,
as in [kuéppo] cuerpo ‘body’ and [aggo] flga ‘something,’ Before
coronals simple gemination occurs in nonstandard varieties (e.g.

[patte] parte ‘part,’ [kusso] curso ‘course,’ [kanne] carne ‘meat’),

 

 

but the standard norm has a retroflex obstruent or a retroflex flap,
and the coronal is retroflex, too, as in [pad't'e] parte, [ku'r's’o]

curso, [kad fie] carne.

 

 

Summry: Regionaflnd Social Vaflaaga
The variants of (s), (r), and (1) relevant to the study of regional
and social variation in Cuban Spanish are the following. The variable
(t) has a basic variant [r], a lateral one [1], a variant arising from
assimilation [assim], and an aspirated one [h]. The variants of (l)
are the basic [1] and assimilated [assim] variants. The variants of

(S) are the basic variant [s], the aspirated one [h], and the deleted

one [a],

24

 

The c
(r), (l) and

the central 1

 

nap in Figui
ofregional
lateralizatic
the speech I
variants site
to base low
younger 5p
liltil’ Speec
noted that
WeSlCl’n 017
Speakers 3
deit‘llofi 0

Sociolinguistic Expectations of the Current Study

The current study investigates the sociolinguistic patterning of
(r), (l) and (s) in the speech of university educated individuals from
the central dialect region. The area studied here is indicated on the
map in Figure 3. Based on the earlier discussion of the distribution
of regional variants, low rates of (r) and (l) assimilation, (r)
lateralization and (s) deletion are expected. As this study examines
the speech of university educated individuals, socially stigmatized
variants such as lateralization of (r) and deletion of (s) are expected
to have lower frequencies than other variants. At the same time,
younger speakers are expected to have some liquid assimilation in
their speech, but older speakers are not (recall that Almendros (1958)
noted that assimilation was spreading to the central region from the
western one). The earlier discussion also suggests that younger
speakers are expected to use more lateralization of (r) and more
deletion of (5). These sociolinguistic expectations will be discussed

in the results and discussion sections of the following chapters.

25

news: .23

vow—:2: 32.5 28.50 on: we “00.30 2: 2 not?» no.3 of at? 3.3 323“. 35:00 of. .miualzmmim

3‘\.$~\2P .1/(«3‘311‘5‘11517-3“ \HI..-{\ .\
1...... it o
kadnb8~58 . 330 on

r...»
waumfion/Mfiésgf

 

26

 

Thn
research qt
unaMes.b|

unanono:

coHection a

 

Twoap
dncnbedin
8mUpofindi
mcmecononi
lohnguismc
mdohnguim
ac0mparisor
”Preach beg
decoyerthe
OWIhatthe.
SignificanCe
Xew YOTker
oi)“ he Talks

Bodio
muamensu

Laboy points

Lindam
diffEre;
Speaker
eSlabHE

METHODOLOGY

This chapter explains the method used to investigate the
research questions outlined in Chapter 1. It describes the contextual
variables, both linguistic and social, that are used to study the
variation of (r), (l) and (3). Also described here are the data

collection and analysis procedures.

The Social Group Versus the Individual

Two approaches to the study of sociolinguistic variation are
described in Labov (1982/1966). The first approach begins with a
group of individuals with common characteristics, such as
socioeconomic status, age or gender, and examines their relationship
to linguistic behavior. This approach renders information about the
sociolinguistic expectations for different social groups and allows for
a comparison of groups in relation to linguistic features. The second
approach begins with a particular linguistic item and then attempts to
discover the social characteristics associated with it. Labov points
out that the second approach “is equivalent to looking for the social
significance of a speech pattern as we first hear it; when we meet a
New Yorker for the first time, what can we infer about him from the
way he talks?” (Labov, 1982/1966, p. 136).

Both of these approaches can be used to determine the social
characteristics which correlate with certain linguistic behaviors, but

Labov points out that the first approach is preferable:

The first approach, through social groups, seems more
fundamental and more closely tied to the genesis of linguistic
differentiation [since] the social group of peers in which a
speaker spends his pre-adolescent years is the main force in
establishing his linguistic pattern. (Labov, 1982/1966, p. 136)

27

 

ll norms o.
early peer ,
intestigatit
the current

regional ta

 

 

The a
norms requ
norms \l'it
includes lei
lnowledge
language t
l978ll'olt

l'ari
lingttistic
CORlTibuq
Mlle [305
l“ Exam
“ll“? c
\‘atiam <

alsfal'or

If norms of linguistic behavior are learned through socialization in
early peer groups, then the social group is the primary source for
investigating sociolinguistic behavior. This was the approach used in
the current study to investigate the variation of (r), (l) and (s) in a

regional variety of Cuban Spanish.

Contextual Constraints on Linguistic Variation

The assumption that groups share and perpetuate linguistic
norms requires that linguistic competence include knowledge about
norms. Within the variationist framework, linguistic competence
includes knowledge about constraints on variation, as well as
knowledge about their hierarchical order, since varieties of a
language tend to share constraints but order them differently (Fasold,
1978,Wolfram & Schilling-Estes, 1998).

Variable rules represent the psycholinguistic expression of
linguistic norms. Speakers have knowledge about which constraints
contribute to the application of variable rules, and which elements
make positive or negative contributions to their application. Thus,
for example, the linguistic competence of speakers of a Caribbean
variety of Spanish includes knowing that application of the aspirated
variant of (s) is favored in the environment before a consonant and
disfavored before a vowel or pause.

Inteml Constraints on Variability

Contextual constraints on variation may be internal or external.
Internal constraints on phonological variation commonly include the
following and preceding phonological environments, the location of

primary stress, the sound’s position relative to a word boundary, the

28

.
A

VT-
-‘h

L031

q
kao
y,"

type of morpheme in which the sound occurs, and the lexical class of
the word in which the variable is found (postulated as a constraint on
the application of aspiration and deletion of syllable final (5) by
Terrell, 1979). These constraints were examined in early variationist
studies on Spanish dialects, including Cedergren, 1973; Poplack,
1981; Terrell, 1976, 1977a, 1977b, 1979; Vallejo-Claros, 1971.
Extergl Constraints on Variability

Variation is also constrained by factors external to language.
External constraints include factors which are relevant to an
individual’s identity. Preston (1989) distinguishes between
individual attributes which are ascribed by society and those which
are acquired by the individual. Preston notes, “Ascribed
characteristics are those over which individuals have the least
control...the list of ascribed individual characteristics includes, then,
those factors most closely bound to identity (1989, p. 53). Ascribed
characteristics include age, gender,9 ethnicity, and region. Acquired
characteristics, on the other hand, reflect an individual’s “dynamic
interaction with society itself” (Preston, 1989, p. 106). They include
the roles individuals play in social settings, and individuality, which
may reflect individual attitudes and network affiliations. Social
status is also an acquired characteristic, although this depends on the
organization of the society in question since status is ascribed rather

than acquired in some societies.

¥

9Gender, as opposed to sex which is biologically determined, is
constructed socially (Wodak & Benke, 1997), but this does not
conflict with its being an ascribed attribute since society associates
certain norms of linguistic behavior with gender.

29

   
  
  

in q.
nosr impo-
These erte

constraints

VARE

quantitative

 

ofenvironm
which lingo
rules. VAR}
not certain

feature; one'
Silengqh 0T
l'iRBavL
diltCTTOn a
UHQWS a

affect, a v;
Value Of '5
calegm’lCa

0.0 indlCaq
These Well
elilanaqio
SOClOllngU

In quantitative sociolinguistics, age, gender and status are the
most important of these ascribed and acquired social characteristics.
These external constraints are examined along with internal

constraints to determine the patterning of sociolinguistic variables.

The Expression of Variable Rules: VARBRUL

VARBRUL is a statistical data analysis tool commonly used in
quantitative sociolinguistics to numerically estimate the contribution
of environmental factors to variation. It measures the degree to
which linguistic and social facts contribute to application of variable
rules. VARBRUL carries out two tasks: first, it calculates whether or
not certain factors have significant effects on the variation of a
feature; and second, it measures the degree of the effect, or the
strength of its contribution to application of the rule. The output of
VARBRUL are numerical values, called weights, which indicate the
direction and strength of the effect. Weights are interpreted as
follows: a weight above .5 indicates that the factor has a positive
effect; a value below .5 indicates that the factor retards the rule; a
value of .5 indicates a neutral effect; a weight of 1.0 indicates
categorical application (called a positive knockout); and a weight of
0.0 indicates the rule never applies (called a negative knockout).
These weights and values are shown in Table 2. For a more in-depth
explanation of the application of VARBRUL analysis to
sociolinguistic research, the reader is referred to Guy (1993) and

Young and Bayley (1996).

30

 

 

Above .5

The
are descr
ClPlanari
descripri.

Used in IT

The
number 0
pieSCl’ipqi
the distan
lheqary Te

The Piesc

Table 2
VARBRUL Weight \Qlues

 

 

Weight Type of contribution to application of rule

Above .5 Positive

At .5 Neutral

Below .5 Negative

1.0 Categorical application (positive knockout)

0.0 Categorical nonapplication (negative
knockout)

 

The Current Study: Method and Procedures

The linguistic and social factors examined in the current study
are described in the following sections of this chapter. A brief
explanation of the selection of sociolinguistic variants precedes a
description of internal and external variants, and of the procedure
used in the data collection and analysis.

The Selection of Variants

The description of Cuban Spanish is attempted in a large
number of published works on the variety, but these are mainly
prescriptive commentaries on the status of the variety which estimate
the distance it has strayed from the Peninsular norm, based on
literary representations of dialects rather than real speech samples.
The prescriptive nature of such dialect studies extended well into the
1 9605. One of the most interesting examples of later work is Bartos'
(1964, as cited in Fails, 1985), which proposed that Cuban Spanish
was so different from the Hispanic norm that not only did speakers of

other varieties have trouble understanding Cuban Spanish, but that

31

Cubans th

 

reason for
Thes

l

Ofllle pilO-t‘
out “the no
less by the
continuum

Thus. the la

 

appear to re

 

 

their classif
syllable fin;
allophones l
others. such
various inte
deletion), 0

The s.-
lllat STUdTES
variants Of
“1051 comm
decision to

Cubans themselves had trouble understanding each other (thus, the
reason for the frequent use of (,dj_m_e_? ‘what?’ and (,Ela? ‘huh?’).

These early studies describe an impressive array of allophones
of the phonemes /r/, /l/ and /s/; however, as Clayton (1985) points
out “the number of segments in our phonetic inventory is determined
less by the real world than by our choice of how many points along a
continuum or near continuum we choose to identify or define” (p. 8).
Thus, the large number of allophones proposed for the Cuban variety
appear to result from little agreement, and a lot of innovation, in
their classification. While some of the major allophones of certain
syllable final consonants, such as the aspirated and deleted
allophones of /s/, are for the most part agreed upon by researchers,
others, such as those arising from liquid assimilation, are subject to
various interpretations (i.e. assimilation is frequently interpreted as
deletion), or may remain unidentified.lo

The scientific nature of quantitative studies, which requires
that studies be reproducible by other researchers, demands that the
variants of sounds selected for study be those that are clearest and
most commonly agreed upon. Terrell (1977a) explains that the
decision to study only the aspirated and deleted variants of /s/ was

based on their clarity and social significance:

 

10This is the case for liquid assimilation in many studies on Cuban
Spanish (i.e. Hammond, 1977; Isba’sescu, 1968; Vallejo-Clams;
1970). Assimilation is usually interpreted as deletion of the segment
because the outcome of assimilation is complete or partial
gemination, which gives the impression that the segment is gone.
Autosegmental accounts of liquid assimilation in Cuban Spanish
suggest that the segment is not deleted, it simply shares the features
of the preceding sound (Harris, 1985). This is more obvious in cases
of partial gemination, ‘parto’ [padto], than in cases of complete
gemination, ‘pardo’ [paddo]. '

32

 

TerreH fur
obstacle tc
nmesthat
Tssue
downin T

variation 0

Tdde 3
Vanable
\

(Y)

I would have been able to investigate other variants, ranging
from strong to weak aspiration, but it was practically
impossible to systematically establish the difference between
them. Still there are other variants that are easier to
distinguish: a mixed aspirated-sibilant variant...slightly voiced
aspiration...and a nasalized aspiration. But these variants
appear to be determined completely by phonetic context,
without any social or stylistic stratification (p. 43).

Terrell further explains that using fine transcription presents an
obstacle to replication. This point is often made by Fails (1985), who
notes that his results are influenced by acoutic analysis tools.

Issues of replication motivated the selection of the variants
shown in Table 3. These variants represent the clearest cases of the

variation of (r), (l) and (5).

Table 3

Variants ofLr). (l) and s

 

Variable Variants

 

(T) if], [11, laSSim],
[h]

(l) [l], [assim]

(S) [S], [h], [9]

For the current study, issues of reliability were addressed by
coding tokens of each variable in several passages from each
respondent two or three times and determining whether the results
were consistent, as suggested in Guy (1993). Furthermore, I also
consulted with two other native speakers of this variety.

Internal Constraints on Variability
The internal constraints on variation examined in this study

are: following and preceding phonological environment, stress,

33

 

poouoni
phonologi
Oflll dtt‘t‘
baucand
(Tads.l9
butthe as
lAhnendr
Pruna.l9
tooccurt
Gonzalez
consonan
19703; G
thewhert
Slllable-

The
commOUq
0fill‘e \‘a
lilVegq'ng
the {0110

T11
disguSSe
ll981) f
Spanish
been n01
Stud}. ex

Sl'llabqe

position in the word, and morphemic category. Following
phonological environment commonly has a strong effect. The variants
of (r) differ according to the environments in which they occur: the
basic and lateral variants appear before a vowel, consonant or pause
(Fails, 1985; Hammond, 1977; Garcia Gonzalez, 1980, Terrell, 1976),
but the assimilated variant only occurs before consonants
(Almendros, 1958; Fails, 1985; Garcia Gonzalez, 1980; Goodgall de
Pruna, 1970a; Guitart, 1976, 1980). The aspirated variant is reported
to occur only before certain consonants (Fails, 1985; Garcia
Gonzalez, 1980). The assimilated variants of (1) appears only before
consonants (Fails, 1985; Garcia Gonzalez, 1980; Goodgall de Pruna,
1970a; Guitart, 1976, 1980), and the basic variant, although it occurs
elsewhere, is examined only in the preconsonantal environment.
Syllable-final (5) occurs preceding consonants, vowels or pauses.

The preceding environment for these coda consonants is
commonly a vowel. The effects of a preceding vowel on the variation
of the variables studied here does not appear to have been
investigated in previous studies, which have concentrated more on
the following segment than the preceding one.

The contribution of stress to the variation of (s) has been
discussed by Cedergren (1973) for Panama City Spanish, Poplack
(1981) for Puerto Rican Spanish, and Alba (1988) for Dominican
Spanish. The role of stress in the variation of syllable-final (r) has
been noted by Poplack (1981) for Puerto Rican Spanish. The current
study examines the presence or absence of primary stress in the

syllable in which the coda consonant occurs.

34

 

 

lb
uncelr
gmmma
explains
tho ext
1981, T
morphe
pretty t
iunctio
person
hrmSt
occurs
person
singul:

the fir

Table

in Ta

Morphological constraints on the variable (r) are considered
since (r) can function as a nongrammatical morpheme or as a
grammatical one representing the infinitive inflection. This is
explained in greater detail in the chapter on (r). The analysis of (s)
also examined morphological constraints (Cedergren, 1973; Poplack,
1981; Terrell, 1977a, 1977b, 1979) because (3) is an inflectional
morpheme that marks plurality on each item of the noun phrase:
Todos los zapatos aauellos bonitos estan alla ‘All the shoes those
pretty are there’ (All those pretty shoes are there). (3) can also
function as an inflectional morpheme carrying information about
person and number on verbs. Table 4 illustrates (s) in present tense
forms of the regular verb me; and the irregular copula s_e; ‘be.’ (3)
occurs in the second person singular tu comes ‘you eat,’ the first
person plural morpheme -mos. nosotros comemos. the second person
singular w ‘you are,’ third person singular as_ ‘he, she, it is,’ and

the first person plural form somos ‘we are.’

Table 4

Number arnd Person Marking on Verbs

 

Regular present tense ‘Ser’ (be) present tense

 

Singular Plural Singular Plural
lst person come comemos soy somos
2nd person comes comen eres son

id person come comen es son

 

Sociolinguistic Constraints on Variability
The social variables investigated in the current study are shown

in Table 5. They are age, gender, parents’ socioeconomic status and

35

 

   
  

070.Rh
MRMSof
beutwidel
the most i
19821966)

conmdered

laMeS

Even
Ofdlalecq a
individUal.
sodoecOnO
.llOreoVef‘ 1
SligcsriOn
PhH‘dETDht

rm] phOnC

r1 .
hsed In th

parents’ origin. For parents’ socioeconomic status rapper indicates

upper middle, middle is lower middle, and Tower is equivalent to

 

working class. The contributions of age and gender to sociolinguistic
variation are well established (Labov, 1972, 1990; Trudgill, 1972,
1974). While status is a well established sociolinguistic variable, the
effects of parents’ socioeconomic status and parents’ origin have not
been widely studied. This is because the peer group is accepted as
the most important contributor to social dialect patterns (Labov,
1982/1966); thus, an individual’s current socioeconomic status is

considered more relevant than that of the parents.

Table 5
The Social Variables

 

 

Age Gender Parent Status Parent Origin
28-35 (Young) Female Upper Urban
60+ (Old) Male Middle Rural

Lower

 

Even though the peer group is accepted as the primary source
of dialect acquisition, social mobility cannot be ignored; an
individual’s adult socioeconomic status may be different from the
socioeconomic status of the group in which he was socialized.
Moreover, the central role of parents is supported by Payne’s (1980)
suggestion that despite being born and raised in a particular
Philadelphia neighborhood, children do not fully acquire certain low-
level phonological rules of the dialect unless their parents were also

raised in the same neighborhood.

36

Information about the social status of parents was included
because many of the respondents had experienced upward social
mobility as a result of wider access to education in post-
revolutionary Cuba (Fitzgerald, 1990). An index based on education
and occupation uncovers only gross differences in social rank and is
not sensitive to underlying facts about the speakers' childhood
associations, especially prolonged and intense contacts with members
of their parents’ social networks, particularly with persons of equal
social status. Examining the socioeconomic status of respondents’
parents will reveal effects of these social network relationships on
speech patterns. Moreover, if the results indicate that parents’
socioeconomic status correlates with the variation, then it must be
explained why the social dialects of parents persist in spite of the
fact that external social class boundaries have been leveled.

An index based on education and occupation is used to
calculate the social status of respondents, as they are overt indictors
of status. Although there are many covert elements relevant to social
status that cannot be easily measured,11 index scores were used to

gauge a person's rank in the local and national society. The

 

11One of these covert elements is the black market economy, in which
individuals with greater resources, obtained primarily through illegal
activities, purchase or trade items. A second element is that
individuals whose families previously belonged to higher status
groups commonly sell off their family's valuables to purchase
automobiles, television sets, refrigerators and other luxury items. A
third element is the economic differences that may exist between
individuals who are members of the Communist party and those who
are not since party membership often has its economic advantages. A
forth element is that individuals who receive financial assistance
from family members outside the country, who are either working in
another country for the Cuban government or living abroad in exile,
also have greater purchasing power. '

37

educatu
Cuba. a
vfluedt

receives

Table 6

 

 

Tl
that pro
occupati
Prior to
Parents

area's Ct

educational index is shown in Table 6. It reflects the fact that in
Cuba, as in most western societies, a university education is highly
valued and gives social respect and power to the individual who

receives it.

Table 6
Education Index

4 = Completed or attended university

3 Completed secondary school
2 = Some secondary school

1 = Grade school or less (lst to 8th grade)

The occupational index is shown in Table 7. It reflects the fact
that professionals are more respected than individuals in other
occupations. It also reflects the agrarian social hierarchy that existed
prior to 1959, which is important for calculating the scores for
parents and older respondents, since agriculture was vital to the

area’s economy.

Table 7

Occupational Index

‘

4 = managers, professionals, proprietors and large landowners

3 = white collar workers, teachers, medium landowners

= skilled workers, small landowners

Tao

unskilled laborers, domestic workers, and peasants

38

one s
natic
used

lowe

aT0:

TTOH
SUSD

Simq

The educational and occupational indexes were combined into
one score that established the individual's rank within the local and
national hierarchy. This index is shown Table 8. The same index was
used for older and younger respondents. No individuals from the

lowest group are represented in this study

Table 8

Index of Socgl Rail;

 

Social Rank

 

7-8 IV (High)
5-6 111

3-4 11

1-2 I (Low)

 

QM Collection

A convenience sample was obtained as a result of problems that
arose when attempting to collect data in the Cuban community in
Miami. One difficulty was that Cubans from different regions are
Spread throughout the county. Thus, I could not simply go to one
neighborhood and expect to find there all the respondents that met
the established criteria.

The most important obstacle in the data collection stemmed
from the political tension that exists in the community, marked by
Suspicion and distrust. Given these circumstances, I used an approach
similar to the one Milroy (1980) used in Belfast, another community
marked by political tension. In order to collect data, I called on
friends of friends, explaining to them that I wanted to tape record a

conversation with them to include in a research project. Although

39

many respondents were suspicious of my intentions at first and
numerous recordings were not suitable because the respondents would
ask me to turn off the tape recorder whenever they wanted to say
something, a sufficient number of good samples of conversational
data were obtained.

The data were not collected in a sociolinguistic interview as
traditionally defined, but rather in a speech event called una visita,
‘a visit,’ in which one goes to someone’s house to visit and stays as
long as the conversation lasts, often several hours. Thus, between
two to four hours of tape recordings were made during each session.
The sessions usually involved four to five participants,” all of whom
belonged to the same social network. One or two of the participants
from a conversation were later used in analysis. My role was that of
participant observer. Since the data was collected during a speech
event with which the respondents were familiar, rather than an
artificial one (i.e. the sociolinguistic interview, as noted in Wolfson,
1976), the observer’s paradox (Labov, 1972) was overcome and a
large amount of casual speech was obtained.

The Respondents

Table 9 lists the respondents. Ten of the respondents were
between the ages of 28 and 35, and two were in their sixties at the
time the data was collected. There are 6 males and 6 females. The
younger respondents had been in Miami, and the United States, for
lengths of time ranging from two weeks to four years. One of the

Older respondents has lived in Miami, and the United States, for two

 

 

12See Wolfson (1976) for a discussion on using peer groups in data
Cellection.

4o

years, and the other has lived in Miami for two years and in the US.
for 28 years, in an area in the country where he had sporadic contact

with Spanish speakers.

41

 

hbh9
Magnifiers
Gen
TV Fen
BC Fen
AA Fer
l3 Tet
lib ye.
11D re
CM 3“
CA M;
13 it.
NC yq
AR yq
RD qq
\

Table 9

 

 

 

Respondents
Gender Age Occupation Parent Parent
SES origin
YM Female 61 Physician Lower Urban
BC Female 28 Dentist Upper Rural
AA Female 30 Engineer Middle Urban
LS Female 30 Attorney Middle Rural
MU Female 30 Math Teacher Lower Rural
MD Female 32 Physician Lower Urban
CM Male 69 Pharmacist Upper Urban
CA Male 33 Engineer Upper Urban
JB Male 34 Engineer Upper Urban
NC Male 31 Physician Upper Rural
AR Male 35 Physics Teacher Middle Rural
RD Male 35 Architect Lower Urban

 

42

5)

5i

. «t\

si

 

Region

All the respondents were born and raised in the northern part of
the central region of Cuba, in the area indicated on the map in Figure
3. Although all the respondents are currently living in Miami, their
speech is representative of the regional standard for two reasons.
First, all of the respondents emigrated from Cuba in their late
twenties or early thirties, well beyond the critical period for second
dialect acquisition. Studies have shown a direct link between
children’s age and their ability to acquire the complex phonological
rules of a second dialect’s (Payne, 1980; Chambers, 1992).13 Payne
(1980) suggests that even if children are born in a neighborhood,
they will not acquire certain of the local phonological patterns if
their parents were not locally born and raised.14

The second reason the respondents are considered
representative speakers is the unique dialect situation to which they
immigrated. Typically regional mobility involves situations in which
an individual moves from one dialect region to another. The dialect
situation in the Miami-Cuban community, however, arose from the

immigration of hundreds of thousands of individuals from all regions

 

13Chambers (1992) studied the acquisition of British English by
Canadian children and found that a child’s age dramatically affected
the rate of acquisition and the ability to successfully acquire rules of
pronunciation: moreover, the older the child, the less likely he or she
was to acquire complex phonological rules. In this case, the rules
were elimination of voicing of a medial /t/ following a vowel or /r/
before an unstressed syllable which produces the minimal pairs
Putting/pudding and petal/pedal (Chambers 1992:682)
.“Payne (1980) not only noted the general inability of children who
lrpmigrated to a Philadelphia neighborhood before the age of eight or
nine to learn the local dialect, she also suggested that “unless a
child’s parents are locally born and raised...the possibility of his
acquiring the short-a pattern is extremely slight even if he were to be
cm. and raised in King of Prussia” [the Philadelphia neighborhood
Studied by Payne] (p. 174).

43

andsoci
ofindiv
sodalnt
uents.:
causesb
and info

Ni
arnvals
coHeagt
panicul
country
establis
inmance
0W1] OT .
umiahz

Tl
Socia1 n
STEEch q
the Use
WhErcas

Slandan

Tl
ldentifp
analyzqr
asdeSCr

and social status groups. Although there is a certain degree of mixing
of individuals from different regions, most Cubans maintain their old
social networks, both formally through clubs that organize social
events, such as picnics and dances, and fundraising events to support
causes back home, such as the restoration of a local church in Cuba,
and informally.

New arrivals rely heavily on social network ties. Younger
arrivals immediately establish contact with their old friends and
colleagues. They also seek out their parents’ network relationships,
particularly since these are usually persons who have been in the
country for many years and have become economically and socially
established. New arrivals often benefit from these relationships. For
instance, new arrivals often find their first employment through their
own or their parents friends, relatives and former colleagues. Most
socializing occurs within the social network established in Cuba.

The sociolinguistic outcome of the re-establishment in exile of
social network ties is the reinforcement and maintenance of local
speech norms. Milroy (1987) and Edwards (1992) have proposed that
the use of nonstandard forms is promoted in multiplex networks,
whereas open social networks are associated with increased use of
standard variants.

Analysis

The analysis involved transcribing the conversational data,
identifying the tokens of each occurrence of the variables, then
analyzing and coding the tokens. Issues of reliability were addressed
as described earlier, by analyzing some passages twice at separate

times to confirm that coding was consistent. I also consulted with

44

two native speakers of Cuban Spanish, and carried out acoustic
analysis on some tokens.

A variable rule analysis was carried out using Goldvarb
(VARBRUL for the Macintosh). The use of this software as a tool in
testing and quantifying the effects of linguistic and social factors
was described earlier.

The remaining chapters of this dissertation will present and

discuss the findings for the variables (3), (r) and (l).

45

 

pf
M

if?

THE VARIABLE (s)

This chapter gives a brief background of the variable (5), and
presents and discusses the results of the variable rule analysis of
word-internal and word-final cases. The results are for the aspirated
and deleted variants; the basic variant [3] is not discussed because it
tends to occur infrequently: 3% (22/686) of word-internal and 3%
(95/2772) of word-final cases.

Background

Aspiration is believed to have begun either around the
thirteenth century as a case of transfer from Arabic (Frago, 1983), or
much later from pressure to eliminate coda consonants; there could
also have been multiple causation. Historical linguists hold two
positions on the relationship between aspiration and deletion. One
argues that aspiration arose before deletion, as a result of the general
tendency to eliminate syllable final consonants, which since the
latter part of the thirteenth century reduced the number of coda
consonants from around twenty-five to five by the XIV century. The
other position holds that aspiration arose after deletion was already
occurring. This breaks the weakening chain (5 —) h —> 6) accepted by
many synchronic linguists as the source of this variation. Ariza
(1994) states that those varieties of Spanish in which (5) varies may
either be “at a stage of intense articulatory weakening” or may be
holding on to “remnants of an old medieval aspiration” (p. 242).

Thus, the study of (s) variation must address the question of
whether or not (3) is actively undergoing change. Increased deletion
by younger speakers is often noted as evidence of change, but

generational differences found in apparent time studies may reveal

46

TL

 

age-grading rather than actual linguistic change. It is possible that
increased deletion by younger speakers is not only caused by
pressure from within the linguistic system, but also by external
factors motivating its use, particularly since covert prestige is
attached to deletion because of its association with working class and
rural speech, and with the new nationaLsm, a point which will be
argued at length below. In the chapter on the variable (r), it will be
shown that its lateral variant, a formerly stigmatized feature
associated with lower status and rural speech, shows a marked
increase in the speech of younger respondents. For the deleted
variant of (s), a marked generational difference would provide
additional evidence that external pressure is toward nonstandandard
phonological variants previously associated with the lower status

norm.

Results and Discussion

The results discussed below are for aspiration and deletion in
word-internal and word-final environments. Separate analyses were
carried out on these because of several important differences: for
instance, word-finally, the following phonological environment may
be a consonant, vowel or pause, and (5) may be an inflectional
morpheme, but word-internally it only occurs preconsonantally and
cannot be an inflectional morpheme.

Word-internal (s)

Linguistic Factors

The analysis of word-internal (3) showed that neither the

following sound (necessarily a consonant) or the preceding one were

47

signifit
represe
reporte
results
voicele
voicele
corona
nasals.
the fol
interna

5
variati
ofa Sll
Part of
results
points

TMEnsi

UngUar
Wall.
maje,

IIllelq.

significant factors in the application of aspiration when deletion
represented nonapplication. Earlier studies (i.e. Poplack, 1981)
reported that deletion was favored by voiced consonants. Similar
results were found here: aspiration was strongly favored by a
voiceless velar consonant [k] (.847), slightly favored by the
voiceless labial [p], and it was slightly disfavored by the voiceless
coronal [t]. The strongest negative effects were from the voiced
nasals, [m] (.263) and [n] (.105). In general, however, the nature of
the following segment did not contribute significantly to the word-
internal variation of aspiration and deletion.

Stress is the only significant factor contributing to the
variation. As the results in Table 10 indicate, when (s) is in the coda
of a stressed syllable, this favors aspiration (.615), whereas being
part of a nonstressed syllable, slightly disfavors it (.437). These
results were also noted by Alba (1982) for Dominican Spanish; he
points out that these findings are not surprising given the greater
intensity and duration created by primary stress. Catalan (1989) also
notes the role of stress on aspiration, but describes it in terms of
dissimilation. Guitart (1982) within a larger argument on the non-
teological nature of linguistic change in Caribbean dialects, argues
that retention, aspiration, and deletion show the effects of
“distributions of energy that are beyond the speaker’s control in
unguarded speech” (p. 68). He notes, “if there is less energy some
or all of the gestures connected with the production of /s/ will not be
made ...” (p. 68). Since stress increases energy, aspiration is more

likely to be retained in stressed syllables.

48

Table 10

Word-internal Aspiration: Stress

 

 

Factor Weight Applications
No stress .437 428/502 (85%)
Stress .615 252/272 (73%)

 

Note. Input = .914; x202, m = 774) =
18.16, a < .05; x2 cell 0.82

 

Social Factors

Social factors made the strongest contribution to the
application of aspiration word-internally. All the factors were
significant, and, with the exception of age, they tended to interact:
gender with status, and status with origin.

The results for age in Table 11 show that older speakers
strongly favor aspiration (.800), and younger ones disfavor it (.365).
Earlier studies found that aspiration was much more frequent than
deletion in this word-internal position: Fails (1984), using data from
the late 1960s, reported 77% aspiration and 10% deletion; Terrell
(1979) found 97% aspiration and 3% deletion. While older
respondents’ use of aspiration is similar to that reported by Terrell
(1979), the younger respondents showed 16% deletion. These results
suggest that deletion is spreading to environments in which it was

previously disfavored, such as the preconsonantal word-internal one.

49

Table ll
W191
Factor

Younger
Older
m lnpu
18 16. [n
The
resuhs inc
toaspirao
tendency,
meyhave
The lower
negative .
CODmonlj

Table 11

Word-internal Aspiration: Aga

 

 

Factor Weight Applications
Younger .365 465/553 (84%)
Older .800 215/221 (97%)

 

Note. Input = .914; x2(12,1_s1_ = 774) =
18.16, [n.s.]; at” cell 0.82

 

The results for gender and status are shown in Figure 4. These

results indicate that upper status females have the strongest tendency

to aspirate word-internal (8). Middle status males show a weaker

tendency, but a positive one, nonetheless, which is unusual because

they have a higher likelihood of using aspiration than females from

the lower and middle status groups. Lower status males show a strong

negative tendency, which is not surprising since this group is

commonly found to use more nonstandard features; this pattern is

confirmed throughout the current study.

50

 

 

Word-int.
[as]; 1'
1157132:
llale lllll

 

FU FM FL MU

MM ML

Eiggra 4. The contribution of gender and status to the application of
word-internal aspiration. Input = .914; x202, fl = 774) = 18.16,
[n.s.]; x2 cell 0.82. Female upper (a = 39/42), Female middle (a =
115/132), Female lower (a = 196/215), Male upper (a = 259/300),
Male middle (a = 62/69), Male lower (a = 9/16).

51

The results for the contribution of status and origin to the
application of word-internal aspiration are shown in Figure 5. Middle
status individuals whose parents are of urban origin have the
strongest tendency to aspirate. Upper status individuals of urban
origin also favor aspiration, but only slightly. Lower urban status has
a neutral effect. Rural origin in general disfavors aspiration, which is
expected since higher rates of deletion is associated with rural
origins.

Upper status of rural origin has the most negative effect on
aspiration in this environment. These results were unexpected
because deletion is commonly associated with lower status and these
are upper status individuals whose parents belonged to the upper
middle class. At the same time, these results can be explained by the
historical association of deletion with rural speech.

The question that arises is why upper status young males whose
parents also belonged to the upper middle class would chose to
identify with a negatively evaluated rural feature rather than
dissociate from it. The answer to this question can be found in the
positive effects of rural ties on an individual’s quality of life in
Cuba, which makes rural origins prestigious within the local culture,
and rural sociolinguistic variants an expression of that prestige.

Individuals from upper middle class homes prefer to identify
with their rural origins rather than with urban ones because having
roots in the surrounding rural area makes life much easier, as
locating food to supplement scare government rations is a basic and
time consuming part of life. One young male respondent after

complaining about how expensive food was (he had remarked earlier

52

0.9 -
0.8 -
0.7 -
0.6 -
0.5 -
0.4 -
0.3 -
0.2 ~

 

 

 

 

 

 

 

 

 

 

 

UU UR MU MR LU LR

Figure 5. The contribution of status and origin to the application of
word-internal. Input = .914; x2(12, N = 774) = 18.16, [as]; x2 cell
0.82. Upper urban (a = 231/258), Upper rural (a = 67/84), Middle
urban (a = 57/59), Middle rural (a = 120/142), Lower urban (a =
60/76), Lower rural (a = 145/155).

53

“(J

in the conversation that half a pound of flour cost 200 Cuban pesos,
but the average person earned 140 pesos a month) noted that it was
actually better to be unemployed than employed in Cuba, because if
you worked you could not make enough money to buy necessary items
and, more importantly, since you worked all day, you did not have

time to obtain them in some other way:

A person who normally makes 148 pesos and has to work during
the eight most important hours of the day, which are from eight
in the morning to five o’clock in the afternoon, what is he
glping to do? Why work? He simply doesn’t have anything, and
t at’s it.

Upper status respondents whose parents were originally from
the country, however, do not have to worry about government rations
because they receive most of what they need (e.g. milk, meat, fruits
and vegetables) from what is left of the family farm. Goods from the
farm are also battered or sold in order to obtain other foods that
cannot be grown there, e.g. clothing, buttons, yarn, thread, auto
parts, gasoline, or just about any item an individual may need to
subsist.

Additional linguistic evidence for the positive association with
rural origins can be found in another rural feature which appears in
the data, but which was not studied in this dissertation: the
aspiration of syllable initial /s/. This is a rural feature much more
strongly stigmatized than deletion because it is mostly restricted to
ruralspeech.

This feature was frequently used by one of the young male
respondents, a physician, whose mother and father, who was also a
physician, were both raised .in the surrounding rural area. Some

examples of his use of it include [nohotro] for [nosotros] nosotros

54

‘we,’ [pahate] for [pasaste] pasaste ‘you passed,’ [pobrehita] for

’

[pobresita] pobrecita ‘poor dim. fem., and [entonhe] for
[entonses] entonces ‘then.’

The only other respondent who used intervocalic [h], was
another young male, an engineer, whose father was an attorney and
mother a teacher. His parents both were from highly respected upper
middle class families. Neither of his parents, nor any of his
grandparents, for that matter, were of rural origin. The young man
was very actively involved in covert economic activities to
supplement the meager income from his profession. Some of the
activities he told me about included subsistence farming, and making
soap and shoes, to name a few. It seems that because of this young
man’s involvement in survival activities, he identifies more with
rural values than with the upper middle class urban ones of his
parents. It should also be pointed out, however, that this young man’s
parents were very involved in revolutionary politics; thus, he may be
expressing not only his stance with regard to rural values, but, more
importantly, his stance in regard to the political system. This
important issue is discussed in more detail below.

Word-final (5)

Two separate analyses of word-final (s) were carried out: the
first examined the variation of (5) according to whether it was a
grammatical or nongrammatical morpheme; and the second one was

carried out on inflectional morphemes only. The results from each

are presented separately.

55

till
tie

for

The Effects of Linguistic Factors on the Vaflation of Word-final (s)

The results of the variable rule analysis of word-final (s) are
shown in Table 12. The aspirated variant [h] represents application
and [a] nonapplication. Preceding sound was eliminated in the
stepwise multiple regression. Interaction was found between the
factor groups following environment (consonant, vowel or pause),
stress (stress or no stress) and morpheme type (grammatical or
nongrammatical) .

1 While previous studies on Cuban Spanish (i.e. Fails, 1984;
Terrell, 1976) found that a following consonant or vowel favored
aspiration word-finally, these results indicate that aspiration depends
not only on the following segment, but also on whether (5) is an
inflectional morpheme and whether it appears in a stressed syllable.

The results in Table 12 show that a following consonant
promotes aspiration if it is a grammatical morpheme in a stressed
syllable (.795), and slightly less if it is a grammatical morpheme in a
nonstressed syllable (.667). A following consonant also favors
aspiration when it is nongrammatical but stressed (.649). It is not
favored, however, when a consonant follows and (s) is

nongrammatical and not stressed (.307).

56

Table 12

Word-final (s): Linguistic Factors

 

 

Factor (Morpheme, Following, Stress) Weight Applications
Grammatical Vowel Stressed .829 113/152 (74%)
Grammatical Consonant Stressed .795 179/259 (69%)
Grammatical Consonant No stress .667 501/906 (55%)
Nongrammatical Vowel Stressed .655 46/86 (53%)
Nongrammatical Consonant Stressed .649 97/189 (51%)
Grammatical Vowel No stress .430 97/306 (32%)
Nongrammatical Consonant No stress .307 40/153 (26%)
Nongrammatical Pause Stressed

.246 19/95 (20%)
Grammatical Pause Stressed
Nongrammatical Pause No stress

.127 45/447 (10%)
Grammatical Pause No stress
Nongrammatical Vowel No stress .074 5/84 (6%)

 

Note. Aspiration represents rule application.

Input = .389; 71205, n = 2677) = 56.85, [sig.]; x2 cell 0.99

 

57

The results for prevocalic (s) are interesting because they
reveal the strongest positive and negative constraints on [b]. When a
vowel follows and (s) is grammatical and not stressed, aspiration is
slightly disfavored (.430), but a very strong negative effect occurs
when a vowel follows and (s) is nongrammatical and not stressed
(.074). At the same time, the strongest positive contribution is made
when a vowel follows and (s) is an inflectional morpheme in a
stressed syllable (.829).

Terrell (1979) noted that deletion occurred more before vowels
than before consonants or pauses, and argued that this could not be
explained with phonological constraints, but rather with functional
grammatical ones. The results described here suggest that
phonological and morphological factors interact to constrain the
occurrence of [a] before vowels; thus, grammatical constraints must
be reconsidered.

As noted in Cedergren (1973), Hammond (1977), and Poplack
(1981) a following pause disfavors aspiration. The results Table 12
suggest, however, that the degree to which it does so depends on the
presence of stress: in a stressed syllable the negative effect is strong
(.246), but not as strong as in a nonstressed one (.127). Whether (s)
is grammatical or not was not significant before a pause. Guitart
(1982) argued that linguistic change in Caribbean Spanish is not
oriented toward maximizing communicative clarity, but rather that it
often occurs “in apparent disregard of their communicative needs” (p.
69). Uber (1984) found that aspiration was too perceptually weak to
convey information about number. Uber (1989) showed that noun

phrases not phonologically marked for plurality are not usually

58

an

ambiguous because information about number can be gleaned from
contextual elements.

The tendency to delete (s) before a pause in spite of its being a
grammatical morpheme could very well be accounted for with
functional constraints since (5) is more likely to be redundant in this
environment than before a vowel or consonant because it may be the
last item of a noun phrase and redundantly marked for plurality, as
shown below in (1) and (2). Some counterexamples are shown in (3)
and (4). (3) shows deletion of every plural morpheme in the noun
phrase. (4) shows retention of prepausal (s) and deletion of the
previous ones.

1. La[h] bota[h] esa[h] verde[e]

I_.2_ts botfiagsesas verdes
The boots those green
‘Those green boots’
2. Entonce[z] imaginate la[h] cara[h] de eso[s] nifio[e].
Entonces imaginate las caras de esos nifios.
Then imagine the faces of those boys
‘Then, imagine those boys’ faces.’
3. El tocaba en toda[o] la[e] puerta[o]
El tocaba en todas las puerta§_.
He knocked on all the doors

‘He knocked on all the doors.’

59

4. Habia dicho que le mandara uno[e] cepillo[o]

semiduro [ s]

Habia dicho que le mamdara unos cefllos semiduros.

 

(3rd pr sg) had said to send him some brushes semihard
‘He said to send him some semihard toothbrushes.’

Previous studies (e.g. Cedergren, 1973; Poplack, 1981; Terrell, 1979)
found that earlier marks of plurality in the noun phrase, usually
determiners, were more likely to be aspirated. Although not reported
on here, I also found for these data that the determiner was the only
noun phrase item that promoted aspiration. Arguing against
functionalist proposals, Guitart (1982) offers a simpler alternative
explanation based on human physiology: “the speaker has more
breath at the beginning of an utterance than in the rest” (p. 69).

Although logical, this does not explain the fact that the
prepausal environment is the one in which the basic variant [5]
occurs most frequently (Fails, 1985; Terrell, 1976). I would like to
suggest that the application of [s] is strongly influenced by
discoursal factors as well as by resyllabification in prevocalic
environment, as noted in earlier studies (e.g. Cedergren, 1973;
Terrell, 1976) and which was found to account for most cases of [s]
in these data, too: [los otroe] los otros ‘the others;’ [diesiseio]
dieciséis ‘sixteen;’ [ehtados unidoh] Estados Unidos ‘United
States.’

Discourse factors, namely reported and emphatic speech, also
Play a role in retention of the basic variant. Examples (1) and (2)
below show retention in reported speech. (1) is from a young male

respondent; it shows [3] only on the final item in the sentence. (2) is

60

from a young female respondent who is paraphrasing her
grandmother.
1. Entonces él preocupado (dice), caballero como sacamo[e]
nosotro[e] de alli lo[h] quintale[h] de arroz[s]?

Then, worried, he says, “min how are wegoing to get those

hundreds of pounds of rice out of there?”

2. Y dice, e[s]ta porque e[s]ta e[h] la mejor.

And she says, “this one because this one is the best.”
Examples (3) and (4) show [8] used for emphasis. (3) is from a
young male respondent relating his excitement at the thought of
making a killing by putting a fragrant soap on the black market. (4)
is from a young female respondent explaining the unexpected size of
an ovarian cyst she had had removed.

3. Si tu te aparece[e] con un jabbn de olor, iacaba[ s]te!
If you show up withwr of soap with fragmicegyou mak_e

a killing!
4. Un qui[h]te de ovario ba[ s]tante grande.

A pretty big ovarian cyst.

In conclusion, these results indicate that grammatical (s) in a
stressed syllable is more likely to be aspirated than a nongrammatical
one in a nonstressed syllable, except when it precedes a pause, in
Which case deletion is favored. Terrell (1979) also found that
nongrammatical cases were deleted more than grammatical ones, but
the difference was not statistically significant. Cedergren (1973) also

found more deletion in nongrammatical cases.

61

 

The Effects of Social Factors on the Variation of Word-final (s)

The results for the effects of social factors are described
below. The factor group representing rural or urban origin was
eliminated after the stepwise multiple regression indicated that it did
not make a significant contribution to the variation. Unlike word-
internal position, origin did not interact with status. Interaction was
found between gender and age, and gender and status.

The results for gender and age, shown in Figure 6 revealed that
older females have a strong positive tendency towards assimilation
(.730), whereas older males have a much weaker one (.573). Younger
females also have a small favoring effect (.532), while younger males
showed a tendency to disfavor aspiration (.362). Males in both age
groups show a less favorable tendency towards use of the aspirated
variant, which is not surprising since males in general tend to use

more nonstandard variants.

62

 

Figure 6. Age and gender and the application of word-final (s). Input
= .389; x205, u = 2677) = 56.85, [sig.]; x2 cell 0.99. Older female
(3 = 199/327), Older male (3 = 125/279), Younger female (3 =
507/1098), Younger male (11 = 311/973).

63

The prestige of aspiration is also evident in the results for
gender and status, shown in Figure 7. It is favored by upper status
females (.746). Other gender and status groups have only a slightly
negative effect: females of middle and lower status (.468) and males
of upper and middle status (.477). The results also show a favorable
weight for lower status males (.728), which is unusual and does not
fit the pattern expected here. This finding is most likely the result of
hypercorrection and the limited number of tokens in this cell.

Vallejo-Clams (1971) found that lower status speakers from the
central region had slightly higher rates of aspiration than upper and
middle status ones. Since Vallejo-Claros found little variation across
social status groups, in all three regional varieties (e.g. the western,
central and eastern ones, indicated in the map in Figure 1), he

concluded that (s) was not a social differentiator.

64

 

 

 

 

 

 

 

 

 

FU FM FL MU MM ML

Figure 7. The contribution of gender and status to the application of
word-final (3). Input = .389; x2(15, 11 = 2677) = 56.85, [sig.]; x2
cell 0.99. Female upper (a = 136/221), Female middle and lower (a =
570/1204), Male upper and middle (a = 423/1223), Male lower (a =
13/29)

65

This suggestion must be questioned because the results reveal
unusual sociolinguistic behavior: for instance, he reports very high
rates of [s] in both interview and casual styles (e.g. for casual style
he found retention rates of 50% for the upper middle class, 47% for
lower middle, and 44% for the working class), and more retention
than aspiration, whereas other studies (i.e. Terrell, 1979; Fails,
1985) reported much lower rates of [s], and more [h] than [s]. For
instance, Terrell (1979) found only 3% retention both word-
internally, and word-finally. Thus, Vallejo-Claros’ findings may not
have been representative of actual sociolinguistic patterns.15

It is suggested below in the discussion of the lateral variant of
(r) that the word-final environment is more salient sociolinguistically
than the word-internal one. The behavior of lower status males with
regard to (s) aspiration provides evidence for this: while aspiration is
favored word-finally (.728), as shown Figure 7, it is very strongly
disfavored word-internally (.177), as shown in Figure 4. Whereas the
word-final environment may be controlled to some degree,
particularly if there is a socially stigmatized variant to be avoided,
as in the case of the deleted variant of (5), control over the word-
internal one is much more difficult to achieve.

There is reason to believe that the individual who represents
the lower status male was exerting some measure of control over his

use of word-final (s). It was obvious that he was aware of his speech

 

”After listening to some of the interviews carried out by Vallejo-
Claros in 1968-1969 for the large dialectology study called Ih_e_
Coordinmd Study ofthe Linguistic Norm of the Principal Cities of
Latin America and of Spain and which he later used in his
dissertation, I believe one of the main problems is that he does not
often manage to overcome the observer’s paradox.

66

and of being recorded: for instance, he commented on how correctly16
he was speaking on the recording in comparison to how he usually
spoke, and he even went as far as to give an imitation of himself,
which included high rates of (r) lateralization and (s) deletion, both
of which tend to be negatively evaluated. He was anxious to have the
tape recorder turned off and requested this not long after taping had
begun. This behavior reveals the respondent’s linguistic insecurity.
He is well aware of [s] as the prescribed norm, an issue which is
addressed even in first grade reading textbooks, and he manages to
control word-final occurrences by avoiding deletion and aiming for
aspiration since he could not have used the full pronunciation in
conversation without sounding affected or effeminate.l7
Inflectional (s): Linguistic Factors

The results in Table 13 show that whether (s) is a verb or plural
inflection is only significant when a consonant follows and (s) is in
the coda of a stressed syllable, in which case a verb inflection
strongly favors aspiration (.804) and a plural one slightly disfavors it
(.463). Cedergren (1973) found that the second person singular
inflection was aspirated more than the plural inflection, and that

deletion occurred more in the plural inflection.

 

l6See Preston (1996) for a discussion of prescription as it relates to
the linguistic awareness of nonlinguists.

17The results, not discussed here, for word-internal [s], for instance,
revealed that gender was the only significant factor group; females
had a.strong positive effect on retention (.768) and males a strong
negatlve one (.236).

67

Table 13

Word-final (s) in Plural and Verb Grammatical Morphemes:

Linguistic fictors Involved in the Application of Aspiration

 

 

Factor (Inflection, Following, Stress) Weight Applications
Plural Vowel Stressed

.807 113/152 (74%)
Verb Vowel Stressed
Verb Consonant Stressed .804 167/228 (73%)
Plural Consonant No stress

.631 500/905 (55%)
Verb Consonant No stress
Plural Consonant Stressed .463 11/30 (37%)
Plural Vowel No stress

.393 97/306 (32%)
Verb Vowel No stress
Plural Pause Stressed

.236 9/39 (23%)
Verb Pause Stressed
Verb Pause No stress

.107 39/396 (10%)

Plural Pause No stress

 

Note. Input = .426; x203, 151 = 2056) = 39.68, [Sig]; x2 cell 0.86

 

68

For the other factors, a following vowel combined with the
presence of stress favors aspiration (.807), but a following vowel
without stress disfavors it (.393). A following consonant with no
stress favors aspiration (.631). When a pause follows, it is
disfavored, although more strongly when the syllable is not stressed
(.107) than when it is (.236).

These results indicate that following environment is not alone
in contributing to variation of word-final grammatical (s). For
instance, a following consonant favors aspiration if the syllable is
stressed and (s) is an inflectional verb morpheme, but not if it is a
plural morpheme. Furthermore, the negative effects of a following
pause on aspiration are stronger in the absence of primary stress.
Social Factors4and the Va_r___iation of Inflectional (s)

Table 14 shows the contribution of social factors to the
variation of (s) when it is an inflectional morpheme. These results
follow the same pattern as for word-final (s). The factor groups
gender and age, and gender and status interacted. Older females have
the strongest tendency toward aspiration (.752) when (s) is an
inflectional morpheme. Older males have a small favoring effect
(.567), as do younger females (.530). As expected, younger males
disfavor aspiration (.358). In both age groups, males disfavor
aspiration more than females, revealing the general tendency for

males to use more nonstandard variants.

69

Table 14
Word-finwnflectiml (s): the Effects of Social Raptors on the

Application of the Aspirafled Variant

 

 

 

Group Factor Weight Applications
Older female .752 171/262 (65%)
Older male .567 98/213 (46%)
Gender and Age

Younger female .530 409/818 (50%)
Younger male .358 258/763 (34%)
Female upper .762 111/163 (68%)
Male lower .704 9/20 (45%)
Male middle

Gender and Status .478 347/956 (36%)
Male upper
Female middle .478 211/467 (45%)
Female lower .455 258/450 (57%)

 

Note. Input = .426; x203, 131 = 2056) = 39.68, [sig.]; x2 cell 0.86

 

7O

The results for gender and status are the same as those for
grammatical versus nongrammatical (s). As anticipated, upper status
females favor aspiration (.762); a small tendency to disfavor it is
found for females of middle status (.478), females of lower status
(.455), and males of upper and middle status (.478). The favorable
tendency found for lower status males was not expected. It was noted
above that this finding most likely resulted from hypercorrection and

that it should be questioned because of the small number of tokens.

Summary Conclusions

These results suggest that stress is the only linguistic factor
that contributes to word-internal variation of [h] and [e]. The
features of the following sound did not effect the variation. In word-
final position, stress interacted with morpheme type and following
environment, revealing significant differences in variation according
to the combination of factors. Thus, the effects of a following
consonant or vowel vary according to which factors they co-occur
with. Before a consonant aspiration is favored except when (s) is
nongrammatical and in a nonstressed syllable, in which case it is
disfavored. A following vowel has a strong positive effect on
aspiration when (s) is a grammatical morpheme and in a stress
bearing syllable, but it has a strong negative effect on rule
application when (s) is nongrammatical and not in a stressed syllable.

The findings for the social factors did not reveal strong
generational differences in either word-internal or word-final
environments. Age had a larger effect on word-internal aspiration,

which was favored by older speakers, than on word-final variation.

71

Age interacted with gender for word-final (s), revealing that males in
both age groups had lesser probabilities for application of aspiration
than females from the same age group: the older males preferred
aspiration less than the older females, and the younger males less
than the young females. Older females had the strongest tendency for
aspiration and young males showed the strongest tendency towards
deletion.

The interaction of gender and status showed that upper status
females strongly favored aspiration, a result which was not surprising
since aspiration is the educated norm (Terrell, 1979), but it was also
expected that middle status females would favor aspiration, too, but
they slightly disfavored it. Aspiration is used by educated females
whose parents belonged to the upper middle class as a symbol of that
status since there are few other means for them to express it. (It will
be noted later that upper status females show a nearly categorical
preference for the aspirated variant of (r), only favored by one other
group, young females of middle status, and that this variant also
appears to be an expression of status.) Trudgill (1972) suggested that
females use more standard variants to compensate for a general lack
of outward symbols of status. This may be especially true in Cuba,
where both genders find their ability to express status limited, at
best.

It was proposed that the strong motivation for young college-
educated males to use more nonstandard variants is an expression of
locality and solidarity. Their lifestyle is more similar to a working
than a middle class one, and their lives are different in nearly every

aSpect from those of previous generations of college-educated

72

Cubans. For instance, young males cannot provide for their families
with only their earnings and government rations. Some of the
respondents received food and other goods from their families,
particularly if the family had a farm. One young male respondent’s
father was a veterinarian at a government office, who illegally
attended to farm animals on the side, receiving meat and produce in
exchange for his services, and passing along goods to his two married
children.

Most young males engage in activities that add to their income
or supplement rations. They often carry out subsistence farming: they
raise chickens, or a pig if they can manage to feed it; they grow
vegetables, beans and other staples. One respondent laughed about
his failed attempts at growing rice. Another, a physician, told me
how a pig for which he had barely been able to find enough scraps
had finally died even after he injected her with medicine he had
taken from his work place. Another respondent, an engineer whose
parents were upper middle class from families that enjoyed a great
deal of respect in the community, told me how he managed to learn to
make shoes with his mother and two brothers and sell them, although
he was indignant that the price they fetched bought only a small

amount of food:

We even managed to sell a pair of shoes for 200 pesos. One
person’s salary for a month. No, but that’s not all, you would
sell the pair of shoes, walk cross the street and it would buy
you a pound of flour, no not a pound of flour, a little can
because that wasn’t a pound, it’s a can of condensed milk. Two
little cans of flour were 150 pesos.

73

This respondent also told me about his experience with making soap,
at which he never succeeded because he had a difficult time finding
the ingredients.

Mode of transportation is another example of the working class
quality of life for college-educated young adults. Most of the
respondents used bicycles as their primary source of transportation.
Only one of the five young males had a car, while another had only
access to a car that belonged to his older brother. Both were from
upper middle class families who had sold off their jewelry and other
possessions on the black market in order to get together enough
money to buy a car for the eldest son. None of the young females
owned or had access to a car. One young female dentist told me how
she would ride her bicycle every morning from one end of town to
the other with her two little boys, the youngest on the handlebars and
the oldest on the back fender, in order to drop them off at relative’s
homes before she pedaled out of town to practice dentistry in the
surrounding countryside. Another female told me that to get to work
in the morning she would first catch a bus, then she would hitch a
ride on a truck and then make the last leg of her journey by tractor.
Another respondent told me how he bought a slaughtered pig and
rode his bicycle for miles carrying it on his back to get it home.

This suggests that the quality of these college-educated
professionals lives is not middle class as this was traditionally
defined in Cuba. The evidence found here for increased by younger
speakers, especially males, may result from solidarity and association
with working class and rural values. Generational differences were

not strong, but external motivations coupled with internal pressure

74

 

toward open syllables appear to be pushing deletion beyond its

previous linguistic and social distribution.

75

Q

THE VARIABLE (r)

This chapter presents and discusses the results for the four
variants of the sociolinguistic variable (r): assimilation [assim],18
lateralization [l], aspiration [h] and the basic variant [r].
Examples of these are shown in Table 15. Further examples of

assimilation are shown in Table l in the introductory chapter.

Table 15

Mmples of the Variants of (r)

 

Variant Examples

 

[r] [carne] carne ‘meat’

 

[assim] [cadne] ‘9 carne ‘meat’

[1] [calne] carne ‘meat’

 

[h] [cahne] carne ‘meat’

 

In the following sections, each variant is discussed separately,
as the binomial version of VARBRUL only allows two factors in the
dependent variable; one variant is selected to represent rule
application, and the other variants represent nonapplication. Thus,
for example, when [assim] is selected as the value for rule
application, the three other variants that could apply in the specific
environment, [1], [h] and [r], are grouped together in the value for

nonapplication.

 

”Choy. Lopez (1986) uses [ACR] asimilaciones en contacto
regresivas sfor all allophones arising from regressive assimilation.
19This interpretation is based on Guitart (1976); other interpretations
of assimilation (Almendros, 1958, Goodgall de Pruna, 1970a)
consider complete gemination the outcome of liquid assimilation
[canne]] carne ‘meat.’

 

76

Position in the Word
Syllable-final (r) can occur word-internally (i.e. parte ‘part,’
persona ‘person,’ Carlos ‘Carlos,’ carne ‘meat’) or word-finally

(buscar ‘search,’ leer, ‘read,’ mejor ‘better,’ mujer ‘woman’). In the

 

word-internal environment, (r) always precedes a consonant, while
word-finally it may precede a consonant, vowel or pause, depending
on the onset of the lexical item which follows. Word-final (r) may be
the inflectional morpheme for the infinitive form of the verb (i.e.
XML ‘live,’ paw ‘eat,’ h_abla; ‘talk,’ asa; ‘use’) or part of a
monomorpheme (primer ‘first,’ pg ‘for,’ M ‘woman’). Given the
different following environments and the fact that word-final
occurrences may be inflectional, cases of word-final and word-
internal (r) were analyzed separately in an attempt to improve the
statistical model of (r) variation.
Infinitive + Clitic Combinations

In Spanish, clitics may attach to infinitive forms of the verb.
The Spanish clitics, shown in Table 16, are object pronouns that
attach themselves to infinitives (llamarte ‘call you’), gerunds
(llamandote ‘calling you’), and other verb forms marked for person,

such as the imperative (vete ‘go away’).

77

Table 16

Phonological Surface Structure of Spanish Clitics

 

3pers 2pers 1pers

 

 

 

 

m f m f m f

ACC SG 10 la te me

PL los las osa nos

SG le te me
DAT

PL les os nos

SG te me
REF se

PL 08 nos

 

Note. From Harris (1995, p. 174)

aos is not used in Cuban Spanish

 

The environment created by the juxtaposition of the infinitive
(r) and the attachment of a clitic is treated as word-internal because
it is always preconsonantal, and because it is not clear what type of
boundary is created by the attachment of the clitic to the infinitive.
Posner (1996) states that most Romance linguists agree that clitics
are “affixes that have become an integral part of the verb,” (p. 174)
and that Romance languages have clearly been moving toward
affixation, with French being at the most advanced stage Table 17
shows minimal pairs illustrating the similarity of the infinitive +
clitic and word-internal environments. Further evidence for the

infinitive + clitic combination as a word-internal environment for (r)

_.- -u _ ..—-..
_ ”Eh—Hi -—-
( __..__._¥--

is found in the distribution of variants in the two environments,

which will be discussed below.

78

Table 17

Infinitive + Clitic as a Word-internal Environment for /r/

 

 

Word-internal Infinitive + clitic
p_a_rta ‘part’ m ‘give you’
M ‘meat’ amt; ‘give me’

fanaa ‘farce’ d_a_rs_e_ ‘give one’s self’
M ‘joke’ glar_l_e ‘give him/her’

 

Note. Examples are from Poplack (1980, p. 152)

 

Distributionand Frequency of (r) Vaflants in Word-internalLWord-
final and Infinitive + Clitic Environments

The variants of (r) that occur in word-internal, word-final, and
infinitive + clitic environments are shown Table 18. All four variants
occur in the word-internal environment. The most frequently
occurring variant is the basic one 48%, followed by the assimilated
variant 41%. The lateral variant only appears in 7% of cases and the
aspirated variant in 4%. The variants that appeared in infinitive +
clitic forms were [r], [assim] and [h], and their frequencies are
more or less similar: [r] 38%; [assim] 31%; and [h] 31%. There
are no cases of the lateral variant in this environment. Word-finally,
the basic variant is the most common, appearing at a rate of 76%.
Lateralization and assimilation occur in 11% and 12% of cases
respectively. The aspirated variant does not occur in this position.
This has been noted by Fails (1985) and Garcia Gonzalez (1985) for
Cuban Spanish, but this differs from other dialects of Caribbean

Spanish, in which aspiration occurs word-finally—for instance, for

79

Puerto Rican Spanish, Poplack (1981) reports 8% aspiration word-

finally.

Table 18
(r) Variants Word-internally and Word-finally

 

Word-internal Infinitive+ Clitic Word-final

 

[r] 48% (379) 38% (26) 76% (478)
[assim] 41% (320) 31% (21) 11% (71)
[1] 7% (56) N/A 12% (76)
[h] 4% (30) 31% (21) N/A

 

A comparison of the distribution of the variants in word-
internal and infinitive + clitic combinations suggests that classifying
the latter as a word-internal environment for (r) is justified because
the aspirated variant occurs in both environments, but not word-
finally.

In conclusion, the behavior of (r) in the environment created by
the juxtaposition of the infinitive and the clitic suggests that this
environment be treated as word-internal. Although it may have been
preferable to do so, a separate analysis could not be carried out for
(r) followed by a clitic because there are only 66 cases, not enough

to build a separate statistical model.

Assimilation
This section provides a brief background on assimilation, and

describes and discusses the results for word-internal and word-final

(r).

80

Background

The assimilation of the rhotic liquid to a following consonant
is reported to have appeared before 1044 (Alonso, 1961). The earliest
cases of assimilation were to the lateral in the onset of a following
clitic. Indeed, assimilation of /r/ to the lateral of a following clitic is
common in many varieties of Spanish, including those of Chile,
Argentina, the Dominican Republic, Puerto Rico and Andalusia.
However, it is important to distinguish between this type-of“)
assimilation, as a relic which is not involved in change, from the
generalized regressive assimilation which characterizes certain
regional varieties of Cuban Spanish.

Liquid assimilation is reportedly the most salient phonological
feature in Cuban regional dialect differentiation. Historically, it was
a feature of the western dialect region, which was nearly categorical
in the speech of individuals from all social status groups, including
the upper middle class (Almendros, 1958; Guitart, 1976, 1982). The
linguistic work which most extensively discusses liquid assimilation,
Guitart (1976), was based on the speech of university educated
individuals from Havana, which is at the core of the western regional
dialect.

Although long considered a feature of the western variety,
Almendros (1958) noted that liquid assimilation was spreading to the
east and claimed that a few cases of it could be heard in Santa Clara,
a central city that is included in the area studied in this dissertation.
Almendros’ prediction was borne out: by the mid-1980’s Choy L6pez
(1986) noted that assimilation occurred in all regional varieties,

although at varying rates, tending to be more frequent in the western

81

dialect than in the central one, but more frequent in the central one
than in other varieties to the east.

The current study examines assimilation in young adults in
their late twenties to mid thirties and in older speakers over the age
of sixty, in order to investigate the degree to which assimilation has
advanced in the central variety. The results of the variable rule
analysis of assimilation in word-internal and word-final positions are
discussed below.

Results and Discussion

The factor groups selected as significant in the stepwise
multiple regression analysis for assimilation in the word-internal and
word-final environments are shown Table 19 and Table 20. Word-
internally the significant factor groups were following environment,
preceding sound, stress, gender, age, and social status of parents.
The factor groups representing syllable boundary vs. clitic boundary,

and respondents’ parents’ origin were not significant.

Table 19

_S_i_gnificant Factor Grogs for Word-internal Assimilaiion

 

Significant Not significant

 

Following environment Syllable vs. Clitic Boundary
Preceding sound Origin
Stress

Gender, Age, Status

 

These results suggest that whether (r) occurs before a syllable

boundary (parte ‘part’) or in the environment created by the

82

juxtaposition of (r) and a clitic (dart_e_ ‘give you’) is not relevant for
the application of assimilation. In light of the earlier discuss on the
infinitive + clitic environment, it can be concluded that this type of
environment does not require a separate analysis since it appears to
behave like the word-internal one. Indeed, these same results were
found for all the other variants of (r), [l], [h] and [r].

Assimilation word-internally is sensitive to linguistic
constraints, including the consonant following (r), the vowel that
precedes it, and whether it is in a syllable bearing primary stress.
The individual factors that favor assimilation will be discussed in
greater detail below.

The factor groups representing the social variables interacted
for assimilation: thus, gender, age and parents’ status together
represent important input for the variation of (r) and the application
of assimilation.

Word-finally, gender and age were the only significant factor
groups. The factor groups which were not significant were following
environment, preceding sound, stress, morpheme type (grammatical

versus nongrammatical), parent status, and parent origin.

83

Table 20

Significant Factor Groups for Word-final Assimilation

 

 

Significant Not significant

Gender Following environment

Age Preceding sound
Stress

Morpheme type
Status

Origin

 

Whereas assimilation is constrained by both linguistic and
social variables word-internally, only the social factors gender and
age are relevant word-finally. In other words, the linguistic
constraints that contributed to word-internal variation were not
significant in word-final position. Word-finally, assimilation is not
constrained by the following or preceding sound, the presence or
absence of stress, the morphological nature of (r), or the status of
parents or their origin. The sections which follow will present and
discuss the results for individual factors in the factor groups shown
above in Table 19 and Table 20.

The following results are presented separately, but they are
from one VARBRUL run with all the factor groups, both linguistic
and social. This analysis was expected to represent the
psycholinguistic impact of the social and linguistic variables and
help determine which ones positively and negatively contributed to

the application of assimilation.

84

Linguistic Factors: Word-internal Position

The linguistic factors involved in the application of the
assimilated variant are following environment, preceding sound and
location of primary stress. Not relevant to the variation was the type
of boundary, whether syllable or clitic, between (r) and the consonant
to which it assimilates.

The results for following consonantal environment and word-
internal assimilation are shown in Table 21. The following
environment most favorable to assimilation was [h] (although there
were only nine tokens in all, of which six were applications). The
voiceless velar [k] and alveolars [t] and [s] strongly favored
application, too. These three segments were grouped into one factor
based on similar weights in a previous run and their sharing of the
features [-voice] and [—labial]. The sounds [b] and [m] are
neutral with respect to this process. None of the other sounds
promotes application of assimilation. Aspiration bleeds assimilation

preceding [n] and [l].

85

Table 21

Following Environment: Word-internal Assimilation

 

 

Factor Weight Applications Percent
[h] .745 6/9 67%
[k, t, s] .637 244/456 54%
[b, m] .484 51/143 36%
[l] .356 17/65 26%
[g] .329 2/10 20%
[p] .249 4/20 17%
[d] .228 15/106 12%
[n] .161 2/23 9%
[f] k.o. 0/2 2%

 

Note. Input = .359; x2(18, N = 851) = 170.14,
[sig.]; x2 cell 1.13

86

These results suggest that in this variety voicing and place, but
not manner, are relevant for assimilation. The features [ —voice,
—labial], of [t], [s], [k] and [h], promote assimilation; the
features [+voice, +labial], found in [b] and [m], are neutral; and
the features [+voice, -labial], of [d], [n], [l] and [g], as well
as the features [—voice, +labial], of [p] and [f], disfavor
assimilation of (r). Table 22 shows these features as they relate to

the application of the assimilated variant.

Table 22

The Contribution of Voicing_and Place to Assimilation

 

Labial Coronal Dorsal

 

vl vd vl vd v1 vd

 

p 9 t d k g
f s h
at n
I

 

Note. Bold indicates that assimilation
is promoted, italics that it is
disfavored, and underlining indicates

a neutral effect.

 

When the conditioning environments for assimilation in
younger and older speakers are compared, it appears that the
diffusion of this sound change has involved its spread from a limited
number of environments in older speakers to a wider range in

younger ones, rather than from one lexical item to another (Labov,

87

1994). The change appears to have progressed in waves, originating
in a specific phonological context and spreading from there (Bailey,
1973, as cited in Preston, (1989).

As Table 23 shows, (r) assimilated to four following
consonants, [k], [l], [m], and [t] in the speech of older
respondents. Assimilation before [k] was the most frequent for older
speakers, applying in 54% of cases. As for the other three
environments, assimilation occurred only once before [1] (1/14) and
[m] (1/15), and in three cases occurred before [t] (3/23). Thus, the
voiceless velar may have been the environment in which assimilation
began, and from there it spread to additional phonological

environments.

Table 23

Percent of Word-internal Assimilation by Followiag Sound and Ag;

 

k t l m n g d b s h p f

 

Older 54 13 7 7 - - - - - - - -
Younger 67 50 31 39 15 25 17 45 45 86 17 -

 

Assimilation before most consonants is reported in earlier
works on the western regional variety (Goodgall de Pruna, 1970a;
Guitart 1976), which is the variety in which assimilation is most
advanced, and the one from which assimilation has spread to the
central region. From this evidence it can be concluded that
assimilation has not only spread to the central variety, as predicted

by Almendros (1958), but that it has advanced vigorously.

88

Although no cases of it were found in the conversational
samples used in this study, the older male speaker often assimilated
before [h], as in [hohhe] Mg; ‘George.’ This would further
suggest that the initial environment for assimilation was before
sounds with the features [-voice] [ -labial], and that assimilation
spread from there to other environments, but, as the results shown in
Table 21 above indicate, i is most strongly favored in this original

environment.

Table 24

Precedirg Sound: Word-internal Assimilation

 

 

Factor Weight Applications Percent
[o] .681 177/286 62%
[a] .495 89/239 37%
[i, e] .343 74/247 23%
[u] .361 1/5 20%

 

Note. Input = .359; 3808, a = 851) = 170.14,
[sig.]; )(2 cell 1.13

 

Table 24 shows the results for preceding sound. Assimilation is
promoted by a preceding [0]; [a] has a neutral effect on it; and [i],
[e] and [a] moderately disfavor it. Word-finally, too, a preceding
[0] promotes assimilation. Table 25 compares application of
assimilation in younger and older speakers according to preceding
vowel. The older speakers assimilated nearly half of all cases of (r)
when [0] preceded (34/73), and younger speakers assimilated in 67%

(143/214) of cases. A preceding [a] contributed to 12% assimilation

89

in older speakers (4/33) and 41% (85/206) in younger ones. After
[e] younger respondents used 29% (70/239) assimilation and older
ones used 3% (2/72). The high vowels had very low rates of
occurrence (the high back vowel had a total of five cases and the
high front one eight) and they do not favor assimilation. These
findings suggest that [c] was an initial environment for

assimilation.

Table 25

I

Word-internal Assimilation by Preceding Sound and Ag;

 

o a e u i

 

Older 47% 12% 3% - -

Younger 67% 41% 29% 33% 25%

 

The results for word-internal assimilation and stress are shown
in Table 26. A stressed syllable slightly favors assimilation and a
nonstressed syllable moderately disfavors it. Goodgall de Pruna
(1970a) noted that assimilation occurred in tonic and pretonic
syllables in her study of young respondents from Trinidad, a central
city that was classified as belonging to the western dialect region
after Goodgall reported that frequent assimilation occurred in the

speech of college-aged students from there.

90

Table 26

Word-internal Assimilationéand Stress

 

 

Factor Weight Applications Percent
Stress .554 263/566 46%
No stress .394 78/207 27%

 

Note. Input = .359; x2(18, a = 851) = 170.14,
[sig.]; 12 cell 1.13

 

The evidence from Puerto Rican Spanish is very different,
however. Poplack (1981) found that the most significant factor in
word-internal assimilation was a following primary stress. This
suggests that assimilation in Puerto Rican Spanish is a
phonologically distinct process, and it supports the claim made in
Almendros (1958) that assimilation as it is carried out in Cuban
Spanish is unique and “perhaps what most distinguishes Cuba within
the Caribbean region” (Almendros, 1958, p. 149).

The type of boundary (r) must cross to assimilate to‘a following
consonant was not found significant in the stepwise multiple
regression. Table 27 shows the VARBRUL results for this factor
group. The probability weights are very similar, which suggests the
individual factors are not significant; furthermore, the weights are
clustered close to .50, which indicates that the factor group does not
have a strong positive or negative effect on the application of
assimilation. These results indicate that the boundary created by the
juxtaposition of the infinitive (r) and the attachment of a clitic is

treated like the word-internal one, which could indicate that Spanish,

91

like French, has vigorously moved toward affixation. Again, the
differences between the Cuban and Puerto Rican varieties are
revealed: Poplack (1981) found that infinitive + clitic combinations
favored assimilation more than the word-internal environment, which
led her to propose that assimilation was morphologically conditioned.
However, the results of the current study indicate that morphological
constraints are not relevant for assimilation in this variety of Cuban

Spanish.

Table 27

Word-internal Assimilation: Syfitble vs. Clitic Boundafl

 

 

Factor Weight Applications Percent
Syllable .504 320/783 41%
Clitic .453 21/68 31%

 

Note. Factor group not significant

 

Linguiistic Factors: Word-final Position:

The most significant finding from the analysis of word-final
assimilation is that it is constrained by social factors, specifically
gender and age. This was not the case for word-internal assimilation,
whose application was constrained by both linguistic and social
factors.

No linguistic factors were found significant in the stepwise
multiple regression for word-final assimilation. Table 28 shows the
results for following phonological environment. Only a following [1]
or [h] favor assimilation. The lateral did not favor assimilation

word-internally since aspiration bleeds assimilation in this

92

Table 28

Word-final Assimilation fld FollowingEnvironment

 

Factor Weight Applications Percent

 

[1] .732 40/92 43%
[h] .693 1/3 33%
[t] .526 7/27 26%
[m] .505 8/24 33%
[k] .458 11/52 21%
[b] .401 2/6 33%
[y] .378 1/6 17%
[d] .374 10/51 20%
[s] .366 6/31 19%
[p] .251 4/28 14%
[f] .156 1/6 17%
[n] k.o. 0/17 -

[g] k.o. 0/4 -

 

Note. Factor group not significant

 

93

environment. Word-finally aspiration does not apply, which results in
the lateral promoting assimilation.

Assimilation of /r/ to a following lateral dates from around the
ninth century. Thus, it is not surprising that the lateral promotes
assimilation word-finally, whereas other sounds do not. Assimilation
before the lateral is common in many varieties of Spanish, including
those of Andalusia, Canary Islands, the Dominican Republic, Puerto
Rico, Colombia, Argentina, Chile and Panama (Alonso, 1961; Catalan
1989). Poplack (1981) found that 85% of cases of (r) before a lateral
either assimilated or were deleted.

Earlier studies on the western regional variety of Cuba reported
high rates of assimilation word-finally before a lateral (Goodgall de
Pruna, 1970a; Haden and Matluck, 1977). Poplack (1981) found that
in Puerto Rican Spanish the rate of assimilation before the lateral in
both word-internal and word-final environments was close to 90%,
which is much higher than the rate found here of 43%.

The voiceless velar fricative had only three cases word-finally;
thus, these results are inconclusive. A following [t], which
promoted assimilation word-internally, and a following [m], have a
neutral effect; [k], which strongly promoted assimilation word-
internally, slightly disfavors it here; and [b], [y], [d] and [s]
([s] also favored assimilation word-internally) moderately disfavor
it; the voiceless labials [p] and [f] strongly disfavor its
application, and [g] and [n] were negative knockouts, with no cases
applying when they followed (r).

The results for assimilation according to preceding sound are

shown in Table 29. As with word-internal assimilation, a preceding

94

[o] promotes assimilation. The other vowels have either a slight or

neutral effect on the application of assimilation.

Table 29

Preceding Sound Ed Word-final Assimilation

 

 

Factor Weight Applications Percent
[e] .445 24/94 26%
[a] .437 27/121 22%
[i] .534 9/31 29%
[o] .644 31/80 39%
[u] k.o. 0/2 -

 

Note. Factor group not significant

 

The results for stress, shown in Table 30, show that the
individual factors are not significant, because they cluster around
.50, and that the factor group has no effect on application of the
assimilated variant. Stress was not significant for word-final

assimilation in Puerto Rican Spanish either (Poplack, 1981).

Table 30

Word-final Assimilation: Stress

 

 

Factor Weight Applications Percent
Stress .503 77/282 27%
No stress .483 14/44 32%

 

Note. Factor group not significant

 

The results shown in Table 31 indicate that the morphemic

status of (r) does not contribute to its variation. Cedergren (1973)

95

and Poplack (1981) found that infinitives promoted nonstandard
variants of (r), while monomorphemes favored the basic variant. This
is contrary to what has been found for English t/d deletion in which
nonstandard variants are disfavored in grammatical morphemes (e.g.

mist vs. missed).

Table 31

Word-final Assimilation: Morpheme Status

 

 

Factor Weight Applications Percent
Infinitive .485 53/213 25%
Monomorpheme .528 38/113 34%

 

Note. Factor group not significant

 

Social Factompd (r) Assimilation

This section discusses the effects of the social variables age,
gender, status and origin, on the application of assimilation in word-
internal and word-final environments. Word-internally, age, gender
and status interacted; the factor group represented by these variables
was selected as significant. Word-finally, gender and age were
significant, but not status; the factor groups did not interact as they
did for the word-internal position. Origin was not significant in
either environment.

In word-internal position, age, gender and status interacted, as
shown in Figure 8. Assimilation tends to be most frequent in young
lower status males (.798), followed by young lower status females

(.675),then

96

 

FYU FYM FYL FO MYU MYM MYL MO

Figure 8. Social variables and word-internal application of (r)
assimilation. Input = .359; x2(18, N = 851) = 170.14, [sig.]; x2 cell
1.13. Female young upper (11 = 27/78), Female young middle (11 =
63/192), Female young lower (11 = 41/75), Female old (11 = 25/109),
Male young upper (n = 127/241), Male young middle (11 = 34/70),
Male young lower (11 = 9/12), Male old (11 = 15/74).

97

young upper status males (.664) and young middle status males
(.649). Young females of upper (.464) or middle status (.418) have
small disfavoring tendencies. Older speakers strongly disfavor
assimilation, but the older male makes a smaller negative
contribution than the older female.

These results provide strong evidence for generational change
in this speech community in the use of the assimilated variant of (r).
As noted earlier, evidence for change can also be found when
following environments are examined according to age. Younger
respondents assimilated in a wide range of consonantal environments.
Older speakers, on the other hand, assimilated in four environments,
and one of these stood out as more advanced than the others,
assimilation occurred in over half the cases of the following
voiceless velar stop. This may suggest that assimilation may have
started in the environment preceding [k] and spread to others from
there. This claim, however, must be investigated further with a larger
amount of data from older speakers.

Evidence of assimilation in the older speakers indicates that
this is not a new change attributable to sociopolitical changes, but
one which was in progress long before the revolution in the late
19508

Assimilation was originally an urban norm. It was widespread
in Havana in all social status groups by the 1950’s (Almendros,
1958); Guitart (1976) noted that its application was nearly
categorical in upper middle class Havana speakers. Thus, its eastward
spread on the island was most likely motivated by the prestige of

Havana as the island’s largest urban center and capital.

98

As an urban feature, assimilation most likely spread from one
urban center to another rather than smoothly progressing throughout
an entire geographic region at a time. The Northern Cities Vowel
Shift has been described as progressing in this manner since it
spreads from one city to the next without covering rural areas in
between (Callery, 1975). The changes radiate out from urban centers
to the surrounding rural areas in a more typical wave-like fashion.

This model explains why assimilation was observed early on in
large central city of Santa Clara (Almendros, 1958). It also explains
the advanced stage of assimilation in Cien Fuegos, a city in the
southern part of the central region, which motivated Choy Lopez
(1986) to classify it in the western dialect area. It also accounts for
the findings of the current study since the respondents are from
towns within a short distance of Santa Clara, the capital of the
central region. It does not, however, fully account for why Goodgall
de Pruna (1970a) unexpectedly found high rates of assimilation in
young speakers from Trinidad, a town in the central region which is
not near a major city. It could also be that the change was at an
advanced state by 1970, but had gone largely unnoticed. An apparent
time study representing more generations than the two used here must
be carried out in order to answer this question.

The above results suggest that males led this sound change.
This is unusual since changes are usually led by females, who tend to
be attracted to the new forms (Labov, 1990). There are few cases of
changes that are led by men, but this is a “minor tendency,”

according to Labov (1990), who notes that a few cases of this include

99

 

 

 

his own study on Martha’s Vineyard (Labov, 1963), and Milroy and

Milroy (1985) in Belfast.

The reason why assimilation was led by males can be traced to
the appeal of the urban center with which it was associated and
unequal access to Havana. Previously, Havana was more accessible to
males from the provinces than to females. For instance, it was
common for young males to go there to attend private schools, the
university or simply to work. Females did not have the same access
to Havana because widespread machismo dictated a traditional role
for them in the family (Smith & Padula, 1990).

Assimilation, as the most distinguishing feature of Havana
speech, may have appealed to males because it represented not only
progressive, cosmopolitan values, but, more importantly, masculine
access to the capital. A perceptual dialectology study carried out by
Preston (1985) in Brazil found that males highly evaluated Rio de
Janeiro based on the positive associations it had as a place where
males, but not females, went to enjoy themselves. Males may find the
speech norms of large urban centers appealing because they associate
them with the social liberty their masculinity is accorded.

As for the current social distribution of assimilation, the
results shown in Figure 8 indicate that young males use more
assimilation than females, with the exception of the young lower
status female. These findings show a typical sociolinguistic pattern:
“in stable sociolinguistic stratification, men use a higher frequency
of nonstandard forms than women” (Labov, 1990, p. 205). The
behavior of young lower status females is an exception to the pattern,

but the young lower status males also do more assimilation than their

100

middle status counterparts. Thus, this is a reflection of more frequent
assimilation in lower status groups.

The findings for (r) in word-final position reveal that
assimilation in this position is not constrained by any linguistic
factors. Age and gender are the only significant factors. The results
for gender are shown in Table 32. These are similar to the word-
internal results in that males strongly favor assimilation and females
strongly disfavor it. Gender makes a strong contribution to
assimilation, as indicated by the spread between the probability
weights for females and males. It was pointed out earlier that this is

a typical pattern for nonstandard variants, which tend to be preferred
by males.
Table 32

Word-final Assimilation: Gender

Weight Applications Percent

 

Factor
Female .260 23/151 15%
Male .711 68/175 39%

 

Note. Input = .264; x2(2, N = 1478) = 5.35,

[n.s.]; x2 cell 1.33

The results for age, shown in Table 33, are also similar to the
results for word-internal occurrences and provide evidence for
generational change. Assimilation is preferred by younger speakers

and strongly disfavored by older ones. This change is advanced in the

Speech of young adults in the central region.

101

Table 33
Age and Word-final Assimilation

 

 

 

Factor Weight Applications Percent
Young .608 80/240 33%
Old .227 11/86 13%

 

Note. Input = .264; x2(2, N = 1478) = 5.35,

[n.s.]; 12 cell 1.33

The results for status are shown in Table 34. Although this
factor group does not contribute to the variation, the results suggest
that it is slightly disfavored by the two middle status groups and
moderately favored by the lowest one. Guitart (1972) noted that in
Havana it was not the frequency of assimilation which reflected
status differences, but rather by production or avoidance of
assimilation in certain phonological environments, and the outcome
of assimilation preceding certain sounds. For instance, complete
gemination correlated with lower status and partial gemination with

higher status. This claim must be investigated further with more data.

Table 34

Word-final Assimilation: Status

 

Weight Applications Percent

 

Factor

Midde .428 26/95 27%
Upper .452 51/161 32%
Low .697 14/70 20%

 

Note. Factor group not significant

 

102

Conclusion

Assimilation in word-internal position is constrained by both
linguistic and social factors. In word-final position, however, only
the social factors age and gender contribute to its application. Word-
internal assimilation is constrained by the voicing and place features,
and the features [ —voice, - labial] promote its application.
Generational differences are evident in the following environments
before which assimilation occurs. Stress had a small effect, with
nonstressed syllables disfavoring assimilation.

Evidence for generational change was found. It was proposed
that assimilation was spreading out from the western region long
before the revolution, and that it was not motivated by sociopolitical
changes. The urban character of the feature was discussed,
particularly as it applies to the manner in which it has spread from
one region to another and how this made the feature appealing to
males. A common sociolinguistic pattern for nonstandard variants

was revealed in the more frequent use of assimilation by young

males.

Lateralization
This section gives a brief background for and presents and
discusses the results of the variables rule analysis for lateralization

in word-internal and word-final environments.

Background

Lateralization of /r/ has existed in Spanish dialects for
hundreds of years. According to Boyd-Bowman (1975), by the Ox"

sixteenth century it was widespread in Andalusia. Today, it is

103

common throughout the Caribbean and coastal areas of the Pacific. It
is stigmatized in many regions, and even where it is not, its
frequency tends to correlate with social status, increasing as social
status decreases. As its variation is mostly social, it does not exhibit
significant regional variation in Cuba.

Earlier studies on Cuban Spanish have reported low rates of
lateralization. Fails (1985), using data collected during 1968-69 from
middle class speakers, found less than 2% lateralization word-
internally and less than 2% word-finally before a consonant, less
than 6% before a pause, and no occurrences before a vowel. Terrell
(1976) reported 7% word-finally and 4% word-internally for middle
status speakers. Vallejo-Claros (1971) examined the lateral variant in
the central dialect in three status groups and reported zero
lateralization for the highest status group, 5% for the middle, and 5%
for the lowest.

Results and Discussion

The factor groups selected as significant in the stepwise
multiple regression analysis carried out on word-internal and word-
final lateralization are shown in Table 35 and Table 36. Word-
internally, the significant factor groups were following environment,
preceding sound, age, gender and origin. Stress and social status

were not significant; and boundary type (syllable vs. clitic) was

eliminated from the analysis because the lateral did not apply before

clitics.

104

 

Table 35

Word-internal Lateralization: Significant Factor Groups

Significant Not significant

Following environment Stress

Syllable vs. Clitic
Preceding sound

Boundarya
Age Status
Gender
Origin

 

'Lateralization did not apply before a clitic boundary

Age, gender and status did not interact for the lateral variant as
they did for the assimilated one, in which interactions were found
between gender, age, and status. Origin is significant for
lateralization, but it was not significant for assimilation, most likely
because the lateral variant has historically been a feature of rural
speech. The social motivations for speakers who have experienced
upward social mobility to use the lateral variant will be discussed
below.

Word-finally, the significant linguistic factor groups were
following environment and stress; preceding sound and morpheme
type were eliminated. There was interaction among the factor groups
gender, age and status. Origin was significant in this position, as it
was for the word-internal one. Status is significant word-finally, but
it did not contribute to word-internal variation. This is perhaps

because the word-final environment is more salient, and the status of

105

the lateral as a negative social stereotype may lead certain speakers

to avoid it in this more salient environment.

Table 36

Significant Factor Groups for Word-final Lateralization

 

Significant Not significant
Following environment Preceding sound
Stress Infinitive vs. monomorpheme

Gender, Age, Status

Origin

These results suggest that (r) as an inflectional morpheme has
no bearing on the occurrence of the lateral variant. These same
results were noted above for assimilation. Poplack (1981) found that
in the Puerto Rican variety she was studying infinitives and
monomorphemes behaved similarly with regard to lateralization,
although in general infinitive (r) favored more nonstandard variants.
Similar results were reported in a study of the Spanish of Panama
City by Cedergren (1973), which found that infinitives favored
nonstandard variants and monomorphemes the standard variant.

In the following sections the results for the individual factors
of the factor groups shown in Table 35 and Table 36 will be
discussed in greater detail.

FollowLng Environment

The results for the lateral variant word-internally are shown in

Table 37. Coda consonants are followed by another consonant when

they occur within a word, but a pause may follow in cases in which

106

Table 37

Word-internal Lateralization: Following Environment

 

 

Factor Weight Applications Percent

Pause .890 41/141 29%
[p] .829 13/52 25%
[g] .792 4/14 29%
[d] .720 30/172 17%
[f] .571 1/8 13%
[k] .515 17/304 6%
[n] .459 2/40 5%
[s] .288 3/84 4%
[t] .218 4/158 3%
[m] .083 1/122 1%
[l] k.o. 0/118 -
[ch] k.o. 0/1 -
[y] k.o. 0/6 -
[b] k.o. 0/42 -
[h] k.o. 0/12 -

 

Note. Input = .036; x205, a = 1095) = 123.13,
[sig.]; x2 cell 0.75

107

pronunciation of the word was halted after the first syllable, as in
Abri lapuer- (Abri la puerta ‘I opened the door’). A following pause
has a very strong positive effect (.890) on lateralization. A following
[p] (.829) or [g] (.792) also strongly favor the lateral. When (r) is
followed by [d], lateralization is more moderately favored (.720).
The voiceless coronals [s] (.288) and [t] (.218) disfavor the lateral;
and it is categorically disfavored, or nearly so, by the voiced labials
[m] (.083) and [b] (k.o.).

The sounds that favor the lateral are the ones that disfavor
assimilation word-internally. The sounds [p, f], [d, n, l] and [g]
disfavored assimilation, whereas lateralization is promoted by [p, f],
[d] and [g]. The sounds [t, s] and [k, h] promoted application of
assimilation, but [s, t] strongly disfavor lateralization, [k] has a
very slight positive effect, and it never applied when [h] followed.

In word-final position, the lateral variant occurs before a
consonant, vowel or pause. Table 38 shows the results for application
of the lateral variant when (r) occurs word-finally. The strongest
contribution to lateralization is made by a following pause (.806),
which was also the case word-internally. Previous studies on Cuban
Spanish have also found lateralization to apply more before pauses
(Fails, 1985; Hammond, 1977; Terrell, 1976), although Garcia
Gonzalez (1980) claims that its occurrence is the same whether
before a consonant or a pause. Garcia Gonzalez also claims that both

vowels and pauses equally favor the lateral.

108

Table 38

Word-final Lateralization: FollowinLEnvironment

 

 

Factor Weight Applications Percent
Pause .806 41/141 29%
[i] .738 5/23 22%
[g] .639 1/4 25%
[n] .565 2/17 12%
[f] .555 1/6 17%
[p] .552 4/28 14%
[o] .512 2/11 18%
[s] .505 2/20 10%
[d] .432 4/51 8%
[a] .412 5/72 7%
[k] .394 4/52 8%
[t] .323 1/19 5%
[u] .296 1/23 4%
[e] .114 1/70 1%
[l] k.o. 0/53 -
[ch] k.o. 0/1 -
[y] k.o. 0/6 -
[b] k.o. 0/6 -
[h] k.o. 0/3 -
[m] k.o. 0/17 -

 

Note. Input = .051; x2(22, a = 537) = 94.07,
[sig.]; x2 cell 0.61

109

The results of the current study, however, indicate that not all
vowels have the same effect on lateralization. The high front vowel
[i] is the only vowel that makes a strong positive contribution
(.738). As for the other vowels, [0] has a very minimal positive
effect (.512); [a] somewhat disfavors application (.412); [11] has a
strong negative effect (.296); and [e] makes an even stronger
negative contribution (.114).

The consonant [g] is the only one that moderately favors
lateralization (.639), and it also favored it word-internally. [k]
(.394) and [t] (.323) moderately disfavor it. [s] has a neutral
effect, and [d] a small negative one. The lateral variant did not
apply before [ch], [y], [b], [h] and [m]. It never occurred before
[1] because all cases of the lateralization before [1] were classified
as assimilation.

Preceding Sound

The phonological environment preceding (r) was significant in
the application of the lateral when it occurred word-internally, but
not when it was word-final.

The probability weights for preceding sound and lateralization
are shown in Table 39. For word-internal cases, the high back vowel
was eliminated from the analysis because it had only five total
tokens, two applications and three nonapplications, and its weight
was .978. The lateral is favored by [e] and [a], but disfavored by
[i] and also by [c], which favored assimilation in both word-
internal and word-final environments. Word-finally, this factor group

was not significant. With the exception of [i], which has a

110

 

moderately disfavoring effect, and [u], which was a positive

knockout, the other vowels do not contribute to the variation.

111

Table 39

Lateralization: Precedifl Sound

 

 

 

Factor Weight Applications Percent

[e] .664 54/399 14%

[a] .574 47/314 15%
Word-internala

[i] .343 5/38 13%

[o] .271 10/344 3%

[e] .517 24/185 13%

[a] .507 32/222 14%
Word-finalb

[i] .384 8/43 16%

[o] .510 10/165 6%

 

aInput = .036; x2(15,y_ =1095)=123.13,[sig.];x2 cell 0.75;

bFactor group not significant

 

112

Sum

Stress was significant for the application of the lateral word-
finally, but not word-internally, as shown in Table 40. For word-final
cases, the lateral is strongly disfavored when (r) is in an unstressed
syllable. Poplack (1981) found that stress was not significant for

lateralization in either word-internal or word-final positions.

Table 40

Qteralization and Stress

 

 

Group Factor Weight Applications Percent
Stress .516 98/812 12%
Internal”
No stress .454 18/283 6%
Stress .545 72/476 15%
Finalb
No stress .194 2/61 3%

 

‘Factor group not significant

bInput = .051; x2(22,_1\1 = 537) = 94.07, [sig.]; x2 cell 0.61

Morpheme Type

The type of morpheme in which (r) appears was not significant
in the stepwise multiple regression; these results are shown in Table
41. This factor group was not significant for the assimilated variant
either. As found by Poplack (1981) and Cedergren (1973), (r) as an
infinitive morpheme tends to favor the nonstandard variant, and as

part of a monomorpheme it disfavors nonstandard variants.

113

Table 41

Word-final Lateralization: Morpheme Status

 

 

Factor Weight Applications Percent
Infinitive .543 53/386 14%
Monomorpheme .430 23/239 10%

 

Note. Factor group not significant

 

Social fictors

Social factors make strong contributions to the application of

f

the lateral variant in both word-internal and word-final positions. As
noted earlier, in Cuba the lateral was historically a stigmatized
variant associated with working class and rural speech. The results
obtained here suggest that it has been involved in change from below.

In word-internal position, the social variables did not interact
as they did for the basic and assimilated variants. The significant
factor groups were origin, age and gender, but not status.

Origin was significant for the lateral variant in both positions.
This is the only variant for which origin is significant, which is not
surprising since it has historically been a negative linguistic
stereotype of working class and rural speech. Indeed, the findings in
Table 42 indicate that lateralization is favored by rural origin. It was
noted earlier in the discussion of (s) deletion, which was also
previously associated with rural and working class speech, that upper
status young males of rural upper middle class parents used deletion
to associate with rural features, rather than dissociate from them, as
would be expected. It was proposed that this is because having rural

ties has positive effects on an individual’s quality of life, and so

114

rural origins are prestigious within the local culture, and rural
sociolinguistic variants as an expression of that prestige, become
prestigious as well.

Furthermore, it is noted throughout this dissertation that the
word-final environment is more salient than the word-internal one,
and that nonstandard forms are more easily avoided in the word-final
environment than the word-internal one. These findings for origin,
however, show the opposite effect: individuals with rural origins
favor the lateral more strongly in the more salient environment (.601
WP vs .565 WI). At the same time, individuals of urban origin
disfavor the lateral more strongly in the more salient word-final
environment than they do in the word-internal one (.369 WP vs. .424
WI). In other words, persons who do not have rural ties do not

express an association with them.

Table 42

Lateralizfiatiorflnd Origin

 

 

 

Group Factor Weight Applications Percent
Urban .424 49/502 10%
Word-internala
Rural .565 67/593 11%
Urban .369 30/233 13%
Word-final”
Rural .601 44/304 14%
1‘Input: .036; 7805,11 =1095)=123.13,[sig.];x2 cell 0.75
bInput = .051; x2(22, a = 537) = 94.07, [sig.]; x2 cell 0.61

 

Gender and age were significant word-internally. These

findings are shown in Table 43 along with status, which was not

115

 

significant. The results indicate that lateralization is favored by
males and disfavored by females, a pattern commonly found for
nonstandard variants (Labov, 1990). The weights for age indicate that
it makes a stronger contribution than gender.

Evidence of generational change is revealed by the results for
age. Older speakers strongly disprefer lateralization (.109), whereas
younger ones favor it (.663). The results for word-final position,
shown in Figure 9, also reflect strong generational differences. The

older speakers strongly disfavor lateralization in this position, and

i

the younger ones have positive probability weights (with the
exception of upper status females).

It was not expected that the lateral would show such strong
generational differences, particularly since these respondents were
college educated adults in their late twenties and early to mid-
thirties, who would be expected to have low rates of stigmatized
items in their speech, as reported in earlier studies of middle class
adults (Fails, 1985; Terrell, 1976; Vallejo-Claros, 1971). In Garcia
Gonzalez (1980), however, it was noted that the lateral was spreading
to the educated, or standard, norm. In this study, the younger age
group had 8% word-internal and 17% word-final lateralization, while
the older one had only 1% in both positions.

The results for social status in Table 43 suggest that, although
social status was not significant, respondents from upper middle
status families disfavor the lateral but those from lower and middle
status ones slightly prefer it, which is not surprising since these
groups tend to lead linguistic change from below, or “in a direction

away from the standard norm” (Trudgill, 1972, p. 191).

116

 

 

. a. .. 390.131.45.31;

Table 43

Word-internal Lateralization: Gender and Aga

 

 

 

 

Group Factor Weight Applications Percent
Female .369 44/593 7%
Gender
Male .653 72/502 14%
Young .663 112/829 14%
Age
Old .109 4/266 2%
Middle .528 44/329 13%
Status“ Upper .446 53/479 11%
Lower .558 19/287 7%

 

Note. Input = .036; 3805, N =1095):123.13,[s1[t;.]-,x2 cell 0.75

*Factor group not significant

 

Gender, age and status interacted in word-final position. These
results are shown in Figure 9. The lateral variant showed nearly
categorical application for young lower status males (.946), though
this finding is based on a very limited number of tokens). It was also
strongly favored by young lower status females (.799), young upper
status males (.799), young middle status males (.774) and young
middle status females (.643). On the other hand, it was found to be
strongly disfavored by young upper status females (.231), older
males (.191) and older females (.066).

117

..w .r

111-..}? :

 

FYU FYM FYL FO MYU MYM MYL MO

Figure 9. Social variables and word—final application of (r)
lateralization. Input = .051; x2(22, N = 537) = 94.07, [sig.]; x2 cell
0.61. Female young upper (11 = 2/38), Female young middle (n =
15/103), Female young lower (11 = 13/49), Female old (11 = 1/95),
Male young upper (11 = 21/114), Male young middle (11 = 12/53), Male
young lower (11 = 9/16), Male old (n = 1/69).

118

. nrlvdztt.:.v§ ..

The fact that the lateral is favored by young males and avoided
by young upper status females can be explained with the notion of
covert prestige, proposed by Trudgill (1972). This notion is meant to
explain the use of nonstandard forms by males, but not by females.
Trudgill proposed that females tend to avoid nonstandard variants
because they are more status-conscious, whereas males are attracted
to nonstandard variants because these are associated with working

class speech and culture, which Trudgill explains,

appears, at least in some western societies, to have
connotations of masculinity probably because it is
associated with the roughness and toughness supposedly
characteristic of WC life which are, to a certain extent,
considered to be desirable masculine attributes. They are not,
on the other hand, considered to be desirable feminine
characteristics. On the contrary, features such as refinement
and sophistication are much preferred. (p. 183)

Trudgill notes, however, that not all females are interested in
projecting refinement and sophistication, younger females, in
particular, show a tendency to “attach covert prestige to WC speech
forms” (p. 192), making nonstandard forms more frequent in these
groups.

The explanatory value of covert prestige with an emphasis on
masculinity vs. femininity is not complete, however. It does not, for
instance, explain why, holding age constant, the upper status females
use less of the nonstandard lateral, while the middle and lower status
ones use more.

As noted above, covert prestige plays an important role in
linguistic change from below. To explain the factors motivating
change in the use of the lateral variant, the notion of covert prestige

must be extended beyond Trudgill’s initial argument, based on the

119

.514 l, . I... 192111.09;

 

 

desire to express the masculine characteristics associated with
working class speech, to include a larger societal desire to associate
with working class speech as a symbol of new nationalism. solidarity
and support for the revolution.

Cuban youth were exposed to a new sociolinguistic norm which
arose as a result of social promotion and the immediate needs of the
revolution. In the early years of the revolution there was a severe
shortage of qualified individuals because political credentials were
more important than professional qualifications. Fitzgerald (1990)
states that “by the end of the 1960s many intermediate posts in Cuba
were occupied by individuals with deficient educational backgrounds
and skills” (p. 192), and that Fidel Castro himself declared in 1970
that “signs of illiteracy and semi-illiteracy can be found in many men
in positions of responsibility” (1990, p. 192).

Along with the strict demand for loyalty to the revolution, the
shortage of qualified individuals was aggravated by the mass exodus
of middle class individuals, which “created a void in the labor force
and an urgent need for administrators, technicians, agronomists,
engineers and other trained personnel” (Bunck, 1989, p. 443). The
centralization of the economy also created a larger need for qualified
administrators and bureaucrats. The demand for teachers with the
appropriate political credentials was imperative since teachers played
a vital role in the socializfion (Baloyra, 1989) of Cuban youth.
Many unqualified personnel were placed in politically sensitive
positions.

The new sociolinguistic norm was based on the norms of groups

that had been institutionally marginalized before the revolution, in

120

H ..- If. 1. slim-gt. .

the sense that they had had less access to economic and social
resources. According to Amaro (1989), these groups were the
strongest supporters of the revolution: “it is certain that the marginal
classes exhibited greater support of the revolution...such support
seems natural since the revolution emphasized marginal individuals”
(p. 57). These groups experienced rapid upward social mobility
because of their positive attitudes toward the revolution. The
sociolinguistic outcome of this situation was to associate nonstandard
features with revolutionary loyalty.

Kontra (1992) explains that the incursion of nonstandard
grammatical features into standard Hungarian occurred in communist
Hungary because many individuals from the working class
experienced rapid upward social mobility, which contributed to the

elimination of previous linguistic markers of social status:

The power structure of society in communist Hungary made it
possible for members of the communist political and economic
élite to retain their vernacular non-standard speech When
they came to power, they may have changed some of their
habits or their clothing, but there was no linguistic threshold to
cross. After all, they represented the people in a pe0ple's

democracy. (p. 219)

When nonstandard linguistic features take on positive value in
the wider society, individuals who experience upward social mobility
do not have to abandon them and adopt standard, or at least
nonstigmatized, forms, which is usually the case for certain
occupations and professions, particularly those that deal with the
public, as suggested by the notion of the linguistic market (Bourdieu
& Boltanski, 1975, as cited in Sankoff & Laberge, 1978). This

motivates the leveling of former linguistic markers of status.

121

 

 

. .o. .. 1a ‘lo-

. 19.1%...

Moreover, the younger generation in these situations is exposed to a
new sociolinguistic norm.

In Cuba, there was strong pressure for young Cubans to
conform to the new norm. Cuban youth had to be integrated if they
were to be successful in their studies and their work, as pointed out

by Baloyra (1989). Baloyra further states:

Cuban youth bear the brunt of a sustained and systematic effort
at socialization in revolutionary values, implemented through a
series of carefully programmed and coordinated activities. The
breadth and intensity of the demands that these activities place
on Cuban youth create very powerful incentives for conformity
and integration [italics added] generic to all totalitarian
societies. (pp. 429-30)

The societal demand for integration and commitment to the
revolution exerted strong pressure to conform to the new
sociolinguistic norm. One of my young female respondents, who told
me she had feigned integration?20 because she wanted to attend the
university, told me that when she was in high school, she was often
criticized by her female peers for speaking like a burguésa
(bourgeois). She noted that she was especially criticized for
“pronouncing the last letters of words.”

The positive correlation between the strength of revolutionary
commitment and the potential for acquiring higher social status is a
factor which has contributed to the formation of identity within post-
revolutionary Cuba (Amaro, 1989). For instance, a university

education can only be obtained by integrated individuals. Admission

 

2oBaloyra (1989) classifies the behavior of Cuban youth into three
groups. One is made up of the true believers; the second group is
made up of the indifferent. He calls the third group the socialistas,
“or dissimulators, who feign allegiance, commitment and compliance
but are merely playing a game of survival.” According to the self-
report of this respondent, she belonged to the third group, the
dissimulators.

122

procedures require a personal interview and letters describing ones
revolutionary commitment. Careers in the social sciences, law,
economics, psychology and the humanities are open only to students
with unquestionable loyalty to the revolution. Careers in the sciences
and technology require integration, but are politically less
demanding.

In conclusion, the lateral variant has been involved in change
from below (Labov, 1966/1982). Sociopolitical changes have created
conditions that have motivated and strengthened its diffusion from
lower status groups to higher ones. On the one hand, the working
class characteristics and values associated with the lateral appeal to
college-educated individuals whose lifestyles and concerns are
essentially working class (please refer to the summary discussion of
the variable (s) for a discussion of this point). On the other hand,
rapid upward mobility and strong pressure to conform to the new
sociolinguistic norm have strengthened the change.

Conclusion

The pattern revealed for the lateral variant of (r) is different
from the one described earlier for assimilation, and that which will
be described below for aspirated and basic variants aspiration of (r).
Perhaps this because the lateral is a very old variant associated with
lower status and rural origin, and, thus, previously subjected to
negative social evaluation. Unlike the assimilated and basic variants,
the lateral is constrained both word-internally and word-finally by
social and linguistic constraints.

The analysis of linguistic constraints revealed that a following

pause strongly promotes the lateral, as does a following voiced velar

123

 

t p l. .s. . ..s . n: uhr.£’gl

stop. The effect of preceding segment was significant word-
internally, with the front mid vowel favoring and the back rounded
mid vowel disfavoring its application. Stress was significant word-
finally only, and it was revealed that a nonstressed syllable has a
strong negative effect. Clitic boundary and morpheme type were not
significant for this variant or for any other one of the variants of (r).
The analysis of the social variables showed strong evidence of
generational change. It was proposed that this is a change from below
motivated by sociopolitical changes. These changes created and gave
strength to a new sociolinguistic norm which was later associated

with revolutionary commitment.

Aspiration
The results for the aspirated variant are discussed below after a
brief background summary. A more complete description of its
history, regional and social distribution can be found in the
introduction.
Background
Aspiration preceding the lateral is found in Cuban Spanish, and
in other Caribbean dialects, including those of Puerto Rico and
Panama, but not the Dominican Republic, as well as in Andalusia,
and the Canary Islands (Alonso, 1961; Catalan, 1989). Choy Lopez
(1986) noted that aspiration was common in the western, central and
east-central varieties. However, studies have reported low rates of it:
in samples from Havana speakers, Fails (1985) found only 1% and

Terrell (1976) reported 3%.

124

Resultsgand Discussion

The variable rule analysis of the aspirated variant did not
select any factor group as significant. This was most likely because
there were only 51 (6%) tokens out of 853 cases of word-internal (r),
not a sufficient number to successfully build a statistical model of its
variation. These results for word-internal aspiration are somewhat
higher than those reported for Cuban Spanish in Terrell (1976) and
Fails (1985). Other varieties of Caribbean Spanish have much higher
rates of aspiration: for Puerto Rican Spanish 18% aspiration was
reported (Poplack, 1981); for Panama it was found to be 2% before
obstruents and nasals, but 23% before the lateral (Cedergren, 1973).

The low rate of occurrence of this variant can be attributed to
the fact that the only following phonological environments that favor
its application are /l/ and /n/: here it occurs in 55% (36/65) of cases
before /1/ and 61% (14/23) before /n/. These results are shown in
Table 44, which shows the rates of application according to

following environment.

125

Table 44

Following Environment and Word-internal Aspiration

 

 

 

 

Following Application Non-application Total
[k] 0 (0%) 252 (100%) 252 (30%)
[l] 36 (55%) 29 (45%) 65 (8%)
[n] 14 (61%) 9 (39%) 23 (3%)
[m] 1(1%) 106 (99%) 107 (13%)
[t] 0(0%) 139 (100%) 139 (16%)
[d] 0 (0%) 121(100%) 121(14%)
[b] O (0%) 36 (100%) 36 (4%)
[s] 0 (0%) 65 (100%) 65 (8%)
[h] 0 (0%) 9 (100%) 9 (1%)
[g] 0(0%) 10(100%) 10(1%)
[f] 0 (0%) 2 (100%) 2 (0%)
[p] 0 (0%) 24 (100%) 24 (3%)
Total 51 (6%) 802 (94%) 853

 

126

 

 

a: ...t... Ir. 3.9-1

Aspiration before /1/ and /n/ is favored in many varieties,
including those of southern Spain, the Canary Islands, the Caribbean
coast of Colombia and Venezuela, Panama, Uruguay, Ecuador,
Dominican Republic, and Puerto Rico (Alonso, 1961; Catalan, 1989;
Cedergren, 1973; Poplack, 1981).

Aspiration before [1] and [n] has long been recognized as a
feature of Cuban Spanish (Isabescu, 1968; Lamb, 1969; Lopez
Morales, 1976; Salines, 1957; Terrell, 1976). Choy L6pez (1986)
claims that although most regional varieties of Cuba aspirate (r)
before these consonants, aspiration is not found in all regional
varieties of Cuban Spanish. The central region, the region under
consideration here, is one of the regions for which aspiration is
reported, and as these results show, it occurs in over half of the cases
of (r) before /n/ and /l/.

There was one occurrence of [h] before [In] in the item
dematitis ‘dermatitis,’ which was previously reported for Cuban
Spanish by Garcia-Gonzalez (1980). However, there are no cases of
aspiration before the clitic ma.

The VARBRUL results for following segment did not reveal
any differences between /1/ and /n/ as conditioning environments for
aspiration. The probability weight for /l/ was .497 and for /n/ it was
.501, indicating that neither contributes more than the other to the
variation (Input = .609; 7802, a = 84) = 48.03, [sig.]; x2 cell 1.23).

Although in Cuban Spanish both /1/ and /n/ favor aspiration
equally, Cedergren (1973) reports that aspiration occurred more
before the lateral (23%) than the nasal (3%) in the Spanish of Panama

City. Poplack (1981) also reports the nasal favoring aspiration (29%)

127

more than the lateral (3%) in Puerto Rican Spanish. In these
varieties, aspiration is not conditioned solely by these two segments,
but may occur in many other environments.

The preceding phonological environment for aspiration is
always a vowel. The VARBRUL results for preceding segment shown
Table 45 suggest that the only vowel that favors aspiration is [a],
but its probability weight of .592 does not indicate a strong
contribution to variation. Aspiration is moderately disfavored by [e]

(.399) and strongly disfavored by [i] (.188).

Table 45

Word-internal Aspiration: Preceding Segment

 

 

Factor Weight Application Percent
[o] k.o. 0/287 0%
[a] .589 35/239 15%
[e] .399 14/311 5%
[u] k.o. 0/5 0%
[i] .188 2/11 18%

 

Note. Factor group not significant

 

Table 45 shows the results for the contribution of stress to
aspiration. These findings suggest that whether the syllable carries
primary stress is not relevant for the aspirated variant. Poplack
(1981) found that stress was not significant in accounting for
aspiration word-internally in Puerto Rican Spanish. In general, more
data are needed to arrive at any conclusions about stress and

aspiration.

128

Table 46

Aspiraftionand Stress

 

 

Factor Weight Applications”l Percent
Stress .509 37/68 54%
No stress .463 12/16 75%

 

Note. Factor group not significant

‘Total cases before [1] and [n]

 

The VARBRUL analysis indicates that aspiration is favored
when (r) precedes a syllable boundary, but disfavored when it
precedes a clitic. However, the factor group was not significant and
these findings should be considered inconclusive. These results are

shown in Table 47.

Table 47

Word-internal Aspiration: Syllable vs. Clitic Boundgy

 

 

Factor Weight Applicationsa Percent
Syllable .608 29/43 67%
Clitic .392 20/41 49%

 

Note. Factor group not significant
‘Total cases before [1] and [11]

Input = .609; x202, a = 84) = 48.03, [sig.]; x2 cell 1.23

 

The factor groups representing the social variables of gender,
age, status and origin were not selected as significant for the
aspirated variant. Nevertheless, the results are discussed here to give

an indication of how aspiration may be constrained by social factors.

129

 

. ”WW ‘0‘“,.": .. ":0 . . .4;
X ' 5.
‘ “ ’ — UK “C 'w‘r 'k‘v‘_ .

 

Age, gender and status interacted, as shown in Figure 10.
Aspiration is favored by females in general, and in particular by
young females raised in upper status families, for whom its
application is nearly categorical (.909). Young middle and lower
status females have moderate probabilities (.616 and .628). The older
female also has a similarly moderate probability (.656). Aspiration is
not promoted by male gender: younger males have weights of .311,
for middle status, and .321 for upper status (the young working class
male only had one case of (r) before /1/ and /n/). The older male had
the most disfavoring weight of all: .194.

The gender differences noted here indicate that aspiration is a
variant associated with female gender, and suggest that it may be a
prestige variant, given its association with upper status females. The
strong preference shown by upper status females expresses status
rather than solidarity and may result from a greater need for this
group to express their social status linguistically since there is a
general lack of status symbols. This point was noted earlier in the

discussions of the deleted variant of (s) and the lateral variant of (r).

130

 

 

 

 

FYU FYM FYL FO MYU MYM MYL MO

Figpra 10. Aspirated variant of (r): Gender, age and status. llTotal
number represents cases 'of [h] before [1] and [n]. Input = .609;
x2(12, 11 = 84) = 48.03, [sig.]; x2 cell 1.23. Female young upper (a =
11/14), Female young middle (a = 7/10), Female young lower (a =
4/7), Female old (a = 3/7), Male young upper (p = 12/19), Male
young middle (a = 4/12), Male young lower (a = 0/1), Male old (a =
8/15).

131

r a . .70.. angst. .r

The results for origin, shown in Table 48, indicate that urban
origin favors application of [h], but rural origin disfavors it. These
results further support the notion of aspiration as a prestige variant
since urban features tend to be more prestigious than rural ones (note
that the lateral variant of (r) is negatively evaluated and that rural

origin favors its application).

Table 48

Aspiration and Origin

 

 

Factor Weight Applications‘ Percent
Urban .727 26/39 67%
Rural .300 23/22 51%

 

Note. Factor group not significant
‘Total cases of before [1] and [n]

Input = .609; x202, 11 = 84) = 48.03, [sig.]; x2 cell 1.23

 

Summary Concluaiona

From the above discussion, it can be concluded that aspiration
is a stable sociolinguistic variant not involved in change, indicated
by its low frequency and restricted environments for application. Its
function within the local community is that of differentiating gender
and status, in particular upper status females of urban origin.

The status of the aspirated variant in other varieties of
Caribbean Spanish is very different. Alba (1992) reports that in
Dominican Spanish, it is stigmatized. Alba found that it was used by
speakers from the lowest socioeconomic status group, and at a low

rate of 3%, but never by speakers from other groups. When

132

_ ...- _~_‘“. ,1, .. A!“
vmfi . -._ ”- ‘- m ’

 

 

respondents were directly asked about aspiration, negative
evaluations were overwhelmingly given by all speakers. Cedergren
(1973) also found that aspiration was preferred by lower
socioeconomic status individuals in Panama City. She also found that
it was used more by older males of rural origins. These results are
the opposite of those described above for this variety of Cuban
Spanish, in which it is preferred by young urban females. Poplack
(1981) also found that older speakers aspirated more in Puerto Rican
Spanish; Poplack also reported that education and gender did not
have any effect on its application.

The variety of Cuban Spanish studied here is more like
Andalusian and Canary Island varieties. This is not surprising since
contributions to the Cuban variety from the Canary dialect has been
substantial and relatively recent. Alonso (1961) claims that in Cabra
and Cordova individuals of urban origin aspirate preceding /n/ and
those of rural origin lateralize which parallels the results found here.
Similarly, Catalan (1989) notes that in Canary Island Spanish, [h]
occurs in urban speech.

Future study of the aspirated variant should use a larger amount
of data in order to build an appropriate statistical model of its
variation, which is somewhat difficult to obtain given the restricted

phonological environments in which it occurs.

The Basic Variant of (r)
The basic variant is the one used in standard speech. The

following section will discuss the results for the variable rule

133

analysis of [r] in relation to the other variants and their
contribution.
Resultsw Discussion

The factor groups the stepwise regression selected as
significant are shown in Table 49. The pattern described above for
word-internal and word-final assimilation is also found for the basic
variant: word-internal variation is constrained by both linguistic and
social factors, but word-final variation is primarily constrained by
social factors. One difference is that for the basic variant, as for the
lateral one, the following phonological environment is also
significant; this was not the case for assimilation. This is most likely
because assimilation is to a following consonant, but the following
environment for the basic, and lateralized, variant includes pauses
and vowels. Since most vowels have a very strong positive effect on
the application of [r], and certain consonants have a very strong
negative effect (i.e. the lateral), these differences contribute to the
factor group’s significance in this environment.

The factor group representing the type of boundary (r) occurs
before, whether a syllable or clitic boundary, was not significant for
the basic variant. This factor group has not been significant for any
of the variants, suggesting that the juxtaposition of (r) and a
following clitic is treated as a word-internal environment.

For word-final environments for (r), these results show that
(r)’s function as the infinitive morpheme or as part of a
monomorpheme does not constrain the application of [r] in any
significant way. The factor groups representing the social variables

age, gender and status interacted in the word-internal analysis, while

134

gender and status, but not age, interacted in the word-final one.
Origin was not significant in either analysis. The results for the
significant factor groups and the individual factors that effect

variation are discussed further below.

Table 49

Significaalt Factor Groups for the Baaic Variant

 

 

 

 

Word-internal Word-fina_l
Significant Not significant Significant Not significant
Following Boundary Following Preceding
Preceding Origin Gender & Status Stress
Stress Age Morpheme
Gender, Age Origin

& Status

 

FollowinggEnvironment

The results for word-internal application of the basic variant
are shown in Table 50. They indicate that the strongest positive
contribution is made by [m, b, d], sounds with the features
[ + voice, —back]. As noted earlier for word-internal assimilation, it
appears that the relevant features are voicing and place, evident in
the similar effects of [t] and [s], even though one is
[ - continuant] and the other [ + continuant]. The sounds [k, h],
which are [ —voice, +back] disfavor the basic variant since, as
discussed earlier, they strongly promote assimilation. The two sounds
that feed aspiration, [n] and [l], have a strong negative effect on

application of the basic variant. In sum, the features that promote

135

[r] are {-1- voice, -back] (and to a lesser degree, [— voice,
+ coronal]), and the features that most disfavor it have the opposite

values [-voice, +back].

Table 50
Following Environment and Word-internal |r|

 

Factor Weight Applications Percent
[m, b, d] .674 170/264 64%
[t, s] .583 116/204 57%
[g] .535 5/10 50%
[p] .521 11/24 46%
[k, h] .377 82/261 31%
[n] .228 7/23 30%
[l] .166 12/65 18%

 

Note. Input = .466; x207. 11 = 851) = 179.20,
[sig.]; x2 cell 1.19

The results in Table 51 show that the most favorable
environment for the application of [r] word-finally is before vowels,
particularly before the front mid vowel [e], which has an extremely
strong positive influence that is nearly categorical (.947). The high
front vowel [u] has a strong positive effect; [a] also makes a strong
positive contribution. The back rounded mid vowel [o] moderately
favors application. The only vowel that disfavors [r] is the high
front vowel, which made a strong contribution (.738) to lateralization

of (r).

136

 

1'4]:

..v..i...v.r.9cdh-1 .

Table 51

Word-final [r]: Followiag Environment

 

 

 

Factor Weight Applications Percent
[e] .947 69/70 99%
[u] .817 22/23 96%
[a] .781 67/72 93%
[o] .641 9/11 82%
[n] .602 15/17 88%
[s] .551 16/20 80%
[g] .509 3/4 75%
[b] .498 4/6 67%
[y] .489 5/6 83%
[f] .459 4/6 67%
[m] .414 12/17 71%
[p] .405 20/28 71%
[i] .399 18/23 78%
[d] .353 37/51 73%
[k] .321 37/52 71%

Pause .317 100/141 71%
[t] .267 13/19 68%
[h] .177 2/3 67%
[1] .100 24/53 45%

Note. Input = .868; x2(24, 11 = 851) = 99.68,

 

E3]; x2 cell 0.85

 

137

Although previous studies have noted that [r] occurs most
often when a vowel follows (Fails, 1985), these results show that not
every following vowel equally favors it, and that the high front
vowel actually disfavors its application. Fails (1985) found that
intervocalic /r/ was realized as [r] in 97.1% (299/308) of all
occurrences, with the remaining 2.9% being cases of deletion in
which the preposition papa was realized as [pa].

The basic variant is disfavored by a following pause, which is
the most favorable for word-final lateralization (.806). Fails (1985)
found that pauses were more favorable environments for [r] than
consonants, but these results suggest that consonants actually provide
more positive input than pauses.

The consonants that positively contribute to application of the
basic variant are [n] and [s], and their effects are not strong. [g]
has a neutral effect, but it was the consonant that most strongly
favored lateralization. Other consonants that contribute positively to
assimilation or lateralization disfavor [r], which is the case of [k],
[t], [h] and [l].

Preceding Environment

The effects of preceding phonological environment on the
application of the basic variant in word-internal and word-final
positions are shown in Table 52. These results show that the high
front and back vowels promote the basic variant word-internally, and
the mid front vowel only slightly favors it. The other two vowels,
[o] and [a] have a small negative effect. As noted earlier, a
preceding [0] tends to promote assimilation, and [a] tends to occur

before lateralization and aspiration. Word-finally, this factor group

138

 

talk—136‘]

Q

,. .. .-, raster.

is not significant; nor was it significant for the other variants of (r)

in this position.

139

 

.t. wrists.

Table 52

Precedingg Sound and Application of the Basic Variant

 

 

 

Group Factor Weight Applications Percent
[o] .435 109/286 38%
[a] .458 94/239 39%
Internala
[e] .573 189/310 61%
[i,u] .813 11/16 69%
[e] .529 142/184 77%
[a] .531 175/222 79%
Final” [i] .517 35/51 71%
[o] .421 124/165 75%
[u] k.o. 0/2 -

 

”Input = .466; 3807,11 = 851)=179.20,[sig];x2 cell 1.19

bFactor group not significant

 

140

. .3......r..4.919-r.

M

The results for stress in Table 53 indicate that when (r) is the
coda of a nonstressed syllable, it is more likely to be realized as the
basic variant than when the syllable is stressed. The factor group was
significant for word-internal occurrences, but not word-final ones,
since carrying primary stress has a neutral effect on application of
[r]. Alba (1988) found opposite results for a variety of Dominican
Spanish: that the standard variant was more frequent in stressed

(51%) than nonstressed (13%) syllables.

Table 53

Stress and the Basic Variant

 

 

 

Group Factor Weight Applications Percent
Stress .451 225/566 40%
Internal”
No stress .596 178/285 62%
Stress .486 418/547 76%
Finalb
No stress .599 59/75 79%

 

”Input: .466; x2(17,fl = 851)=179.20,[sig.];)(2 cell 1.19

bFactor group not significant

 

Boundary Type

The type of boundary was not significant in the stepwise
multiple regression. Table 54 shows the VARBRUL results for this
factor group. Although the clitic boundary favors retention of [r],
the weight for the syllable boundary is close to .50, indicating that it
does not affect the variation. As noted above, these results indicate
that the boundary created by the juxtaposition of the infinitive (r)

and an attached clitic is linguistically handled as a word-internal

141

. . f, V L: «:23: t

one, suggesting that the process is one of affixation and that

morphological constraints are not relevant in this variety.

Table 54

Word-internfljj]: Syllable vs. Clitic Boundazy

 

 

Factor Weight Applications Percent
Syllable .484 377/783 48%
Clitic .676 26/68 38%

 

Note. Factor group not significant

 

Morpheme Type

The type of morpheme in which (r) appears was not significant
in the stepwise multiple regression. These results are shown in Table
55. This factor group was not significant for the assimilated or the
lateral variants either. When (r) is an infinitive morpheme it slightly
disfavors the basic variant, while as part of a monomorpheme it
favors it. The case of the infinitive favoring nonstandard variants
was discussed by Cedergren (1973), who found that word-final (r)

was morphologically constrained, and Poplack (1981).

Table 55

Word-final [r]: Morpheme Type

 

 

Factor Weight Applications Percent
Infinitive .486 299/385 78%
Monomorpheme .523 178/237 75%

 

Note. Factor group not significant

 

142

.4;
I.t(

.. .. rut

It

‘1: 11.

iv!

2.

.. flutter: ..r r

The linguistic factors discussed in this section will be
discussed further in the Summary and Conclusion section of this
chapter. The following section discusses the social factors and their
effects on the basic variant.

Sofia] Factor;

The results for word-internal [r] showed interaction between
gender, age, and socioeconomic status, which was also the case for
assimilation. The results for word-internal position are shown Figure
_1_1. For word-final (r), gender and status interacted with each other
but not with age. The results for word-final applications by gender,

status, age are shown in Table 56.

143

n no:

. «JIP‘IV’I‘I I. .

 

 

 

 

 

 

 

 

 

 

 

 

 

FYU FYM FYL F0 MYU MYM MYL MO

Figure 11. Contributions of age, gender and status to the application
ofthe basic variant in word-internal position. Input = .466; x2(17, 11
= 851) = 179.20, [sig.]; x2 cell 1.19. Female young upper (a =
37/78), Female young middle (a = 106/192), Female young lower (a =
29/75), Female old (a = 81/109), Male young upper (11 = 75/241),
Male young middle (p = 24/70), Male young lower (a = 2/12), Male
old (a = 49/74).

144

.. .1 ......... rvwoiflwrl

The results for word-internal [r] are typical of standard
variants: they are preferred by females, older speakers and higher
status groups. The results for word-internal position show that older
females have a very strong tendency to use the basic variant (.819);
older males also prefer the basic variant, although not as strongly
(.740). Young middle status females tend to slightly favor it, as do
upper status females, but their effect is slightly less. Lower status
females and males of all status groups strongly disfavor the basic
variant. By far the strongest positive contribution is made by the
older male and female; even though middle status females make a
positive contribution, it is very small.

The results for the basic variant in word-final position in Table
56 indicate a strong preference for it by older speakers, but younger
ones strongly disfavor it. In addition, it is preferred in this position,
too, by middle and upper status females, and strongly disfavored by

males and lower status females.

145

1‘ Zilliu 5.1!]
:34

. . ..htL‘; 5’14!"qu

Table 56

Word-final Baaic Variant: Age, Gender and Status

 

 

 

Group Factor Weight Applications Percent
Age Young .318 303/437 69%
Old .858 174/185 94%
Gender Female middle .576 94/115 82%
and Male middle .359 35/61 57%
Status Female upper .525 44/50 88%
Male upper .291 163/221 74%
Female low .372 134/156 86%
Male low .240 7/19 37%

 

Note. Input = .868; 12(24, N = 851) = 99.68, [sig.]; x2 cell 0.85

 

146

 

Conclusion

The basic variant word-internally is constrained by linguistic
and social variables, whereas word-finally the social variables have a
greater effect and the only significant linguistic factor is following
environment, which appears to be a result of the strong positive
contributions made by vowels. Word-internally, preceding sound has
a strong effect, with the high vowels having a very strong positive
effect on its application. The effects of stress were small, but they
showed that a nonstressed syllable favored the basic variant. The
factor group representing the type of boundary, whether a clitic or
syllable, and the one representing (r) as an infinitive morpheme or
part of a monomorpheme were not significant for the basic variant
nor for any of the other variants described earlier.

The social variables showed the patterning expected for
standard variants: older speakers, females and higher status
individuals favor it. The factor group representing rural or urban

origin was not significant for the basic variant.

Summary Conclusions
Following Phonologigll Environment

The following phonological environment was significant for all
variants when they occurred in the word-internal position. In the
word-final position, however, it was significant for the lateral and
the basic variants, but not the assimilated one. It was proposed
earlier that this was because vowels can follow the basic and lateral
variants and their strong positive contribution makes the factor group

significant.

147

 

Word-internal (r)

The sounds that most strongly favor each variant word-
internally are shown in Table 57. The voiceless labials [p] and [f]
favor lateralization; the voiced labials [b] and [m] favor the basic
variant; the voiceless coronals [t] and [s] favor assimilation; the
voiced coronal [d] promotes both the basic and lateral variants.
Aspiration occurs only before [n] and [l].21 The voiceless dorsal
consonants [k] and [h] favor assimilation, but the voiced one [g]

favors lateralization.

 

21Although it was noted above that following sound was not selected
as significant for the aspirated variant, the fact it only applies before
[n] and [1] makes following sound highly significant.

148

 

::
T: e Eb. TH E
2;. 75:2 e Se. H853 a gm. 5 a
o2. E

«as. E m 26. 7533 6. «S. E 6 Se. 253a: fie. E e an”. E a

 

F» _> c> _> w> _>

_Iaasla 943.3 343.

 

 

 

 

3353:7303 3523/ Cw wcto>mm mwesom $153275

hm 03:.

149

Word-final (r)

The effects of following environment on word-final variation of

 

(r) are shown in Table 58. The basic and lateral variants occur before
vowels, and these results indicate that the high front vowel [i]
promotes lateralization, [o] favors assimilation, and [e, u, a]
strongly favor the basic variant.-

A following pause makes a strong positive contribution to the
application of the lateral variant, but disfavors the occurrence of the
basic one. Earlier studies (i.e. Fails, 1985; Hammond 1976) also
found the lateral to occur more frequently before a pause.

As for the contribution of a following consonant to the
variation of (r), these results indicate that assimilation is strongly
favored when [l] or [h] fol-low. Since aspiration does not occur
word-finally, the lateral provides input for assimilation. Similarly,
the nasal promotes the basic variant in the absence of the aspiration

rule. The voiced velar stop [g] promotes the lateral variant.

150

.. llt¥.il I. 1..
r

a .1 M c. stirri'yflfualfl 1

 

Table 58

Word-fiaal Variants of (r) and Followingjnvironment

 

 

 

 

[assim]' [r]b [l]c

[e] - .947 .144

[u] - .817 .296

Vowels [a] - .761 .412

[o] - .641 .512

[i] - .399 .738

Pause - .317 .806
[1] .732 .100 -

Consonant [h] .693 .177 k.o.

[n] kai .602 .565

[g] k.o. .509 .639

 

aNot significant; ”Input = 051; °Input = .868

 

151

Precediag Phonological Environment

 

The results for preceding phonological environment in word-
internal position are shown in Table 59. They show that certain
vowels favor the application of certain variants: a preceding [0]
favors assimilation; [a] favors aspiration; [e] favors lateralization;

and the high vowels favor the basic variant.

Table 59

PrecedingaSegalent in Word-internal Application of (r)

 

 

Assimilationa Lateralizationb Basicc Aspirationd
[o] .681 .271 .435 k4;
[a] .495 .574 .458 .589
[e] .343 .664 .573 .399
[i] .343 .343 .813 .188
[u] .361 kgo. .813 kgo.

 

Note. Factor group not significant for aspiration

‘Input==.359;”1nput==.036;°Input==.466;”Input==.609

 

The results for word-final (r) are shown in Table 60. Preceding
sound was not significant for any of the variants in this position. The
strongest effect is of a preceding [0], which favors assimilation, but

otherwise the effects are neutral.

152

Table 60

Word-final Application of (r) and Preceding Sound

 

 

Assimilation Lateralization Basic
[e] .445 .517 .529
[a] .437 .507 .531
[i] .534 .384 .517
[o] .644 .510 .421
[u] kgo. kgo. er

 

Note. This factor group was not significant for any of

these variants.

 

m
The results for the effects of stress on the variation of (r) are
shown in Table 61. This factor group has a small effect on variation,
as indicated by the fact that the weights are clustered around .50 or
not far from this score. Two exceptions are the low weights for
absence of stress on word-internal assimilation and word-final

lateralization.

Table 61

The Effects of Stress on Variation of(r)

 

 

 

 

Assimilated _B_a_si_c_ Lateral Aspirated
wi wf wi wf wi wf wi wf
Stress .554 .503 .451 .486 .516 .545 .509
No Stress .394 .483 .596 .599 .454 .194 .463
Significance sig. n.g. sig. n.s. n.s. sig. n.s. -

 

153

 

Social Factors

The analysis of social factors related to application of the
variants shows that they make a strong contribution to variation. The
assimilated and lateral variants have been involved in generational
change. Assimilation was a change in progress before the revolution:
it was suggested that it was a change led by males, rather than
females. The lateral is a change from below motivated by
sociopolitical changes; it reflects locality and solidarity as an
expression of new nationalism. The aspirated variant is used more by
females; it was suggested that it is used to express status rather than
solidarity. The basic variant reflects the standard norm, and it was
used more by older respondents, higher status individuals, and

females.

154

 

THE VARIABLE (l)

The phoneme /1/ has been reported as quite stable in Spanish. A
phonological study of the lateral phoneme in Cuban Spanish by Fails
(1985) revealed that it was realized as the basic lateral allophone in
99.4% of cases. Haden and Matluck (1977) note that in Cuban
Spanish the lateral liquid, compared to the rhotic liquid, is
remarkably stable, except for a tendency to neutralize to [r]. The
neutralization of the liquids has been identified as a characteristic of
the Cuban variety by a number of linguists, including Isba'sescu
(1968) and Choy Lopez (1986). However, quantitatively-oriented
phonological studies by Fails (1985) and Hammond (1977) did not
find any occurrences of [r] as a variant of /l/.

The lateral tends to be rather stable in other varieties of
Caribbean Spanish, too. Alba (1988) found categorical application of
the standard variant in the speech of upper middle class speakers of a
regional variety of Dominican Spanish. Chela-Flores (1986) notes
that the lateral is rarely realized as [r] in the urban dialect of
Caracas, Venezuela.

Not commonly discussed in many studies is the tendency of the
lateral to be involved in regressive assimilation in preconsonantal
environments. An [r] allophone of the lateral does arise, however,
when the lateral liquid assimilates to a following /r/. Examples from

Alonso (1961, p. 243) include [errei] for el rey ‘the king,’ and

 

[errico] for el rico ‘the rich one.’
Regressive assimilation of the lateral in preconsonantal
environments tends to occur across varieties of Spanish. Alonso

(1961) notes that the lateral may assimilate to a number of following

155

consonants, including /r/, /e/, /t/, /d/, /dz/, /y/, and /fi/.22 As noted
above, Catalan (1989) also reports assimilation to /h/ in rural speech
in the Canary Islands. In general, the linguistic patterning of
assimilation remains largely uninvestigated. Assimilation of the
lateral would be a low-frequency feature, if we accept earlier reports
of its stability in both Cuban and other Caribbean varieties.

As for the social patterning of assimilation, whether it is
constrained by external variables, such as region or social status,
also appears to have been largely uninvestigated, although Catalan
(1989), for example, claims that an urban versus rural distinction
constrains assimilation of the lateral before [h], where [ehhafo] e_l
ja_rr_g ‘the jug’ can be observed in rural, but not in urban, speech.

Although assimilation of the lateral occurs cross-dialectally in
Spanish, it is more advanced and generalized in Cuban Spanish than
in other dialects of Spanish. The assimilation of both /r/ and /l/ are
considered identical phonetic processes by Guitart (1976, 1980). In
Guitart’s description of assimilation, which is the one used here,
either liquid can assimilate to any following consonant. An earlier
study by Goodgall de Pruna (1970a, 1970b) describes a more limited
range of following consonantal environments for assimilation:
preceding the consonants /p, t, k, b, d, g, s, m/. This study, however,
was on a regional variety which did not traditionally have
assimilation of the liquids, and the limited environments in which
assimilation was found may reflect an early stage of change,

particularly when compared to Guitart’s findings on the dialect of

 

22/e/ and /dz/ occur in Peninsular varieties, but not in Cuban Spanish
unless it results from hypercorrection. In Cuba, /9/ > /s/ and /d2/ >

/y/.

156

 

l . flit" $700691

upper middle class Havana speakers, which reflect a more advanced
stage of assimilation.

There are, then, two differing views on the lateral’s realization
in Cuban Spanish: one holds that the lateral is extremely stable, with
nearly categorical application of the major allophone (e.g. Fails,
1985; Hammond, 1977); the other reports that in certain regional
dialects the lateral behaves like the rhotic liquid, frequently
undergoing regressive assimilation in most preconsonantal
environments (e.g. Guitart, 1976, 1980; Goodgall de Pruna, 1970a,
1970b)

This study considered assimilation of the lateral as a process
parallel to that of assimilation of the rhotic liquid, thus accepting
Guitart’s (1976) view of assimilation as a frequent and widespread
feature of the western dialect of Cuba. The current study attempts to
quantify the assimilated and basic variants of the lateral in a central
region variety in order to determine its frequency. More importantly,
this study examines the linguistic and social constraints which may
be involved in the variation of (l). The results of the variable rule
analysis are presented and discussed in the following sections of this
chapter. As with the variable (r), the analysis of (l) was carried out
separately for word-internal and word-final occurrences. As (1) has
two variants, a basic and an assimilated one, the VARBRUL analysis
reflects a binomial choice, in which the assimilated variant reflects

rule application and the basic one nonapplication.

157

Results and Discussion

Only social factors were significant for the variation of (1).
None of the linguistic factors made a significant contribution to the
application of assimilation. As shown in Table 62, the social factors
age, gender and status were significant and interacted in both word-
internal and word-final environments, but the factor group
representing origin was not significant. The linguistic factors of
following and preceding phonological environment and stress were

not selected as significant.

Table 62

Significant Factor Groups for Word-internflnd Word-final (l)

 

Significant Not significant

 

Gender, Age, Status Origin
Following segment
Preceding segment

Stress

 

These results indicate that the assimilation of (l) is not
constrained by the same factors as the assimilation of (r). For
instance, for (r) in word-internal position, linguistic factors, such as
following segment, preceding segment and stress, all contributed to
the application of assimilation (this was shown above in Table 19),
whereas these factors are not significant for (l) assimilation.
Furthermore, the results for (r) assimilation showed that constraints
were different depending on whether the segment appeared in word-

internal or word-final position, with linguistic variables making a

158

contribution to variation of word-internal occurrences, but not to
word-final ones (see Table 20 above). For (1), however, the same
constraints are relevant for both word-internal and word-final cases
of the variable, and the significant factors are social rather than
linguistic variables.

Followinag Environment

The results for word-internal and word-final assimilation are
shown in Table 63. Following consonantal environment was not
significant for assimilation in either environment. A clear pattern of
the consonants which favor or disfavor assimilation does not emerge
from these results.

Word-internally, the strongest positive contributions to
assimilation are made by [s] (.896) and [p] (.767); more moderate
ones are made by [k] (.659) and [m] (.606); and minimal
contributions are made by [g] (.531) and [t] (.518). When compared
to (r), in which following consonant was significant in this word-
internal environment, there are mainly differences rather than
similarities: [k], [s], and [t] (.637) favored assimilation of (r), but
[p] (.249) and [g] (.329) disfavored it, and [m] (.484) had a small
disfavoring effect.

The results for assimilation in word-final position are shown in
Table 63. In this environment, assimilation is favored by [ch]
(.841), [y] (.791), [r] (.694), [t] (.663) and [p] (.620). Even
though this factor group was not significant for (r), either, the
effects of following consonants were very different for that variable

since only [1] (.732) and [h] (.693) promoted assimilation of (r),

159

...I\.:

«I, I: 1111.

. ....:i.r..or.0.lrur.

Table 63

Application of Assimilation by Following Environment

 

 

 

“Mfight
Factor Word-internal“ Word-finalb
[k] .659 .505
[t] .518 .663
[g] .531 .466
[rn] .606 .466
[b] .278 .511
[r] .212 .694
[f] .387 .347
[d] .407 .433
[p] .767 .620
[s] .896 .429
[n] k.o. .083
[ch] k.o. .841
[y] - .791
[h] - .416

 

Note. Factor groups not significant

“Total number = 183, ‘Total Number = 575

 

160

. .. .I rflrgfll‘I

[t] (.526) had a small positive effect, and [y] (.378) and [p] (.251)
disfavored it.

Following environment, although not significant for the
application of assimilation for (l), is significant for assimilation of
(r). Perhaps at the root of this difference is the fact that (r)
assimilates to only /1/ and /n/ in most Spanish dialects, but (1) tends
to assimilate to a wide range of consonants in these same varieties.
In the Cuban variety, the addition of further environments to the
already wide range of possible environments for assimilation was not
socially significant. The spread of assimilation of (r), however, from
the two initial environments, before /1/ and /n/, to a wider range of
environments did become socially significant.

In general, these results indicate that following environment
does not contribute to assimilation of (l) in word-final or word-
internal positions. These results are different from those for (r),
which suggests that the assimilation of the two liquids are distinct
sociolinguistic processes which have the same phonetic outcome.

Preceding Segment

The results for preceding segment were not significant;
nevertheless, they are shown in Table 64 in order to highlight another
difference between the two liquid variables which further suggests
that assimilation of (l) and (r) are sociolinguistically distinct.
Preceding sound was significant for (r) in word-internal position, but
not in word-final; for (1) it is not significant in either case. The back
mid vowel [o] (.681) favored assimilation of (r), [a] was more or
less neutral, but the front mid and high vowels [i, e] retarded

assimilation by (.343) and the back high vowel [u] by (.361).

161

Similar results were found for word-final occurrences, although the
factor group was not significant in that position. Again, the back mid
vowel [o] (.644) favored assimilation; the other vowels had small

positive or negative effects—[i] (.534), [a] (.437), [e] (.445)——that

are not strong enough to constrain the variation.

162

, In!!! «is...

Table 64

Assimilation of (1) ad PrecedingaSound

 

 

Factor Weight Applications Percent
[a] .550 45/106 45%
[o] .238 4/23 17%
Word-internal [e] .552 10/23 43%
[u] .525 14/27 52%
[i] .370 1/4 25%
[e] .484 137/490 28%
[a] .589 26/85 31%
Word-final
[o] k.o. 0/9 --
[i] k.o. 0/5 --

 

Note. Factor groups not significant

163

Sm
The results for stress are shown in Table 65. This factor group
was not significant for either word-internal or final assimilation.
Word-internally, the results indicate that stress slightly favors
assimilation, but word-finally, stress disfavors its application.
Goodgall de Pruna (1970b) reported that stressed syllables favored
assimilation of the lateral, but those findings were not reproduced

here.

Table 65

Stressand the Application of Assimilation

 

 

 

Group Factor Weight Applications Percent
Stress .555 56/124 45%
Internal
No stress .387 21/59 36%
Stress .410 24/119 20%
Final
No stress .524 139/456 30%

 

Note. Factor group not significant

 

The differences in the results for (l) and (s) suggest that
different factors constrain the application of assimilation. Whereas
absence of stress was found to retard assimilation of word-internal
(r), the presence or absence of primary stress is not significant in the
variation of word-internal (1).

Age. Gender and Status
Word-internal (I)
The results for age, gender and status for the lateral in word

internal position are shown in Figure 12. Generational difference are

164

 

all

.4‘

highlighted for (l). Assimilation was strongly disfavored by older
male and female speakers. The older female had a lower probability
for assimilation than the male (.120 vs. .226). This is perhaps an
indication that males were less conservative at an earlier stage of the
change than females.

These results indicate that there has been extensive
generational change in the frequency of assimilation of (1). Age is
the only factor that significantly disfavors assimilation, which is
favored by all younger speakers, but disfavored by the older ones.
Figure 13 shows the probability weights for both liquid variables.
The results for (r) coupled with these results provide strong evidence
for the progression of assimilation in this speech community. The
older speakers have remarkably less assimilation in both cases than
the younger speakers. There is a difference between older and
younger speakers in the amount of assimilation that applied for each
variable. For instance, older speakers assimilate the lateral less than

the

165

Probability Weight

 

FYU FYM FYL F0 MY MYM MYL MO

Figure 12. Social variables and word-internal application of lateral
assimilation. Input = .391; 12(6, N = 183) = 0.00, [n.s.]; x2 cell 0.00.
Female young upper (11 = 11/15), Female young middle (11 = 16/35),
Female young lower (11 = 6/9), Female old (11 = 2/25), Male young
upper (11 = 27/56), Male young middle (11 = 12/24), Male young lower
(11 = 9/9), Male old (11 = 3/19).

166

 

a. .OIVItgll‘I‘

 

rhotic, but for the younger speakers there tends to be more
assimilation of the lateral. These differences may point to the
heightened social significance of (r) within the last generation.

The results for the young male speakers Figure 12 show a
categorical application of the assimilation rule in the young lower
status male. This is, however, based on only nine tokens and further
data would be needed to confirm and replicate this behavior. Males in
the upper and middle status groups behave similarly with respect to
this variable. Application of the assimilation rule is favored by both,
with the middle status males having a slightly more favorable weight
(.609) than the upper status males (.592). Similar results were found
for the young males for the variable (r). The comparison of (r) and
(1) can be found in Figure 13. Assimilation of (r) does not apply
categorically in the lower status male, although he does have the
highest probability of using assimilation (.798). Similar weights are
again found for the upper and middle status males (.664 vs. .649).
Their tendency to apply assimilation is somewhat lower than their
lower status counterpart.

The probability weights obtained for the young females are
among the most interesting of these results. In particular, it was
unexpected that young upper status females would have a strong
tendency to assimilate (.811). This is radically different from the
assimilation of word-internal (r), as shown in Figure 13, which is
slightly disfavored (.464) by young upper status females.

The strong tendency for these speakers to assimilate the lateral,
but not the rhotic, suggests that assimilation of (l) is not socially

significant in the same way as assimilation of (r).

167

 

 

+ (I)
+ (r)

 

 

\
\;

 

 

 

 

 

.9
\1

o
as
\

 

 

/ /
// V
J

5,!

.o
A

 

Probability Weight
0
\II

.0
w

 

 

.99
—-N

 

 

 

O

 

I 1 I 1 I

FO MO FYU MYU FYM MYM FYL MYL

Figure 13. Probability weights for word-internal assimilation of (l)
and (r). For (r), input = .359; x2(18, 11 = 851)=170.14,[sig.];)(2
cell 1.13. For (1), input = .391; x2(6, N=183)= 0.00, [n.s.]; x2 cell
0.00

168

 

.1. n :rrb.\l '11

1 .1. t. It L.I,Pl'r.,go

This reduced significance would make lateral assimilation a safe way
of signaling their participation in the local community’s speech
norms without compromising their attention to prestige and
conservative stance, which have been evidenced in their disfavoring
the stigmatized lateral variant of (r), while favoring the prestigious
aspirated variant of (r), and favoring the aspirated variant of (s) over
the negatively evaluated deleted one.

Young middle status females only slightly favor assimilation
(.568) of the lateral. When compared to (r), for which they disfavor
assimilation (.418), they have a slightly higher tendency to
assimilate the lateral, as shown in Figure 13. In general, however,
assimilation is not strong in this group for either variable. Lower
status females showed a strong tendency to assimilate the lateral
(.757). They showed a similar tendency to assimilate (r) (.675).
Again, the lateral is more likely to undergo assimilation than the
rhotic, which suggests that it is more advanced in the speech
community and is not charged with strong social evaluations.
Word-final (I)

Figure 14 contains the results for assimilation when the lateral
variable occurs in word-final position. Strong evidence is also found
in this position for generational differences. As with word-internal
assimilation the older speakers disfavor assimilation of the lateral,
and the older female disfavors it somewhat more strongly than the
male (178 vs. 245). The assimilation rule had positive probabilities
of application for all of the younger speakers. This suggests that
assimilation has progressed quite vigorously in the local speech

community during the last generation.

169

Probability Weights

 

FYU FYM FYL FO MYU MYM MYL MO

Figure 14. The word-final application of assimilation according to
gender, status and age. Input = .246; x2(7, N = 589) = 0.00, [n.s.]; x2
cell 0.00. Female young upper (a = 10/36), Female young middle (a =
46/143), Female young lower (a = 30/58), Female old (a = 5/76),
Male young upper (a = 44/127), Male young middle (a = 18/59), Male
young lower (a = 2/6), Male old (a = 8/84).

170

In word-final position, lower status females have the strongest
tendency to assimilate (.767) the lateral, and they tend to assimilate
more than the young lower status males (.606), who had categorical
application of word-internal cases. The young upper status males had
a slightly higher probability weight than their lower status
counterparts (.619), but they do much less assimilation than the
young lower status females. The young upper status females (.593)
behaved more or less like the young males (upper .616, lower .606,
middle .574). The young middle status females only slightly favored
assimilation (.542).

It has been noted throughout this dissertation that the word-
final environment is more salient, which makes it a less favorable
environment for the application of nonstandard variants. Thus, word-
internal assimilation of the lateral is very strongly favored by upper
status young females, but is only slightly favored word-finally. The
lower status females, however, are perhaps using word-final
assimilation to enhance their status.

In general, the assimilation of the lateral in word-final position
does not appear to be as socially significant as the assimilation of
word-final (r), which was strongly favored by young males lower
status males, and disfavored by females, older speakers, and upper

and middle status groups.

Conclusion
The variable rule analysis of (1) showed that linguistic factors
are not significant in the application of assimilation. The variation of

(l) is constrained primarily by the social factors, age, gender, and

171

 

 

i. . . ....~.... ,0 .9.
..tg'u?

status. This pattern differs from the one found for assimilation of (r),
which is constrained word-internally by linguistic factors, including
following segment, preceding segment, and stress.

These results suggest that the lateral and rhotic liquids as
sociolinguistic variables behave very differently. From a
phonological perspective, assimilation has previously been
considered the same for both (r) and (l), with identical phonetic
outcomes (Guitart, 1976, 1980). Similarly, from the perspective of
dialectology, assimilation of (l) and of (r) have been together
considered a feature of the western region of Cuba. This study has
shown, however, that the application of assimilation for (r) and (l) is
constrained by very different factors.

Another important difference between (1) and (r) is that the
assimilation of (r) is socially significant, whereas the assimilation of
(1) appears not to be charged with social evaluation. Evidence for
this was found in the tendency of young upper status females to show
a stronger tendency than most other speakers to assimilate in word-
internal environments. It was suggested that since the lateral carries
little social significance, assimilation of the lateral is a safe manner
for the young upper status females to express their loyalty to the
speech norms of the local community without compromising their
attention to prestige and conservative stance.

The analysis of (1) provides further quantitative evidence of
generational change in the use of the assimilated variant, which has
become more widespread in local speech. Evidence of change was
found in both word-internal and word-final positions. It was shown

that the application of assimilation is strongly disfavored by older

172

. .mfl'fifip u.“

 

 

speakers. The fact that the older female disfavors it more strongly
than the male may perhaps be an indication that males led the spread
of assimilation at an earlier point in time; but, of course, more data
is needed to get a better understanding of how the change spread in

the community.

173

...r .. ......l Hi.
09 1

la].

 

 

.. it funky? ..

SUMMARY CONCLUSIONS

This chapter summarizes the most important conclusions from
the chapters on (r), (l) and (s) in order to address the research
questions stated in the introduction to this dissertation. Questions
were posed about the extent and direction of sound change, and about
the nature and role of external and internal constraints on variation.
These questions are addressed individually below.

The first question addressed the extent of sound change from
one generation to the next. This study found strong evidence for
generational change in all three variables. The strongest evidence for
change was found in the liquid variables, (r) and (l). The findings for
(8) indicated that deletion has increased in the speech of younger
speakers, particularly in the word-internal environment, in which
older speakers favored aspiration, but younger ones did not. In
general, the deleted variant was preferred. by younger male speakers.
The nonstandard assimilated, lateral and aspirated variants of (r)
showed strong effects for age. The previously stigmatized lateral
variant had increased significantly in the speech of young college-
educated individuals, particularly males, as compared to the older
speakers. The aspirated variant of (r) was used by older speakers
regardless of gender, but in the younger group it was preferred by
females whose parents had been upper middle class before the
revolution. The assimilated variants of (r) and (1) showed strong
contributions from age: older speakers had few occurrences of it in a
limited number of environments, while younger speakers showed a
strong tendency to assimilate in a wide range of environments, which

suggests that this is not a new change, but rather one which has been

174

in progress for many years. Liquid assimilation, as indicated by these
generational differences, has become a feature of the local dialect.

The second question addressed the direction of change,
specifically focusing on the frequency and distribution of the
previously stigmatized variants, the lateral variant of (r) and the
deleted variant of (s). The findings indicated that both lateralization
of (r) and deletion of (8) had increased significantly in the younger
group. Assimilation of (r) and (1) also showed a marked increase in
younger speakers. Assimilation of the lateral liquid was not found to
be as socially significant as the assimilation of the rhotic liquid.

The third question addressed the factors involved in the spread
of these changes. It was argued that sociopolitical changes had
created conditions which motivated and strengthened the diffusion of
variants previously associated with working class and rural speech.
Two main points were made. First, it was argued that young Cubans
have been exposed to a new sociolinguistic norm, which has arisen
because of social promotion and the needs of the revolutionary
regime. The socialization of children in revolutionary values together
with the expectation that young Cubans be committed to the
revolution in order to succeed exert strong pressure on young Cubans
to conform to the new norm.

The term covert prestige (Trudgill, 1972) was extended beyond
an individual desire to express characteristics associated with
masculine and working class speech to include a larger societal
desire to associate with working class speech as a symbol of aavy

aationalism and solidarity with the revolution.

 

175

 

. $19.51 1.11
. in a. .

.trrtnh'g

 

Second, it was proposed that rural features have become
prestigious because they are associated with the economic benefits of
having rural origins. It was argued that young college-educated
individuals in Cuba are oriented toward rural and working class
variants because their lifestyles and concerns are essentially working
class.

It was suggested that the increase of the lateral variant of (r)
and the deleted variant of (s) in the speech of college-educated young
adults resulted from sociopolitical changes. Assimilation of the
liquids, however, was a change in progress at the time of the
revolution, and the spread of liquid assimilation cannot be attributed
directly to changes brought about by the revolution. It was suggested
that the spread of (r) assimilation was led by male speakers, who
were attracted to the its urban association. It was proposed that the
spread of assimilation was from one urban center to another, and that
from each city, it spread outward in waves, rather spreading out in a
linear manner, covering each region completely as it progressed.

The fourth question addressed the linguistic constraints on
variation of (s), (r) and (l), which had not been previously studied in
this variety. An important finding of this research is that assimilation
of the lateral and rhotic liquids are distinct sociolinguistic processes,
which have the same phonetic outcome. Linguistic factors are
significant for the assimilation of (r), but they do not play a role in
the assimilation of (l).

The results of this study indicated that following phonological
environment was the most important linguistic factor. Morphological

distinctions were studied for (r), but they were not found to

176

 

significantly contribute to its variation. In word-internal cases,
whether the variable was an infinitive morpheme at a clitic boundary
or nongrammatical preceding a syllable boundary did not contribute
to variation; word-finally, whether (r) was an infinitive marker or
part of a monomorpheme was not significant, either, although
morphological considerations were significant for the variable (s),
which revealed a three-way interaction between morpheme type,
following segment and stress. Stress was found to make small-to-
large contributions to the application of most variants. For (r) in the
word-internal environment, stress had a moderate effect on the
application of the basic variant, a small effect on assimilation, but no
effect on either lateralization or aspiration. Stress constrained the
variation of (s), either by itself, as in word-internal cases, in which
the presence of stress promoted aspiration, or in conjunction with
other factors, such as morpheme type and following sound, which was
the case for word-final (8).

An important finding of this study is that linguistic factors
tended to contribute more to variation in the word-internal
environment than the word-final one, where social factors tend to be
more important than linguistic ones. For instance, assimilation of (r)
was constrained by following and preceding segment, and by stress
word-internally, but none of these factors was significant word-
finally. Similarly, the application of the basic variant was affected
only by following segment word-finally, but following segment had a
strong effect word-internally and preceding segment and stress had

moderate effects.

177

 

 

 

......I.V.. ‘1
1

 

. _Jtrrbw?o¥.

This study suggests that the internal constraints on variation in
Cuban Spanish are different from those in other varieties of
Caribbean Spanish.

The first paragraph of this dissertation stated that a perceptual
dialectology study (Alfaraz, in press) in which Cubans in Miami were
asked to rate Cuban Spanish on the island on dimensions of status
and solidarity found that, regardless of their length of residence in
the US, Cubans downgraded Cuban Spanish on the island. That
study raised questions about linguistic change in Cuba, about the
nature of sound changes which had taken place within the last thirty
to forty years. This dissertation found evidence of generational
change in previously stigmatized and nonstigmatized variants of (r),
(l) and (8).

Whether the variants examined here, particularly the previously
stigmatized ones, contribute to negative perceptions of Cuban
Spanish on the island needs to be addressed in future research. Future
research should also compare the findings discussed here for young
college-educated speakers to other groups of speakers in order to
better understand the connection between social mobility, social

class leveling and social stratification in noncapitalist societies.

178

 

 

1:: .l ”I .'90

 

Bibliography

179

 

 

. .11»... rrvtIrQn-flv-l

BIBLIOGRAPHY

Alba, 0. (1982). Funcién del acento en el proceso de elision de
la /s/ en la Republica Dominicana. [The function of stress on the
deletion of /s/ in the Dominican Republic]. In 0. Alba (Ed.), _El
espanol del Caribe (pp. 15-26). Santiago de los Caballeros, R.D.:
UCMM.

Alba, 0. (1988). Estudio sociolingfiisticos de la variacion de
las liquidas finales de palabra en el espafiol cibaefio. [Sociolinguistic
study of the variation of word-final liquids in the Cibao variety of
Spanish]. In R.M.. Hammond & M.C. Resnick (Eds.), Studies in
Caribbean Spanish dial_ectology (pp. 31-6). Washington, DC: George
Washington University Press.

Alba, O. (1992). Diferenciacion objetiva y valoracion social del
debilitamiento de dos segmentos consonanticos en el espafiol
dominicano. [Ojective differentiation and social evaluation of the
weakening of two consonant segments in Dominican Spanish]. In M.
Vaquero & A. Morales (Eds), Homenaie a Humberto Lopez Morales
(pp. 67-74). Madrid: Editorial Arco/Libros.

Alfaraz, G. G. (1999). Generational chflge in Cuban Spanish:
Liquid assimilation in the speech of older and younger middle cla__ss
speakers in Miami Paper presented at 17th Conference on Spanish in
the United States, Miami, FL.

Alfaraz, G.G. (in press). Miami Cubans perceptions of varieties
of Spanish. In D. Long & D.R. Preston (Eds), Handbook of
perceptual dimctologya Amsterdam and Philadelphia: John
Benjamins.

Almendros, N. (1958). Estudio fonético del espafiol en Cuba.
Boletin de la Academia Cubaaa de la Lengua, 7, 138-76.

Alonso, A. (1961). Estudios lingfiisticos: Temaa
hispanoamericanos (2nd ed.). [Linguistic studies: Latin American
Topics]. Madrid: Gredos.

Amaro, NR. (1989). Mass and class in the origins of the Cuban
Revolution. In I.L. Horowitz (Ed.), Cuban communism (7th ed., pp.
39-62). New Brunswick: Transaction Publishers.

Ariza, M. (1994). Sobre fonética historica del espafiol [On
Spanish historical phonetics]. Madrid: Arco Libros.

Baloyra, EA. (1989). Political control and Cuban youth. In I.L.

Horowitz (Ed.), Cuban communism (7th ed., pp. 429-42). New
Brunswick: Transaction Publishers.

180

Bayley, R., & Young, R. (1996). VARBRUL analysis for second
language acquisition research In R. Bayley & D. R. Preston (Eds. ),
Second language acauisition and linguistic variation (pp. 253- 306)
Amsterdam and Philadelphia: John Benjamins.

Boyd-Bowman, P. (1975). A sample of sixteenth century
'Caribbean' Spanish phonology. In Colloquium on Spanish and
Portaguese Linguistics (pp. 1-11). Washington: Georgetown
University Press.

Bunck, J.M. (1989). The Cuban Revolution and Women's
Rights. In I.L. Horowitz (Ed.), Cuban communiam (7th ed., pp. 443-
64). New Brunswick: Transaction Publishers.

Callery, RE. (1975). Phonological change and the development
of an urban dialect in Illinois. Language in Society, 4, 155-70.

Catalan, D. (1989). El espafiol: Origenes de su diversidad
[Spanish: The origins of its diversity]. Madrid: Paraninfo.

Cedergren, H]. (1973). The interplay of social and linguistic
factors in Panama. New York: Cornell University.

Chambers, J.K. (1992). Dialect acquisition. WM),
673-705.

Chela-Flores, G. (1986). Las teorias fonologicas y los dialectos
del Caribe hispanico [Phonologial theories and Spanish Caribbean
dialects]. In R.A. Nunez Cedefio, I. Paez Urdaneta & J.M. Guitart
(Ecllls), Estudioa sobre la phonologia del Caribe. Caracas: La Casa de
Be o.

Choy L6pez, LR. (1986). Sistema fonético y sistema
fonologico. Neuphilologische Mitteilungen, 3. 400-13.

Clayton, M.L. (1985). Trying to make the data stand still:
Continuity, gradience and indeterminacy in Spanish phonology. In

F.H. Nuessel, Current issues in Hispanic phonology and morphology
(pp. 1-13). Bloomington, IN: Indiana University Linguistics Club.

 

Edwards, W.F. (1992). Sociolinguistic behavior in a Detroit
inner-city black neighborhood. Language in Society, 21, 93-115.

Fails, WC. (1985). An analysis of the consonantal phonemes of
the educated norm of Havana (Doctoral dissertation, The University
of Texas at Austin, 1984). Dissertation Abstracts InternationaL
fl(07),

Fairclough, N. (1989). Language and Power. New York:
Longman.

181

 

 

 

. .1510 a ‘11) r
r

r . ...‘i... Fins:-

Fasold, R.W. (1978). Language variation and linguistic
competence. In D. Sankoff (Ed.), Linguistic variation: Models a_na
methods (pp. 85-96). New York and London: Academic Press.

Fitzgerald, PT. (1990). Cuba's new professionals. In S.
Halebsky & J.M. Kirk (Eds.), Iaansformaalon and struggle: Cuba
faces the 1990s (pp. 189-203). New York: Praeger.

Frago, IA. (1983). Materiales para la aspiracion de /-8/
implosiva en las hablas andaluzas. LEA. 1983. 153-71.

Garcia Gonzalez, J. (1980). Acerca de la pronunciacon de R y L
implosivas en el espanol de Cuba: Variantes e influencias. Islas 65
115-17.

 

Goodgall de Pruna, R. (1970a). La geminacion de consonantes
en el habla de Trinidad. Islas 11(34), 125-28.

 

Goodgall de Pruna, R. (1970b). La pronunciacon del idioma
espafiol en el centro de Cuba. Islas 11(37), 155-60.

 

Guitart, J.M. (1976). Markedness and a Cuban dialect of
Spanish. Washington, DC: Georgetown University Press.

Guitart, J.M. (1980). Aspectos del consonantismo Habanero:
Reexamen descriptivo [Aspects of Havana consonants: Descriptive
reexamination]. In G.E. Scavnicky (Ed.), Dialectologia
hispmamericana: Estudios actuiales (pp. 32-47). Washington, DC:
Georgetown University Press.

 

Guitart, J.M. (1982). On Caribbean Spanish phonology and the
motivation for language change. Current research in Romance
languages (pp. 63-70). Bloomington: Indiana University Linguistics
Club.

Guy, GR. (1993). The quantitative analysis of linguistic
variation. In D.R. Preston (Ed.), American dialect research (pp. 223-
49). Amsterdam and Philadelphia: John Benjamins.

Haden, E.F. & Matluck, J.H. (1977). El habla culta de la
Habana: Analisis fonologico preliminar [The cultivated speech of
Havana: Preliminary phonological analysis]. In J. Lope Blanch (Ed.),
Estudios §obre e1 espafiol haflado en la_s principales ciudades de
América (pp. 13-3 7). Mexico: Universidad Nacional Autonoma de
Mexico.

Hammond, RM. (1977). Some theoretical implications from
rapid speech phenomena in Miami-Cuban Spanish (Doctoral
dissertation, University of Florida, 1976). Dissertation Abstracts
InternationaalL37(6444A), .

182

Harris, J. (1985). Autosegmental phonology and liquid
assimilation in Havana Spanish. In L. King & C. Maley, Papers from
the 13th Lingistic Symposium on Romance Languages (pp. 127-48).
Amsterdam: Benjamins.

Harris, J. (1995). The morphology of Spanish clitics. In H.
Campos & P. Kempchinsky (Eds), Evolutionaand revolution in
liaguistic theory (pp. 168-97). Washington, DC: Georgetown
University Press.

Isbasescu, C (1968). El emafiol en Cuba: Observaciones
fonéticas y fonologicas [Spanish in +Czuba Phonetic and phonological
observations]. Bucharest: Sociedad Rumana de Lingfiistica Romanica.

Kontra, M. (1992). Class over nation - linguistic hierarchies
eliminated: The case of Hungary. Multilingua, 11-2, 217-221.

Labov, B. (1972). Sociolinguistic Patterns. Philadelphia:
University of Pennsylvania Press.

Labov, W. (1963). The social motivation of a sound change.
Word. 19, 273-309.

Labov, W. (1982). The social stratification of English in New
York City. Ann Arbor: UMI Books on Demand (Original work
published 1966)

Labov, W. (1990). The intersection of sex and social class in
the course of linguistic change. Language Variationand Change, 2,
205-54.

Labov, W. (1994). Principles of lingalsitic change: Internal
factors. Cambridge, MA: Blackwell.

Lafford, BA. (1986). Valor diagnostico del uso de ciertas
variantes de /s/ en el espaflol de Cartagena, Colombia [Diagnostic
value of the use of variants of /s/ in the Spanish of Cartagena,
Colombia]. In R.A. Nunez Cedei’lo, I. Paez Urdaneta & J.M. Guitart
(Eds), Estudios sobre la phonologg del Caribe (pp. 53-74). Caracas:
La Casa de Bello.

Lamb, A.J. (1969). A phonological study of the Spanish of
Havana, Cuba (Doctoral dissertation, University of Kansas, 1968).
Dissertation Abstracts International, 29(2244A).

 

Lipski, J.M. (1994). Latin American Spanish. London:
Longman.

Lopez Morales, H. (1970). Estudios §obre e1 espaflol de Cuba
[Studies on the Spanish of Cuba]. New York: Las Americas
Publishing.

183

 

Lopez Morales, H. (1987). M88 alla de la regla variable: La
velarizacion de /r/ en el espafiol de Puerto Rico. [Beyond the variable
rule: The velarization of /r/ in the Spanish of Puerto Rico]. In T.A.
Morgan, J.F. Lee & B. VanPatten (Eds), Language and language use:
Studies in Spanish (pp. 31-6). Lanham, MD: University Press of
America.

Lopez Morales, H. (1992). El espafiol del—caribe [The Spanish
of the Caribbean]. Madrid: Editorial MAPFRE.

Megenney, W. (1989). An etiology of /-s/ deletion in the
Hispanic Caribbean: Internal processes or substratum influence?.
Estudios aobre e1 espajiol de Américaa/ liaguisticgfroamericaaa
(pp. 300-27). Bogota: Instituto Caro y Cuervo.

Milroy, L. (1987). Language and Social Networka. Oxford:
Basil Blackwell.

Montori, A. (1916). Modificaciones populares del idioma
pastellano en Cuba. [Popular modifications to the Spanish language
in Cuba]. Havana: Imprenta Cuba Pedagogica.

Payne, AC. (1980). Factors controlling the acquisition of the
Philadelphia dialect by out-of-state children. In W. Labov (Ed.),
Locating language in time and spa (pp. 143-78). New York:
Academic Press.

Poplack, S. (1981). Function and process in a variable
phonology. Ann Arbor: University Microfilms International.
(Original work published 1979)

Posner, R. (1996). The Romance languages. Cambridge:
Cambridge University Press.

Preston, D. R. (1985). Mental maps of language distribution in
Rio Grande do Sul (Brazil). The Geographical Bulletin. 47, 46-64.

Preston, DR. (1989). Sociolinguistics and second language
acquisition. Cambridge and Oxford: Basil Blackwell.

Preston, DR. (1996). Whaddayaknow? Folk Linguistic
Awareness. Language Awareness. 5(1), 40-74.

Prieto, M.E., & Moreno, L.N. (1990). The changing class
structure in the development of socialism in Cuba. In S. Halebsky &
J.M. Kirk (Eds), Transformation raid struggle: Cubafitces the 19908
(pp. 205-18). New York: Praeger.

Sankoff, D., & Laberge, S. (1978). The linguistic market and
the statistical ex lanation of variability. In D. Sankoff (Ed),
Liaguistic va_riatlon: Models and methods (pp. 239-50). New York
and London: Academic Press.

184

 

Smith, L.M., & Padula, A. (1990). The Cuban family in the
1980s. In S. Halebsky & J.M. Kirk (Eds), Transformation ya
struggle: Cuba faces the 19908 (pp. 175-188). New York: Praeger.

 

Terrell, T. (1976). Variacion fonética de /r/ y /rr/ en el espafiol
cubano. Revista de Filologia Espanola, LVIII(1-4), 109-32.

Terrell, T.D. (1977a). La aspiracion y elision en el espafiol
cubano: Implicaciones para una teoria fonologica dialectal
[Aspiration and deletion in Cuban Spanish: Implications for a
phonological dialect theory]. In J. Lope Blanch (Ed. ), Estudios sobre
el espafiol hablado en las principales ciudades de América (pp. 39-
48). Mexico: Universidad Nacional Autonoma de Mexico.

Terrell, T.D. (1977b). Constraints on the aspiration and
deletion of final /s/ in Cuban and Puerto Rican Spanish. The
Bilingual Review. 4(1), 35-51.

Terrell, TD. (1979). Final /s/ in Cuban Spanish. Hispania, 62,
599-612.

Torreblanca, M. (1989). La /s/ implosiva en espaflol: Sobre las
fechas de su aspiracion. Thgaurus: Boletin del Instituto Cflo y
Cuervo. Bogota. Colombia. 44(2), 281-303.

Trudgill, P. (1972). Sex, covert prestige and linguistic change
in the urban British English of Norwich. Language in Society, 1, 179-
95.

Trudgill. (1974). The §ocial differentiation of English.
Cambridge: Cambridge University Press.

Uber, DR. (1984). Phonological implications of the perception
of- -s and -n in Puerto Rican Spanish. In P. Baldi (Ed. ), P_apers from
the XIIth Linguistic Symposium on Romance Languages s(pp. 287- 99).
Amsterdam/Philadelphia: John Benjamins.

Vallejo-Claros, B. (1971). La distribucion y estratificacion de
/r/ If/ y /s/ en el espafiol cubano (Doctoral dissertation, The
University of Texas at Austin, 1970). Diasertaaion Abstracts
Internation. 31(6039).

Wodak, R. & Benke, G. (1997). Gender as a Sociolinguistic
Variable: New Perspectives on Variation Studies. In F. Coulmas

(Ed.), The Handbook of Sociolinguistics (pp. 127-50). Oxford:
Blackwell.

Wolfram, W., & Schilling-Estes, N. (1998). American English.
Oxford: Blackwell.

185

 

Wolfson, N. (1976). Speech events and natural speech: Some
implications for sociolinguistic methodology. Language in Society,
;(2), 189-209.

186

 

.11 ....1.....1u1.. 1.: $1.. 5; l