Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Prosody in contact: Spanish in Los Angeles
(USC Thesis Other)
Prosody in contact: Spanish in Los Angeles
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
PROSODY IN CONTACT:
SPANISH IN LOS ANGELES
by Sergio Robles-Puente
______________________________________________________________________________
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the Requirements for the Degree
DOCTOR OF PHILOSOPHY
(LINGUISTICS (HISPANIC LINGUISTICS))
August 2014
TABLE OF CONTENTS
Dedication i
Acknowledgements ii
Abstract iii
Chapter 1: General introduction 1
Chapter 2: Rhythm 7
2.1. Introduction 7
2.2. Background: Quantifying rhythm 8
2.3. Spanish and English rhythm in Los Angeles 12
2.3.1 Participants 12
2.3.2 Methodology 16
Voicing ratios 16
Normalized Pairwise Variability Index 18
Results 19
Voicing ratios results 19
nPVI results 23
Voicing ratios vs. nPVI: Comparing results 27
Voicing ratios vs. nPVI: Discussion 29
2.3.3 Vowel durations as rhythm shapers 30
Vowel durations in English 31
Vowel durations in Spanish 34
Vowel durations: Conclusions 39
2.4. Rhythm: General discussion 40
Chapter 3: Intonation 44
3.1. Introduction 44
3.2. General Methodology 46
3.3. Declaratives: Pre-nuclear pitch-accents in broad focus 48
3.3.1 Introduction 48
3.3.2 Methods 49
3.3.3 Results 50
3.3.4 Discussion 55
3.4. Declaratives: Pre-nuclear pitch-accents in narrow focus 56
3.4.1 Introduction 56
3.4.2 Methods 57
3.4.3 Results 57
3.4.4 Discussion 60
3.5. L+H* vs. L+>H* 60
3.5.1 Introduction 60
3.5.2 Methods 60
3.5.3 Results 61
3.5.4 Discussion 64
3.6. Declaratives: Nuclear pitch-accents 66
3.6.1 Introduction 66
3.6.2 Methods 68
3.6.3 Results 69
3.6.4 Discussion 73
3.7. Imperatives 73
3.7.1 Introduction 73
3.7.2 Methods 76
3.7.3 Results 77
3.7.4 Discussion 85
3.8. Wh- questions 87
3.8.1 Introduction 87
3.8.2 Methods 87
3.8.3 Results 87
3.8.4 Discussion 90
3.9. Yes/No questions 90
3.9.1 Introduction 90
3.9.2 Methods 91
3.9.3 Results 92
3.9.4 Discussion 96
3.10. Vocatives 97
3.10.1 Introduction 97
3.10.2 Methods 97
3.10.3 Results 98
3.10.4 Discussion 102
3.11. Intonation: Overall results 102
3.12. Intonation: General discussion 103
Chapter 4: Discussion and conclusion: rhythm and intonation compared 108
Appendix 120
References 133
Notes 148
i
DEDICATION
This work is dedicated to my family for their constant support
ii
ACKNOWLEDGEMENTS
First of all I would like to thank my dissertation committee for taking the time to read and
discuss my research with me, especially my advisor, Louis Goldstein, for showing me his
support at not only the academic but personal level. Also thank you to Maria Luisa Zubizarreta,
Carmen Silva-Corvalán and Mario Saltarelli, for their guidance since day one in the Hispanic
Linguistics Program and the Spanish Department. I am also indebted to Rachel Walker for all the
times that she has devoted in providing me feedback and the interesting discussions within the
six years I have spent at USC. Also thanks to Phonlunch and SPAN members for their valuable
comments and feedback. And of course, thank you, to all the people in the Linguistics
Department (teachers, students, and staff) that, for one reason or another, have been part of this
trip.
I am also indebted to the Dornsife Del Amo Foundation and the Spanish Ministry of Science and
Innovation for their economic support. Special thanks to Yolanda Congosto, coordinator of the
PROESLA-AMPER project and to all the participants that helped me with my research and took
part in the recordings. This dissertation would not have been possible without them.
Also, thank you, to all the people in the Spanish Department: Deborah, Amelia, Martha,
Consuelo, Andrea, Vianey and a very long etcetera. Special thanks to Gayle Vierma and David
Zarazua for their help, while I was working in the Spanish Department. The last thank you in the
academic front is for Jon Franco, for his endless support for over ten years now. If it had not
been for his advice, I would probably never have ended up being a linguist.
Thanks to all the people I have met in Los Angeles in these six years for all the good (and bad)
experiences because you made my life more interesting. Particularly thank you to the Richards,
my Spanish friends, the Gak House residents and my Futsal teammates. All those meals/dinners,
outings, goals, and championships were necessary breaks to make my academic life more
productive. Special thanks to my coaches, Henrique and Ryan, for their friendship and for being
“life coaches” and to Hector, for his wise pieces of advice and sharing his knowledge with me,
not only in linguistics, but in all kinds of topics.
And finally thanks to my family and my friends from Bilbao, because although they are far, they
are always with me.
iii
ABSTRACT
Spanish and English rhythm and intonation have been described in detail; however, not much is
known about them in language contact situations and bilinguals. This study analyzes the
phonology and phonetics of these prosodic properties hand in hand in Mexican Spanish/English
bilinguals in Los Angeles from a sociolinguistic point of view and reveals how both prosodies
can suffer more or less noticeable attrition and transfer processes depending on the age or length
of exposure to the L2.
Two techniques were used to analyze the rhythmic properties of speech: i) nPVI (Low et al.
2000), consisting of measurements of durations of successive pairs of vowels, and ii) voicing
ratios (Dellwo et al. 2007), consisting of a Matlab function that calculates the voiced and
voiceless portions of the signal. The speech of forty nine participants in five groups was
examined: Control English Group - G1) twelve native speakers of English; Adult Early
Bilinguals - G2) eight Mexican Sp/Eng adult bilinguals who have been in L.A. since childhood;
L.A. Born Bilinguals - G3) eleven Mexican Sp/Eng young bilinguals descendants of Mexican
immigrants born in L.A.; Adult Late Bilinguals - G4) seven Mexican Sp/Eng adult bilinguals
who moved to L.A. as adults and Control Spanish Group - G5) eleven Mexican Spanish speakers
who have stayed in L.A. for less than a year. Participants read the passage “The North Wind and
the Sun” in Spanish and English. Both methodologies indicate that G1 and G2 show English-like
rhythm (more variability in vowel length and less voicing in the signal) in both languages. In
contrast, G4 and G5 groups present Spanish-like rhythm in both languages. G3s accommodate
rhythm depending on the language they produce.
The same participants produced ten neutral declarative sentences in semi-spontaneous speech in
Spanish and English for a total of 192 pre-nuclear pitch-accents and 226 nuclear pitch-accents.
These were analyzed following the Autosegmental-Metrical model of Intonational Phonology
(Pierrehumbert et al. 1988). Results indicate that for nuclear pitch-accents, H* tones were
exclusively used by G1, G2 and G3, while L*+H were only found in G3, G4 and G5. In nuclear
pitch-accents, upstepped configurations (L+¡H* L%) and sustained boundary tones (!H%)
typical of Mexican Spanish (Butragueño 2006), were found in all groups except in G1, indicating
the retention of such contours even by heritage speakers in G2 and G3. No group produced
iv
different F0 contours depending on the language. Given the tendency of Mexican Spanish to
present higher F0 values in nuclear than pre-nuclear pitch-accents, another analysis looked at the
differences in F0 between pitch-accents. Results indicate that G3, G4 and G5, contrasting with
G1, produce nuclear pitch-accents that are higher than pre-nuclear ones. Speakers in G2 show an
intermediate behavior. Finally, G2, G3, G4 and G5 produce vocatives with intonational patterns
characteristic of Mexican Spanish while G1 does not.
Overall results demonstrate that although rhythm and intonation are suprasegmental features
under the domain of prosody, they show different levels of accommodation, transfer and
attrition, indicating that they may differ in their grammatical representation.
1
CHAPTER 1: GENERAL INTRODUCTION
Lately there is a growing interest in describing the prosodic characteristics of Spanish (and
Romance languages in general) worldwide. Descriptions have generally concentrated on
intonational properties perhaps due to the salient production differences that can be seen between
varieties even to convey the same pragmatic meaning. Nevertheless, not many investigations
have studied Spanish intonation in language contact situations and bilinguals in detail (Elordieta
2003, Colantoni & Gurlekian 2004, O’Rourke 2005, Lleó et al. 2004, Alvord 2006, Robles-
Puente 2012, Miglio et al. 2013). At the same time, there has been an ongoing debate on whether
languages can be labeled as belonging to different rhythmic classes or whether such
classification is unfounded (Ramus et al. 1999, Low et al. 2000, Dellwo et al. 2007, Nava 2010).
The close inspection of two languages that a priori do not share their rhythmic properties but that
coexist within the same community like the case of Mexican Spanish and English should shed
some light on this matter. As a more general goal, the study of the suprasegmental properties of
intonation or rhythm in a language contact situation like the one found in Los Angeles can help
to determine not only whether prosodic properties can undergo change or not, but also how that
change can be motivated by a second language and sociolinguistic factors and what the final
prosodic product looks like.
The lexicon is said to be the most permeable part of language (Thomason & Kaufman 1988) and
consequently the most likely to be affected and undergo changes in language contact situations.
Various studies have demonstrated that lexical transfers are common in Spanish speaking
communities in contact with English in the U.S. due to linguistic pressures from the latter. These
include loanwords, calques and semantic extensions (Otheguy et al. 1989, Zentella 1990,
Otheguy & Garcia 1993, inter alia). Nevertheless, linguistic changes due to language contact
have also been detected at a deeper structural level of the language.
Thomason and Kaufman (1988) argue that language contact can provoke changes at any level of
the grammar (phonology, morphology and syntax) and that it is the sociolinguistic history of the
speakers, and not the structure of their language, that is the determinant of the linguistic output of
language contact. Silva-Corvalán (1994) adopts a slightly different posture and posits that the
language structure, which is constrained by cognitive processes, governs the introduction and
diffusion of new elements in the linguistic system while the sociolinguistic history of the
2
speakers determines the direction and degree of the changes. She demonstrates that this is the
case for Spanish syntax and morphology in the Mexican community of Los Angeles. Some
examples of change she mentions are the placement of clitics, the Subject-Verb order or the lack
of use of the complementizer que ‘that’. Phonological and phonetic changes are not considered
in her study, however.
Following a similar line of research, Lipski (2008) argues that transitional Spanish-English
bilinguals in Mexican-American communities present deviations from standard Spanish
grammar. These include the use of wrong nominal and adjectival inflections, incorrect
conjugations in verbs, errors in prepositional usage, incorrect use of definite and indefinite
articles or the use of redundant subject pronouns among others. The author also notices some
phonetic differences in the Spanish of these speakers: /e/ is laxed to [ɛ], a weakened
pronunciation of /j/ and /x/ and reduction of unstressed vowels. Thus, the analysis of language
change in Spanish/English contact situations generally makes reference to the lexicon, syntactic
and morphological components or the phonetic (phonological) particularities at the segmental
level of the language. Not much attention has been paid to the characteristics of Spanish in
contact with English at the prosodic level, however. This study aims to explain how the prosodic
features (rhythm and intonation) of a second language might be gradually affecting the
phonological system of the first language (Spanish). The acquisition of prosody in a second
language (English) will be studied as well.
Previous studies have demonstrated that the existence of two competing languages may result in
alterations at the syntactic level and the segmental (phonetic) level. Consequently, it is
reasonable to think that prosodic systems can be modified as well. Intonation and rhythm are
considered two suprasegmental components of prosody that play a central role in perceiving an
accent as native or foreign and also serve as a cue for listeners to distinguish languages (Ramus
2002). These two prosodic components can be said to differ in various aspects, however. First,
they inherently refer to different dimensions of prosody. On the one hand rhythm is a temporal
characteristic of language and a by-product of lengthening and reduction processes of speech
elements (due to accentuation and foot structure) or phonotactics (Dauer 1987, Low & Grabe
1995, Low 1998, Grabe et al. 1999, Low et al. 2000, Dellwo et al. 2007, Nava & Zubizarreta,
2009 and Nava 2010). On the other hand, intonation refers to structured melodic patterns linked
3
to stressed syllables and boundaries that occur over segments to systematically convey different
pragmatic meanings (Pierrehumbert & Beckman 1986, Ladd 1996, Gussenhoven 2004, Beckman
et al. 2005, inter alia). Second, rhythm and intonation differ when it comes to their pragmatic
roles. No matter how different the rhythm is (unless there is a vowel length contrast), a sentence
will still have the same meaning. However, the pragmatic meaning of the sentence can be
completely different depending on the intonational patterns or melodies used at the
suprasegmental level. With a falling intonation it would normally be a statement. With a rising
intonation it would normally be a question. Thus, rhythm and intonation may operate at the
suprasegmental level, but the importance of their roles in communication differs. Additionally,
there is a recent consensus that although it might be dependent on different foot structures or
phonotactics, rhythm needs to be considered a gradual property (more or less syllable-timed /
more or less stress-timed) rather than a discrete one, resulting in a classification of languages
along a continuum (Dauer 1983, Ramus et al. 1999, Low et al. 2000, inter alia). In contrast,
intonation and intonational patterns are normally treated as categorical units (discrete tonal
movements composed by low -L- and high -H- tones) that are used or not in a given language.
Finally, the theory of Prosodic Phonology proposes a hierarchy of domains where the
phonological patterns of rhythm and intonation can be located at different levels. According to
this framework there is a prosodic representation that is related but still independent from the
syntactic structure. The following figure (Figure 1) summarizes the hierarchy (Selkirk 1980,
1984, Nespor & Vogel 1986, Hayes 1989, Schiering et al. 2007, Prieto 2006).
1
the intonational phrase
|
the phonological phrase
|
the phonological word
|
the foot
|
the syllable
Figure 1. Hierarchy in Prosodic Phonology.
4
According to the theory, rhythm and stress happen at the lowest levels of the hierarchy and
intonational contours happen at the level of the intonational phrase (which at the same time is
composed of phonological or intermediate phrases and phonological words that can have their
own tonal movements). The domains of the foot and the syllable are essential to distinguish
English and (Mexican) Spanish rhythms (see chapter 2 for more detailed descriptions). It is
important to highlight that the temporal characteristic of rhythm and the domains of the foot and
the syllable can be influenced by boundaries at the phrase level. As will be discussed in section
2.3.2., vowels belonging to the last syllable or foot in a phrase or utterance can undergo
lengthening (see Oller 1979, Prieto et al. for Spanish). Nevertheless, clear differences between
the languages at the foot and syllable levels have been traditionally established. English has been
analyzed as a more stress-timed language where stresses happen at regular intervals. For this to
happen, stressed syllables are lengthened and unstressed syllables are reduced forming feet of
similar lengths. Spanish, a more syllable-timed language, uses the syllable as the rhythmic unit
instead and the difference between stressed and unstressed syllables is not as noticeable as in
English. Additionally, stresses do not have to be evenly distributed (Abercrombie 1967, Ramus
et al. 1999, Nava & Zubizarreta 2009, Nava 2010, inter alia). This core phonological difference
that holds between most Germanic and Romance languages can be challenged in language
contact situations and bilinguals since two competing domains or rhythmic units (foot vs.
syllable) coexist. More specifically the stress-timed properties of English could shift towards
more syllable-timed in the continuum of rhythm due to Spanish or vice versa. A third possible
outcome is that the rhythm of bilinguals is between that of English and Spanish. These
possibilities are represented in the diagram below (Figure 2):
2
Figure 2. Possible rhythmic outcomes in English/Spanish bilinguals within a continuum.
English monolingual
Stress-timed rhythm
• Vowel reductions
• Syllables of
different lengths
Spanish monolingual
Syllable-timed rhythm
• No vowel
reductions
• Syllables of similar
lengths
?
English/Spanish bilingual
5
In the case of intonational contours it is well known that Mexican Spanish has properties that
make it different even from other Spanish varieties, the use of circumflex patterns and final
sustained tones being the most salient ones (see chapter 3 for more detailed descriptions and
other differences). These contours are not part of the phonological inventory of English but, as is
the case with the rhythmic units, it is possible that bilinguals show altered prosodic productions
where Spanish tonal movements have entered the English system or vice versa. An alternative
outcome is that bilinguals end up coming up with new tonal movements not available in either
language:
Figure 3. Possible intonational outcomes in English/Spanish bilinguals.
The study of rhythm and intonation hand in hand in different generations of Spanish/English
bilinguals will determine which of the levels described in the hierarchy proposed by Prosodic
Phonology exhibit changes due to language contact given that English and Spanish belong to
different rhythmic families and that they employ slightly different tonal patterns in certain
pragmatic contexts. If any of those levels exhibit changes, the speech analysis of different
generations and levels of bilingualism will determine a) if the degree of modification is
consistent across levels (rhythm and intonation) and b) in which groups modifications are more
salient. By consistent I mean, for example, if the productions of a speaker have English-like
rhythm they would have English-like intonation as well. If the levels are treated as independent
components of prosody, different combinations would be possible. (e.g. English-like rhythm but
Spanish-like intonation).
English monolingual
English intonational contours
• e.g. low boundary
tones in declaratives
Other contours
Spanish monolingual
Spanish intonational
contours
• e.g. sustained
boundary tones in
declaratives
?
use of
English/Spanish bilingual
6
The inclusion of the sociolinguistic component is motivated by previous research, which
demonstrates that the length of time in a language contact area as well as the linguistic exposure
to the second language are correlated with speech modifications and innovations in language
contact situations (Silva-Corvalán 1994, Vann 2009). The (Mexican) Spanish community in Los
Angeles is not homogeneous due to the variety of personal histories and linguistic backgrounds.
Since the community has grown for the last fifty years, Mexican Spanish speakers can have
different linguistic profiles; From fluent Spanish / English bilinguals to quasi monolinguals in
either language. If the prosodic system is permeable, Spanish speakers that have had an extended
contact with English are more likely to show deviations from their original grammars than those
that have had reduced or null contact. Another possibility is that speakers that have been equally
exposed to both prosodies should have the ability to maintain two different grammars.
Results on the L.A. community will shed light on both the role sociolinguistic factors play in the
shaping of suprasegmental features as well as on the structure and characteristics of the prosodic
domain. Thus, this descriptive and comparative work will provide a better understanding of the
phonological and phonetic properties of Spanish and English and will also contribute to the fields
of sociolinguistics, language contact and bilingualism.
7
CHAPTER 2: RHYTHM
2.1. INTRODUCTION
The aim of this section is to provide a prosodic description of the rhythmic properties of the
Spanish and English spoken in the Los Angeles County area (California), where 47.7% of the
population has Hispanic origins (U.S. Bureau of the Census, 2010 Census). The study focuses on
the Mexican community, the most numerous and representative one (35.8% of the Hispanic
population - U.S. Bureau of the Census, 2010 Census), and tests to what extent the two
languages in contact (Spanish-English) have had an effect on each other. Thus, this project is not
only descriptive but contrastive. Given the linguistic situation in the Los Angeles area
(bilingualism, diglossia and contact of the Spanish and English languages), the investigation is
carried out from a sociolinguistic perspective and therefore various social variables are taken into
account. These include place of origin (Mexico or U.S.), generation (1
st
generation of
immigrants, 2
nd
generation, etc.), age, mother tongue and second language.
The reason to study the prosodic feature of rhythm is the different rhythmic properties of English
(stressed-timed) and Spanish (syllable-timed). Two complementary methodologies (the
Normalized Pairwise Variability Index and voiceless to voiced ratios) are used to analyze the
English and Spanish productions of five different groups: 1) English Control Group - Native
speakers of English from Los Angeles County, 2) Adult Early Bilinguals - Mexican
Spanish/English adult bilinguals who moved to Los Angeles County early in their childhood and
were raised there, 3) L.A. Born Bilinguals - Mexican Spanish/English young bilinguals
descendants of Mexican immigrants who were born and raised in Los Angeles County, 4) Adult
Late Bilinguals - Mexican Spanish/English bilinguals who moved to Los Angeles County when
they were fifteen or older and 5) Spanish Control Group - Mexican Spanish speakers who have
been in Los Angeles County or the U.S. for a short period of time or have never been to the U.S.
3
Results reveal that the rhythm in English of the English Control Group, the Adult Early
Bilinguals and the L.A. Born Bilinguals is more stressed-timed than that of the Adult Late
Bilinguals or the Spanish Control Group. In Spanish, the English Control Group or the Adult
Early Bilinguals show less syllable-timed rhythms than the L.A. Born Bilinguals, the Adult Late
8
Bilinguals and the Spanish Control Group. These results suggest the prosodic feature of rhythm
can be altered due to language contact.
This chapter is divided as follows: In Section 2 the main methodologies used in previous studies
to quantify rhythm and classify languages according to this prosodic feature are described. In
Section 3, I present the population under study (2.3.1.), the two different methodologies used to
quantify rhythm together with their results in Spanish and English (2.3.2.) and the role vowel
durations play in shaping rhythm (2.3.3.). Section 4 summarizes all the findings reported in the
previous sections to reach a conclusion, raise future questions and integrate the current results in
the literature of rhythm.
2.2. BACKGROUND: QUANTIFYING RHYTHM
Rhythm in linguistics has been defined as the systematic occurrence of strong / stressed / heavy /
prominent and weak / unstressed / light / less prominent speech elements (e.g. vowels or
syllables) over time. Early work on rhythm (Pike 1945, Abercrombie 1967) establishes a dual
classification of languages: stress-timed or syllable-timed. While syllable-timed languages would
have syllables of similar lengths (i.e. isochronous syllable durations), stress-timed languages
would exhibit more variation in syllable length so that stresses occur at regular intervals (i.e.
isochronous durations between stresses). According to this first classification, Spanish and other
languages belonging to the Romance family would be syllable-timed, while English and other
Germanic languages would be stress-timed. Although the two extremes of the continuum remain
somehow unaltered, there have been several attempts to refine this classification since isochrony
is not supported by acoustic evidence (e.g. Roach 1982, Beckman 1992).
Dasher and Bolinger (1982) posit that rhythm is not a phonological primitive, but a by-product of
the phonological characteristics of a language. Similarly, Dauer (1983) suggests that the syllable-
timed vs. stress-timed dichotomy is a simplification, and that languages can be somewhere in
between depending on phonotactics, vowel reductions or the phonetic realization of stress.
Ramus and colleagues (1999) use the term ‘intermediate language’, a concept that captures that
9
the original dichotomy should give way to a continuum where languages can be more or less
stress-timed and more or less syllable-timed.
In order to organize languages in the continuum and quantify rhythmic differences, several
techniques have been developed. Ramus and colleagues (1999) analyze eight languages (English,
Dutch, Polish, French, Spanish, Italian, Catalan, Japanese) by taking three phonetic variables of
the speech signal into account: 1) the proportion of the duration of a sentence composed by
vowels (%V), 2) the standard deviation of the vocalic intervals (∆V) and 3) the standard
deviation of the durations of the consonantal intervals (∆C). Their conclusion is that the variety
of syllable structures accounts for rhythmic differences and that therefore 1) and 3) are the most
reliable acoustic correlates to organize languages in the continuum of rhythm. They argue that
since Germanic languages have more syllable types and therefore more variability in the number
of consonants and higher consonant/vowel ratios (i.e. less %V) than Romance languages,
English and Spanish are rather distant as exemplified in Figure 4 (adapted from Ramus et al.
1999). Based on this contrast, Japanese, a mora-type language, would be even more distant from
English due to the simpler syllable structures.
Figure 4. Percentage of vowels (%V) and SD of consonantal intervals (∆C).
Low and Grabe (1995) propose the Pairwise Variability Index (PVI), a technique that can
account for rhythmic differences by comparing the length of successive vocalic and intervocalic
(or consonantal) intervals. Low and colleagues (2000) present a slightly different version and
claim that it is a better method to quantify rhythm than the one presented by Ramus and
0.03
0.04
0.05
0.06
35 40 45 50 55
∆C
%V
Spanish
English
Japanese
10
colleagues (1999) since speaking rate is taken into account. It is the so-called normalized
Pairwise Variability Index. To get the final score (nPVI), the absolute value of the difference in
duration of two successive intervals is divided by the average duration of the pair. Then, all the
resulting values of an utterance are averaged and multiplied by 100 to avoid fractional values.
This technique was used to confirm perceptual differences and successfully quantify that
Singapore English is more syllable-timed than British English.
Dellwo and colleagues (2007) present a different approach to study rhythm. They depart from
measuring vowels or intervocalic intervals and argue that the duration of voiced and voiceless
intervals in speech is a better indicator of rhythmicity. According to them, higher voiceless ratios
in the acoustic signal would be characteristic of stress-timed languages, while lower voiceless
ratios would be found in syllable-timed languages. Although they do not include Spanish in their
study, they find that English (and German) shows shorter voiced intervals than Romance
languages like Italian or French.
All these studies and techniques opened the door to the study of rhythm in situations where
something more than a simple stress-timed vs. syllable-timed distinction is necessary between
languages. These include language contact situations and second language learning, among
others. O’Rourke (2008) studies two different varieties of Peruvian Spanish (Lima and Cuzco
Spanish) following the methodology introduced by Ramus and colleagues (1999). Although
studies like the ones mentioned before could cause us to expect both varieties to be categorized
as syllable-timed and the rhythm of Quechua has not been studied in detail, O’Rourke considers
the possibility that bilingual Quechua-Spanish speakers from Cuzco may show rhythmic
properties different from those of monolingual Spanish speakers from Cuzco and Lima. She
found that Spanish monolingual and Quechua-Spanish bilingual speakers from Cuzco showed
similar results no matter what their language background was but they presented significantly
lower %V and higher ∆C durational values than those of the Lima variety presumably due to the
weakening of codas in Lima Spanish and vowel reductions in Cuzco Spanish. More interestingly,
both varieties fall outside the stress-timed – syllable-timed continuum and closer to mora-timed
languages (i.e. closer to Japanese in Figure 4) due to their high percentage of vocalic sequences
and low consonantal variation. This finding, together with an identical one in European and
11
Brazilian Portuguese (Frota & Vigário 2001), lead O’Rourke to point out the necessity to study
Spanish and Romance languages in other contact situations to see to what extent they should still
be located towards the syllable-timed side of the continuum.
Nava and Zubizarreta (2009) and Nava (2010) notice that vowel reduction is common in English
but nonexistent in Spanish and that the difference in duration between stressed and unstressed
vowels is greater in English than Spanish. Finally, syllable structures in English are more varied
than in Spanish. This phonotactic contrast results in differences in voicing and the variability of
voiceless intervals. They choose the measurement technique proposed by Dellwo and colleagues
(2007) to test if rhythm is the direct result of the aforesaid differences. In their study they
included thirty adult native speakers of English, twenty Spanish monolinguals and forty-five
adult L1Spanish/L2 English speakers. Ages varied between nineteen and fifty-five and dialectal
variations were not considered. Native speakers of English showed the highest standard
deviations of voiceless intervals and the least voicing in English while Spanish monolinguals
showed the highest voicing ratios and the least variability in voicing in Spanish. Their results
show that second language learners of English with Spanish as their mother tongue had rhythmic
values in English between those of monolingual speakers of English and Spanish. Similarly,
Carter (2005) compares the rhythm of the English spoken by the Hispanic community in North
Carolina to that spoken by English monolinguals by using the nPVI technique. His results show
that the rhythm of Hispanic English is not as syllable-timed as that of Spanish, but not as stress-
timed as the English of English monolinguals either.
Thus, studies on the rhythmic properties of Spanish and English are not scarce; however, as
O’Rourke (2008) or Frota and Vigário (2001) demonstrate, it should not be assumed that all the
varieties of the same language can be always categorized under the same label. From previous
studies it is also clear that the classification of languages along a continuum is the result of an
effect of various phonological structural properties on duration (phonotactics, lengthening and
reduction processes, etc.). New techniques provide researchers with multiple tools to get more
detailed measurements, allowing more refined descriptions and the identification of those
phonological structures. In this study, two complementary techniques are used: the nPVI applied
to vowels and the measurement of voiceless and voiced portions of the acoustic signal. The
12
former was chosen in order to control for potential differences in reading rates and to compare
the current results to those in the bilingual community of South Carolina reported in Carter
(2005). The latter was included to look for rhythmic similarities and differences between native
speakers and L1 Spanish/L2 English speakers presented in Nava and Zubizarreta (2009) and
Nava (2010) and the different groups in the Los Angeles community.
2.3. SPANISH AND ENGLISH RHYTHM IN LOS ANGELES
The study concentrates on the Spanish and English spoken in Los Angeles County due to the
unique linguistic situation the area offers. The Mexican community has members with various
language backgrounds and different amounts of exposure to English depending on where they
were born, when they moved to the area, etc. Given that Spanish and English are languages of
different rhythmic nature, the analysis of the productions of these speakers will unveil how these
languages may have influenced each other’s rhythmic properties to a greater or lesser extent
depending on sociolinguistic factors.
2.3.1 PARTICIPANTS
According to the U.S. Bureau of Census 2010, there are 4,687,889 persons of Hispanic or Latino
origin in Los Angeles County (47.7%). The largest Spanish-speaking population in the county is
the Mexican one (35.8%) and that’s the reason why this community is the focus of the current
study. A valuable characteristic of the Mexican community is that it has been formed over the
last fifty years. According to the U.S. Bureau of Census 2000, 486,147 Mexican citizens moved
to Los Angeles County before 1980, 489,047 between 1980 and 1989 and 549,963 between 1990
and 2000. Most of the descendants of these immigrants have been born in the area too and
therefore different generations and speakers with different language backgrounds coexist. The
study of the Spanish produced by these speakers will determine to what extent and how the
prosodic properties of this Romance language have been influenced by English. At the same
time, the English of the same speakers is analyzed to a) compare it to their Spanish productions
and b) to look for patterns considering sociolinguistic variables.
13
Labov (1966) is one of the first linguists that tries to demonstrate that variability is part of
linguistic systems. It is expected that language variation will take place even between individuals
that have very similar linguistic backgrounds because each person talks in a particular way
(idiolect), with his/her own vocabulary and expressions, grammar or pronunciation. Of course,
all these particularities are likely to meet at least the minimum requirements that are necessary to
communicate in the speech community the speaker belongs to. A very exhaustive linguistic
analysis could consider each speaker individually but the role of sociolinguistics is precisely to
capture linguistic behaviors that are common to a group normally defined by certain social
factors (age, sex, origin, class, etc.). By grouping speakers that share specific extralinguistic
characteristics as opposed to others, clear patterns should emerge in the use of sociolinguistic
variants given that the extralinguistic grouping that has been selected is relevant. Thus, selecting
appropriate ways to group speakers is crucial to identify linguistic trends.
In the present study, and since the period and the years of exposure to English are a priori the
most determining factors to shape the rhythmic properties of Spanish and English in the L.A.
community, the following groups are considered:
4
- English Control Group (Non-Mexican native speakers of English from Los Angeles) –
Twelve native speakers of English from Los Angeles County or nearby areas.
- Adult Early Bilinguals (Mexican immigrants at childhood) - Eight Mexican
Spanish/English bilinguals (adults) who were born in Los Angeles County or moved
there early in their childhood and were raised there.
- L.A. Born Bilinguals (American children and adolescents descendants of Mexican
immigrants) - Eleven Mexican Spanish/English bilinguals (children/adolescents)
descendants of Mexican immigrants who were born and raised in Los Angeles County.
5
- Adult Late Bilinguals (Mexican immigrants at adulthood) - Seven Mexican
Spanish/English bilinguals (adults) who moved to Los Angeles County when they were
fifteen or older.
- Spanish Control Group (Mexican Spanish speakers with little or null influence of
English) – Eleven Mexican Spanish speakers (adults) who have been in Los Angeles
County or the U.S. for a short period of time or have never been to the U.S.
6
14
One problem in the elaboration of the groups is that since they are necessarily composed by
several individuals, they cannot always be 100% homogeneous (Silva-Corvalán 2001). Most of
the participants belonged to what can be considered a middle class but since different social
histories may have an impact in the results, all subjects were asked to report the percentage of
time they use(d) Spanish and English at home, with friends, at school (when applicable) and at
work (when applicable) in the questionnaires in different periods of their lives (1-14, 15-25, 26-
50 and +50). They were also requested to point out any relevant information regarding their
language backgrounds (i.e. going to a bilingual school, studying abroad, etc.), when they started
using each of the languages and what languages their parents spoke to make sure the grouping
was as homogenous as possible. This information is collected in Appendix A.
7
The first and last groups serve as control groups of English and Mexican Spanish respectively.
8
In the English Control Group, four of the speakers had high proficiency in Spanish while eight
had just basic knowledge from high school and should be considered English monolinguals. The
Spanish Control Group was formed by two different subgroups. The first subgroup was formed
by three members of a Mexican family who were visiting a relative in Los Angeles for a few
days and four other subjects who live in Mexico.
9
All these subjects reported that they had little
command of English and that they had had null or very little exposure to American English
before. Therefore they could be considered representatives of Mexican Spanish without
significant L2 influences.
10
The second subgroup is composed by four Mexican students
studying in Los Angeles. They had been in the area just for a maximum of two years so they can
still be considered good representatives of speakers of Spanish with little or null influence of
English as well. Given these small differences between subgroups, potential differences were
considered at the time of the analyses; however, all subjects behaved similarly.
The other three groups are of great relevance since they are formed by native speakers of Spanish
that have moved to Los Angeles County at different stages in their lives. This means that these
speakers have had different degrees of exposure to English at different ages but all of them are
Spanish-English bilinguals. The Adult Early Bilinguals Group is similar to the L.A. Born
Bilinguals in that subjects were raised in California from an early age (average of 3.3 years and
SD of 2.5 for the Adult Early Bilinguals and 0 for the L.A. Born Bilinguals) and different in that
15
the Adult Early Bilinguals are already adults (34.3 years old on average with SD of 12.2) while
the L.A. Born Bilinguals are not (16.6 years old on average with a SD of 3.6). This can be
considered “a sort of apparent time” in Silva-Corvalán’s (1994) terminology which in this case
results in different amounts of exposure to English. The exposure of the Adult Early Bilinguals
has been longer than that of the L.A. Born Bilinguals since the former has lived in L.A. for a
longer time (30.8 years on average with a SD of 12.5 vs. 16.6 years on average with a SD of 3.6
for the L.A. Born Bilinguals). The main difference between the Adult Early Bilinguals and the
L.A. Born Bilinguals when it comes to the % of the languages used is found in the first years.
The L.A. Born Bilinguals show more use of English than the Adult Early Bilinguals (always
over 50% with an average of 75.2% and a SD of 14.8% vs. 54.6% and a SD of 23.6%) because
they were born in L.A. In most cases, the use of English in both groups increases over time. All
subjects in the Adult Early Bilinguals Group considered Spanish their mother tongue since they
were exposed to English and started speaking the language only when they moved to L.A.
Interestingly, two of the speakers in the Adult Early Bilinguals Group called themselves
“pochos” because they considered their Spanish bad. They reported that they used to speak it
fluently but that with time English became the language they used the most. This captures the
trend that seems to be taking place with young speakers. Although Spanish is the home and
family language, there is an increasing pressure of acquiring English (and using Spanish less)
over time. In the L.A. Born Bilinguals Group both Spanish and English were considered the
mother tongues of most of the speakers. Although all subjects in the L.A. Born Bilinguals Group
had at least one parent in the Adult Late Bilinguals Group and the use of Spanish was high at
home, they reported that English was also used very frequently. Some of the parents in the Adult
Late Bilinguals Group with children in the L.A. Born Bilinguals Group reported that although
they wanted their children to learn and retain Spanish, they knew learning English was important
growing up in L.A. in order to study or work.
The Adult Late Bilinguals Group is composed by adults that moved to the Los Angeles area
when they were 15 or older (21.7 years old on average upon arrival with a SD of 3 and 43.7
years old on average with a SD of 6.4 at the time of the recording). The amount of exposure to
English has been long as well (average of 22 years with a SD of 5.1) but the exposure started
much later than in the Adult Early Bilinguals or the L.A. Born Bilinguals. Thus, the different
16
backgrounds of the three groups provide the necessary combinations to test how the length of
exposure and the time of exposure to English may have influenced Spanish rhythm and the
ability to produce native-like English in the Mexican community. If the age at the time of
exposure affects the ability to acquire English rhythm, then the Adult Late Bilinguals should
have a less stressed-timed rhythm than the Adult Early Bilinguals or the L.A. Born Bilinguals.
Also, if the length of contact with English plays a role in shaping Spanish rhythm, the Adult
Early Bilinguals Group should show Spanish productions that are less syllable-timed than those
by the L.A. Born Bilinguals and the Adult Late Bilinguals. Two different methodologies were
used to test these hypotheses.
2.3.2 METHODOLOGY
Participants were recorded reading the passage “The North Wind and the Sun” in its English and
Spanish versions (“El Viento Norte y el Sol”) unless they were monolinguals, in which case they
read the passage only in the language they speak.
11
They were given a few minutes to get
familiarized with the texts and ask questions (e.g. meanings of words). Then they were instructed
to read the texts at a normal pace first in Spanish and then in English.
The reason why readings were chosen instead of natural speech productions was to control for
the words and sentences produced. In fact, these readings are widely used in phonetic studies
(e.g. Grabe & Low 2002, Nava & Zubizarreta 2009, Nava 2010, inter alia). By reading the same
passage, measurements of the same syllables in the same positions could be taken, ensuring that
any differences in the results came from the actual rhythmic properties and not the elicitation
materials or other factors like lemma frequencies.
12
As discussed before, there is no general
consensus on what the best technique to quantify rhythm is. Thus, two main approaches were
used to get complementary results.
VOICING RATIOS
First, following Dellwo and colleagues (2007), Nava and Zubizarreta (2009) and Nava (2010)
the voiceless and voiced portions of the acoustic signal were analyzed. In order to do so, the
17
software voicebox (vocDetect – custom-adapted) (Brookes 2003) was used. The software was
also customized in such a way that calculations took pauses into account and separated them
from actual voiceless sequences. As in Nava (2010), any voiceless speech stream over 250ms
was considered a pause. Every acoustic signal was visually inspected to make sure the program
was dividing the voiceless and voiced intervals correctly.
13
Two different values were calculated for each speaker: the ratio of the sum of all voiceless
intervals over the sum of all voiced intervals (excluding pauses) and the standard deviation of the
duration of voiceless sequences (excluding pauses). Based on the phonotactics of both languages
(English has a more varied cluster size) and previous studies like Dellwo and colleagues (2007),
Nava and Zubizarreta (2009) and Nava (2010), it is expected that native English productions will
show high voiceless to voiced ratios (i.e. less voiced intervals) and more variability in
voicelessness overall. In contrast, Spanish productions with no English influence should be more
voiced overall and voiceless sequences should show less variability (Figure 5).
Figure 5. Voicing ratios and variability in English and Spanish according to previous studies.
Voiceless/voiced ratio
SD of voiceless intervals
-
+
+
-
English
Spanish
18
NORMALIZED PAIRWISE VARIABILITY INDEX
The alternative method employed to measure rhythm was the so-called Normalized Pairwise
Variability Index (nPVI). As pointed out in Grabe and Low (2002), the nPVI can be applied not
only to whole syllables but to vowels (and intervocalic intervals) as well (Low & Grabe 1995,
Low 1998, Grabe et al. 1999, Low et al. 2000). I follow Deterding (2001) and Carter (2005) and
compare successive pairs of vowels in order to get comparable results with the latter study. Note
that this method is different from the one described and used in the previous section, as the
division between the voiced and voiceless parts of the signal implied treating certain consonants
differently from others. For example /m/ would be part of the voiced stream like any vowel, but
/p/ would be included in the voiceless part. With the nPVI technique only a few non-vowels were
considered for the analysis for the reasons explained below.
In order to separate vowels from consonants or vowels in other syllables, waveforms
(periodicity) and wide band spectrograms were inspected with the speech analysis tool Praat
(Boersma & Weenink 2009). The accuracy of the divisions was also confirmed by listening to
the audio recordings. The following decisions were taken for the measurements: English and
Spanish diphthongs (e.g. cloak [kloʊk]; viento [bien.to] – ‘wind’) were measured as one vowel.
Vowels in hiatus (e.g. sería [se.ɾi.a] – ‘it would be’) were considered two vowels. However,
when phonetic reduction was noticeable they were treated as diphthongs (see Aguilar 1995 for a
detailed explanation of this phenomenon in Spanish). When two vowels belonging to different
words were in contact, (e.g. la empresa – ‘the attempt’) the division depended on whether they
were perceived as a diphthong ([laem.pɾe.sa]) or as separate vowels ([la.em.pɾe.sa]) in both
languages. In the latter case two measurements were made; in the former case only one, but no
zero value was given to the missing syllable (as in Carter 2005). Liquids in a coda were difficult
to separate from the preceding vowels in English and were included in the measurement (e.g.
north [nɔrθ]). Glides were also considered part of the vowel (e.g. wind [wɪnd]). Finally, although
they were not too common, syllabic /n/s were considered vocalic portions as well. In the rest of
CV and VC sequences the onset and offset of the vowel was determined by looking at the
beginning and end of formants bars and the periodicity of the waveform; in cases with formant
transitions the division was made in the middle of the transition. Vowels belonging to the last
syllable or foot that were located before a pause or a hesitation were measured, but were not
19
included in the analysis in order to avoid lengthening effects (Prieto et al. 2010). These normally
corresponded to words at the end of intonational or intermediate phrases: stronger, cloak, off,
other, could, blew, him, attempt, warmly, cloak and two in English and fuerte, capa, capa,
poderoso, furia, soplaba, viajero, empresa, ardor, viajero and sol in Spanish. Examples of
creaky voice were scarce and were always located at the end of utterances and therefore were not
analyzed either. In the cases where participants substituted words or introduced new words (e.g.
“obligated” instead of “obliged” or “el más poderoso” instead of “más poderoso”) or where
the stressed of the word was erroneously misplaced (e.g. [aˈθeɾto] instead of [aθeɾˈto]), all words
were analyzed.
After applying the previous decisions to all the productions, a total of 5524 pairs of syllables
were compared in English (1450 for the English Control Group, 935 for the Adult Early
Bilinguals, 1306 for the L.A. Born Bilinguals, 780 for the Adult Late Bilinguals and 1053 for the
Spanish Control Group) and 6397 pairs in Spanish (1096 for the English Control Group, 1135 for
the Adult Early Bilinguals, 1525 for the L.A. Born Bilinguals, 1035 for the Adult Late Bilinguals
and 1606 for the Spanish Control Group).
14
nPVI values were calculated by dividing the absolute
difference in duration of the pair of vowels in successive syllables by the average of the total
duration of the pair and then multiplying the resulting value by 100:
nPVI = (abs (Vowel A - Vowel B) / ((Vowel A + Vowel B) / 2)) x 100
Based on previous studies on rhythm it would be reasonable to expect that English will show
higher nPVI scores than Spanish.
15
RESULTS
VOICING RATIOS RESULTS
Figure 6 shows the average voiceless to voiced ratios with 95% confidence interval bars for the
five groups in both languages.
16
A one-way analysis of variance (ANOVA) determined that there
was a statistically significant difference between groups in English (F (4,38) = 14.452, p < .001).
Post-hoc Scheffe tests revealed significant differences between the English Control Group and
the Adult Late Bilinguals (p = .003), the English Control Group and the Spanish Control Group
(p < .001), the Early Adult Bilinguals and the Spanish Control Group (p = .002) and the L.A.
20
Born Bilinguals and the Spanish Control Group (p < .001). A significant difference was also
found between groups in Spanish (F (4,36) = 9.114, p < .001). Post-hoc Scheffe tests revealed
that the significant differences were between the Control groups (p < .001), the Adult Early
Bilinguals and the Spanish Control Group (p < .001) and the L.A. Born Bilinguals and the
Spanish Control Group (p = .042).
Figure 6. Voicing ratio averages per group in Spanish and English.
As expected, the English Control Group shows the highest ratio in English (0.45) and the
Spanish Control Group the lowest in English and Spanish (0.27 and 0.26 respectively).
Interestingly, the Adult Early Bilinguals have almost identical ratios in English and Spanish
(0.40 vs. 0.41). These are not statistically different from those of the English Control Group
(0.45 for English and 0.40 for Spanish). Although the ratio for English for the Adult Late
Bilinguals is similar to that of the Spanish Control Group (0.31 and 0.27 respectively) and the
Spanish ratio is considerably higher in the former (0.35 vs. 0.26) these two groups did not show
statistical differences in either language. Finally, the L.A. Born Bilinguals show an English ratio
that was not statistically different from those of the English Control Group and the Adult Early
Bilinguals (0.41 vs. 0.45 and 0.40). The Spanish ratio was identical to the one found in the Adult
Late Bilinguals (0.35 vs. 0.35) but statistically different from that in the Spanish Control Group
(0.35 vs. 0.26). It is of great relevance that the L.A. Born Bilinguals is the only group where
0.45
0.40
0.41
0.31
0.27
0.40
0.41
0.35 0.35
0.26
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Group 1 Group 2 Group 3 Group 4 Group 5
Voiceless to voiced ratio
English
Spanish
Control
English
Control
Spanish
L.A. Born
Bilinguals
Adult Early
Bilinguals
Adult Late
Bilinguals
21
confidence intervals do not overlap and rhythmic scores are similar to those by the control
groups, demonstrating the ability of young bilinguals to adapt their speech to the rhythmic
characteristics of each language. A Wilcoxon signed-rank test confirmed this observation
revealing that the English Control Group (p = .018) and the L.A. Born Bilinguals (p = .007)
show statistically significant different ratios in Spanish and English.
Figure 7. Average standard deviations of voiceless intervals per group in Spanish and English.
Figure 7 shows the average standard deviations with 95% confidence interval bars for the five
groups in both languages. A one-way ANOVA detected that there was a statistically significant
difference between groups in English for the SD of voiceless intervals (F (4,38) = 3.489, p =
.016) but post-hoc comparisons revealed that the only difference was between the L.A. Born
Bilinguals and the Spanish Control Group (p = .026) due to the lower variability in the latter.
Another one-way ANOVA did not detect a group effect in the standard deviations of voiceless
intervals for Spanish. Thus, it seems that the voicing ratios are a more reliable way of detecting
differences within languages.
Using the voiceless and voiced portions of the acoustic signal as indicators of rhythm seems to
be effective to distinguish at least extreme groups both in English and Spanish. Results also
partially resemble those reported in Nava (2010) using the same methodology, since L2 English
53.4
55.0
48.7
56.6
59.0
52.8
49.6
45.2
47.6
46.3
0
10
20
30
40
50
60
70
Control
English
Adult Early
Bilinguals
L.A. Born
Bilinguals
Adult Late
Bilinguals
Control
Spanish
SD of voiceless intervals
SD English
SD Spanish
22
speakers’ English productions, pattern between L1 English and L1 Spanish. Figure 8 shows the
similarities between studies.
17
Figure 8. Voicing ratios and variability in Spanish and English in the current study (left) vs.
Voicing percentages and variability in Spanish and English according to Nava 2010 (right).
Table 1 summarizes where significant differences between groups were found in both languages
in the current study:
Table 1. Significant differences (rhythm-voicing) between groups in English and Spanish.
Significant differences in English concentrated in the voicing ratios. The English Control Group,
the Adult Early Bilinguals and the L.A. Born Bilinguals are different from the Spanish Control
40
45
50
55
60
0.2 0.3 0.4 0.5
SD of Voiceless Intervals
Voicing ratio
Control English - En
Adult Late Bilinguals - En
Control Spanish - Sp
0
0.004
0.008
0.012
0.016
0.02
65 70 75 80 85
SD of Voiceless Intervals
% Voicing
English
L2 English
Spanish
Significant
differences
(voicing)
Control English Adult Early Bilinguals L.A. Born Bilinguals Adult Late
Bilinguals
Control
Spanish
Control
English
Adult Early
Bilinguals
L.A. Born
Bilinguals
Adult Late
Bilinguals
Voiceless/Voiced En.
Control
Spanish
Voiceless/Voiced En.
Voiceless/Voiced Sp.
Voiceless/Voiced En.
Voiceless/Voiced Sp.
Voiceless/Voiced En.
SD in voicing En.
Voiceless/Voiced Sp.
23
Group in this indicator revealing that due to the lack of exposure to the target language, the
Spanish Control Group fails to produce a stress-timed rhythm characteristic of English. The fact
that the same indicator is significantly different in the English Control Group and the Adult Late
Bilinguals suggests that speakers of Mexican Spanish who became exposed to American English
in their adulthood, may lack that ability as well as opposed to the Adult Early Bilinguals and the
L.A. Born Bilinguals. Averages presented in Figure 6 suggest that the L.A. Born Bilinguals and
the Adult Late Bilinguals have Spanish rhythms more similar to the Spanish Control Group than
the English Control Group and the Adult Early Bilinguals do, but significant differences in
voiceless to voiced ratios were found between the English Control Group, the Adult Early
Bilinguals and the L.A Born Bilinguals and the Spanish Control Group. The Normalized
pairwise Variability Index was used as an alternative method to validate these results.
NPVI RESULTS
Figure 9. nPVI averages per group in Spanish and English.
Figure 9 shows the average nPVI values in English and Spanish for the five groups with 95%
confidence interval bars. A one-way ANOVA indicated a significant effect of group in English
(F (4,42) = 13.777, p < .001) and post-hoc Scheffe tests revealed significant differences between
the English Control Group and the Adult Late Bilinguals (p < .001), the English Control Group
and the Spanish Control Group (p < .001), the Adult Early Bilinguals and the Adult Late
53.3
54.5
52.2
39.7
41.3
44.1
41.5
34.2
32.2 32.0
0
10
20
30
40
50
60
Group 1 Group 2 Group 3 Group 4 Group 5
nPVI
English
Spanish
Control
English
Control
Spanish
Adult Late
Bilinguals
L.A. Born
Bilinguals
Adult Early
Bilinguals
24
Bilinguals (p < .001), the Adult Early Bilinguals and the Spanish Control Group (p < .001), the
L.A. Born Bilinguals and the Adult Late Bilinguals (p = .002) and the L.A. Born Bilinguals and
the Spanish Control Group (p = .005). A one-way ANOVA was also run to compare the nPVI
values in Spanish and it revealed a significant effect of group (F (4,40) = 24.535, p < .001). Post-
hoc Scheffe tests indicated significant differences between the English Control Group and the
L.A. Born Bilinguals (p < .001), the English Control Group and the Adult Late Bilinguals (p <
.001), the English Control Group and the Spanish Control Group (p < .001), the Adult Early
Bilinguals and the L.A. Born Bilinguals (p < .001), the Adult Early Bilinguals and the Adult Late
Bilinguals (p < .001) and the Adult Early Bilinguals and the Spanish Control Group (p < .001).
As expected, results show that the English Control Group has the highest nPVI values in English
(53.3) and Spanish (44.1). The Adult Early Bilinguals Group has values that are not statistically
different from those (54.5 in English and 41.5 in Spanish). The L.A. Born Bilinguals present
high nPVI results in English as well, but not as high as those by the English Control Group or the
Adult Early Bilinguals (52.2). Significant differences between these groups were not found,
however. Interestingly, the L.A. Born Bilinguals show nPVI scores similar to the Adult Late
Bilinguals in Spanish (34.2 vs. 32.2) but a much higher one in English (52.2 vs. 39.7). Finally,
the Spanish Control Group has the lowest nPVI result for Spanish (32) and the second lowest one
for English (41.3). A Wilcoxon signed-rank test revealed that all groups show statistically
significant different nPVI ratios in Spanish and English (.012 for the English Control Group;
.011 for the Adult Early Bilinguals; .003 for the L.A. Born Bilinguals; .046 for the Adult Late
Bilinguals and .007 for the Spanish Control Group).
Table 2. Significant differences in nPVI scores between groups in Spanish and English.
Overall nPVI results suggest that the English Control Group, the Adult Early Bilinguals and the
L.A. Born Bilinguals present similar rhythmic properties in English as opposed to the Adult Late
Significant
Differences (nPVI)
Control English Adult Early Bilinguals L.A. Born
Bilinguals
Adult Late
Bilinguals
Control
Spanish
Control English
Adult Early Bilinguals
L.A. Born Bilinguals nPVI Sp. nPVI Sp.
Adult Late Bilinguals nPVI En.
nPVI Sp.
nPVI En.
nPVI Sp.
nPVI En.
Control Spanish nPVI En.
nPVI Sp.
nPVI En.
nPVI Sp.
nPVI En.
25
Bilinguals and the Spanish Control Group. In Spanish, the L.A. Born Bilinguals show average
nPVI scores closer to the Adult Late Bilinguals and the Spanish Control Group and distant from
those by the English Control Group and the Adult Early Bilinguals. Individual generational
analyses within each family support these findings and provide valuable information since they
capture how gradual changes in rhythm may be happening without relying on the group level.
The best example of generational change is embodied by g5s7, g4s3 and g3s8. These are three
subjects that belong to the same family (grandmother – Spanish Control Group, father – Adult
Late Bilingual, and daughter – L.A. Born Bilingual) and to three different groups as represented
in Figure 10:
Figure 10. nPVI values for family 1 (Control Spanish – Adult Late Bilingual – L.A. Born
Bilingual).
nPVI values for English increase from generation to generation (32, 45, 53). nPVI values for
Spanish remain similar from the Spanish Control to the Adult Late Bilingual (notice from
Appendix A that g4s3 has one of the lowest nPVI in Spanish) and increase slightly in the L.A.
Born Bilingual. This increase is not a noticeable as in English, however, indicating that Spanish
rhythm is being retained (or at least is far from being English-like). Nevertheless, it could be
argued that the higher nPVI values in the last generation are the result of the quality of the
Spanish and English of the mother, who belongs to the English Control Group (g1s10 – ratio of
49 in English and of 41 in Spanish). A similar trend is found between g4s1 and g3s3 (Figure 11).
The father (not recorded) would belong to the English Control Group and that could explain why
the English of the son is higher than that of the mother.
32
45
53
31
28
36
0
10
20
30
40
50
60
g5s7 (grandmother) g4s3 (father) g3s8 (daughter)
English
Spanish
26
Figure 11. nPVI values for family 2. (Adult Late Bilingual – L.A. Born Bilingual).
Nevertheless, other subjects show evidence that the very same trend still happens when none of
the parents are part of the English Control Group or the Adult Early Bilinguals (Figure 12 and
Figure 13):
Figure 12. nPVI values in family 3. (Adult Late Bilingual – Adult Late Bilingual – L.A. Born
Bilingual).
Figure 13. nPVI values in family 4. (Adult Late Bilingual – L.A. Born Bilingual).
Both in Figures 12 and 13, children show scores in Spanish similar to those by the parents. In
contrast, their nPVI scores in English are higher even when all the parents belong to the Adult
Late Bilinguals Group, showing the same trend as in Figures 10 and 11 and overall group results.
35
51
36
34
0
20
40
60
g4s1 (mother) g3s3 (son)
English
Spanish
43
36
53
30
36 36
0
20
40
60
g4s4 (mother) g4s5 (father) g3s7 (son)
English
Spanish
38
46
36
33
0
10
20
30
40
50
g4s6 (father) g3s6 (daughter)
English
Spanish
27
VOICING RATIOS VS. NPVI: COMPARING RESULTS
In section 3.2, I have presented the rhythmic similarities and differences between the five groups
under study in English and Spanish. The following charts summarize where significant
differences were found with each indicator of rhythm (Voiceless/voiced ratios, SD in voicing,
and nPVI) in each language:
Table 3. Significant differences (rhythm-all) between groups in English.
For English, the nPVI and the voiceless to voiced ratio differentiate between the rhythm of the
Spanish Control Group and those of the three groups that were exposed to English since their
childhood (the English Control Group, the Adult Early Bilinguals and the L.A. Born Bilinguals).
The same indicators also differentiate between the rhythms of the English Control Group and the
Adult Late Bilinguals but only the nPVI detects significant differences between the Adult Early
Bilinguals and the Adult Late Bilinguals and the L.A. Born Bilinguals and the Adult Late
Bilinguals. The standard deviation in voicing also detected differences between the L.A. Born
Bilinguals and the Spanish Control Group.
Table 4. Significant differences (rhythm-all) between groups in Spanish.
Significant
Differences
in English
Control English Adult Early Bilinguals L.A. Born
Bilinguals
Adult Late
Bilinguals
Control
Spanish
Control English
Adult Early
Bilinguals
L.A. Born Bilinguals
Adult Late
Bilinguals
nPVI
Voiceless/Voiced
nPVI nPVI
Control Spanish nPVI
Voiceless/Voiced
nPVI
Voiceless/Voiced
nPVI
Voiceless/Voiced
SD in voicing
Significant
Differences
in Spanish
Control English Adult Early Bilinguals L.A. Born Bilinguals Adult Late
Bilinguals
Control
Spanish
Control English
Adult Early
Bilinguals
L.A. Born
Bilinguals
nPVI nPVI
Adult Late
Bilinguals
nPVI nPVI
Control Spanish nPVI
Voiceless/Voiced
nPVI
Voiceless/Voiced
Voiceless/Voiced
28
In Spanish, the nPVI and the voiceless to voiced ratio revealed significant differences between
the English Control Group and the Spanish Control Group and the Adult Early Bilinguals and the
Spanish Control Group. The nPVI also detected significant differences between the English
Control Group and the Adult Early Bilinguals vs. the L.A. Born Bilinguals and the Adult Late
Bilinguals. Finally, the voiceless to voiced ratios turn out to be significantly different between
the L.A. Born Bilinguals and the Spanish Control Group.
A linear regression was run to determine if the voicing ratios could be predicted from the nPVI
scores (including both languages). A significant model emerged even considering obvious
outliers: R
2
= 0.380, intercept of 17.8 and slope of 68.6; F(1,82) = 50.308; p < .001.
18
Figure 14. Voicing ratios vs. nPVI values.
Figure 14 demonstrates the relation between the nPVI values and the voicing ratios. The general
trend is that if a subject presents a big difference in duration between vowels (nPVI score) it is
likely that he/she will also produce a signal with more voiceless than voiced portions (voicing
ratio). This relation highlights once again the validity of two techniques which rely on different
measurements (vowel durations vs. voicing ratios) and procedures (manual vs. automatic) to
quantify rhythm. In section 2.3.3. it will be demonstrated how the lengthening and reduction of
vowels is the main reason for this relation.
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
0.6
25 30 35 40 45 50 55 60 65
Voiceless to voiced ratios
nPVI values
29
VOICING RATIOS VS. NPVI: DISCUSSION
By looking at all the results, conclusions can be made about the techniques used to quantify
rhythm. The variability in voicing only captured a significant difference between the L.A. Born
Bilinguals and the Spanish Control Group in English. This result does not imply that this
technique is not useful to quantify rhythm but it is probably the case that measurements that rely
on standard deviations require different languages to obtain clear results due to inter subject
variability. Additionally, different levels of language proficiency and speaking rates may have
played a role in this variability measure. The voicing ratios detected at least the most noticeable
differences between groups in the same language, meaning that differences between two
languages with different rhythms should be detectable as well. Finally, the normalized pairwise
variability index was the technique that detected the most differences. These observations were
also supported by the results obtained with the Wilcoxon signed-rank tests, where differences
between languages within the same group were always significant for the nPVI technique but
only for the English Control Group and the L.A. Born Bilinguals with the voicing ratios. These
results indicate that nPVIs scores might be capturing differences in vowel length effects (due to
stress or foot size) while voicing rations are not. As discussed before, measurements of rhythm
rely on phonological structural properties and these two techniques may be more or less sensitive
to those. While voicing ratios should capture contrasts yielded by phonotactics (e.g. the more
syllable types, the more irregular the durations), nPVIs should capture the lengthening of
stressed syllables, the shortening of unstressed syllables or any effects in duration caused by feet
size (e.g. polysyllabic shortening).
Results obtained with the normalized Pairwise variability index are also similar to the ones
reported in Carter (2005) with natural speech, highlighting the reliability of this technique to get
comparable data across populations. The nPVI scores resulting from the current experiment
agree with previous studies in that English is clearly more stress-timed (or less syllable-timed)
than Spanish. Carter (2005), based on Thomas and Carter (2003) employs the same technique but
applied to natural speech to study the English and Spanish spoken in North Carolina and gets
very similar nPVI scores in English (53.2 vs. 53.3 in the current study). Also, agreeing with
current results, he also points out that Hispanic English (42.6 vs. 41.3 in the current study) is not
as stress-timed as English and that Spanish has the lowest nPVI scores (27.9 vs. 32 in the current
30
study).
19
These similarities between studies demonstrate the validity of the nPVI technique to
study rhythmic differences across bilingual communities. Additionally, results at the family level
are very relevant since they replicate those found at the group level.
2.3.3 VOWEL DURATIONS AS RHYTHM SHAPERS
So far it has been demonstrated with complementary methodologies that speakers with different
language backgrounds show rhythmic variation in English and Spanish. But what is causing this
variation? The measurements used to calculate the nPVI scores in the passages in both languages
also provide the necessary data to look for an answer at the level of the vocalic segments.
As mentioned in section 2.1, one of the main factors why English and Spanish can be categorized
as having different rhythms is the amount of variability in vowel durations found in each
language. While vowel reductions and deletions are common in English, especially in unstressed
syllables and function words, Spanish and other Romance languages exhibit more moderate
reductions or deletions if any (Grabe et al.1999, Dauer 1983). Also, in English, stressed syllables
are on average as twice as long as unstressed ones (Hoequist 1983, Dauer 1983 Williams &
Hiller 1994), while in Spanish they are only 10% longer (Dauer 1983). Finally, English exhibits
polysyllabic shortening (i.e. the more syllables in a foot, the shorter they are) (Kim & Cole
2006). Some of these contrasts have been reported to be difficult to acquire when learning a
second language, due to the influence of the first language. For example, Gut (2003)
demonstrates that speakers of Romance languages like French or Italian tend not to reduce or
delete vowels when speaking German as a second language while English speakers show too
much reduction. Carter (2005) and Nava (2010) show that vowel reduction or deletion is not
common among speakers of Spanish with English as their second language either. nPVI and
voicing ratio results in this paper capture the variability found between languages in the groups
under study as well. Nevertheless, it is necessary to determine where exactly contrasts in rhythm
are emerging from.
Three kinds of vowels were considered to determine the source of vowel duration variability:
vowels in content words with primary stress, vowels without stress in content words and vowels
31
in function words.
20
The same as in the nPVI calculations, vowels belonging to the last syllable
or foot that were located before a pause or a hesitation were not included to avoid lengthening
effects (Prieto et al. 2010). Also, since classification of vowels produced as a diphthong but
belonging to two different words was not possible, they were excluded from the analysis.
VOWEL DURATIONS IN ENGLISH
Based on the nPVI scores and previous studies on English rhythm, it is expected that in the
English Control Group, stressed content vowels will be as twice as long as function vowels. The
durations of unstressed content vowels will be somewhere in between. Also, it is expected that
the Adult Early Bilinguals and the L.A. Born Bilinguals will show a very similar pattern as the
English Control Group. In contrast, in the Adult Late Bilinguals and the Control Spanish Group,
differences between vowels should not be that extreme presumably due to the lack of vowel
reduction.
Figure 15. Normalized vowel durations per group in English.
Figure 15 shows the normalized average durations in English for the three vowel classes in the
five groups with 95% confidence interval bars.
21
First, it is worth noticing that all groups
produced stressed content vowels of similar lengths when compared to the other two kinds;
however, the lengths of the other two vowel classes varied from group to group. As expected, on
the one hand, the English Control Group, the Adult Early Bilinguals and the L.A. Born
Bilinguals show function vowels that are twice as short as the stressed content vowels. On the
0
0.002
0.004
0.006
0.008
0.01
0.012
Control
English
Adult
Early
Bilinguals
L.A. Born
Bilinguals
Adult Late
Bilinguals
Control
Spanish
content &
stressed
content &
unstressed
function
32
other hand, the Adult Late Bilinguals and the Spanish Control Group also show a difference in
duration, but not as big as in the other three groups.
In order to capture the difference in length between vowel types in each group, Figure 16 shows
the following ratios: average duration of stressed content vowels (in ms) divided by the average
duration of unstressed function vowels (in ms), average duration of stressed content vowels (in
ms) divided by the average duration of unstressed content vowels and the average duration of
unstressed content vowels (in ms) divided by the average duration of unstressed function vowels
(in ms).
Figure 16. Vowel ratios per group in English.
Notice that in Figure 16 higher bars indicate a bigger difference in vowel duration. The biggest
difference for all groups is found in the stressed content vowel vs. function vowel ratio, followed
by stressed content vowel vs. unstressed content vowel ratio. As expected, the difference
between unstressed content vowels and function vowels, although noticeable, was the smallest.
A one-way analysis of variance (ANOVA) determined that there was a statistically significant
difference between groups in English for the content stressed vowel / function vowel ratio (F
(4,42) = 9.687, p < .001). Post-hoc Scheffe tests revealed significant differences between the
English Control Group and the Adult Late Bilinguals (p = .002), the English Control Group and
the Spanish Control Group (p = .005), the Adult Early Bilinguals and the Adult Late Bilinguals
1.95
1.99
1.83
1.47 1.54
1.43
1.59
1.45
1.28
1.30
1.37
1.26 1.26
1.17
1.19
0
0.5
1
1.5
2
Control
English
Adult Early
Bilinguals
L.A. Born
Bilinguals
Adult Late
Bilinguals
Control
Spanish
content &
stressed /
function
content &
stressed /
content &
unstressed
content &
unstressed /
function
33
(p = .002), the Adult Early Bilinguals and the Spanish Control Group (p = .004) and the L.A.
Born Bilinguals and the Adult Late Bilinguals (p = .035). Another one-way analysis of variance
(ANOVA) showed that there was a statistically significant difference between groups in English
for the content stressed vowel / content unstressed vowel ratio (F (4,42) = 5.266, p = .002). Post-
hoc Scheffe tests revealed significant differences only between the Adult Early Bilinguals and
the Adult Late Bilinguals (p = .012) and the Adult Early Bilinguals and the Spanish Control
Group (p = .012). A final ANOVA detected no differences between groups for the unstressed
vowel / function vowel ratios.
Linear regression analyses were run to determine if the stressed content vowel / function vowel
ratio (the most different one between groups) was a good predictor of rhythm in English. A
significant model emerged for the nPVI scores: R
2
= 0.701; F(1,45) = 105.460; p < .001 (Figure
17).
Figure 17. nPVI values vs. content/unstressed vowels in English.
For voicing ratio results, another significant model emerged: R
2
= 0.335; F(1,41) = 20.618; p <
.001 (Figure 18).
30
35
40
45
50
55
60
65
1 1.5 2 2.5
nPVI English
Duration of Stressed Vowels in Content Words / Duration of
Unstressed Vowels in Function Words
Group 1
Group 2
Group 3
Group 4
Group 5
Control English
Adult Early
Bilinguals
L.A. Born
Bilinguals
Adult Late
Bilinguals
Control Spanish
34
Figure 18. Voicing ratios vs. content/unstressed vowels in English.
The stressed content vowel / function vowel ratio (.701) can account for 70% of the nPVI scores
and almost 40% the voiceless to voiced ratios even without removing outliers.
22
Thus, in
English, the difference in duration between vowels, especially between stressed content vowels
and function vowels, seems to be one of the main reasons why different rhythms are found
between groups.
VOWEL DURATIONS IN SPANISH
Based on the findings in English and the nPVI and voicing ratio scores in Spanish, it is
reasonable to expect that in Spanish the English Control Group and the Adult Early Bilinguals
will also show bigger differences in duration between stressed syllables in content words and
function words than the other three groups.
0.15
0.25
0.35
0.45
0.55
0.65
1 1.5 2 2.5
Voiceless/Voiced Ratio
English
Duration of Stressed Vowels in Content Words / Duration of
Unstressed Vowels in Function Words
Group 1
Group 2
Group 3
Group 4
Group 5
Control English
Adult Early
Bilinguals
L.A. Born
Bilinguals
Adult Late
Bilinguals
Control Spanish
35
Figure 19. Normalized vowel durations per group in Spanish.
Figure 19 shows the normalized durations of the three kinds of vowels for the five groups in
Spanish with 95% confidence interval bars. In English, after normalization (Figure 15), stressed
content vowels had similar lengths in the five groups. In this case, however, groups show more
variability and comparisons are harder to make based solely on this graph. Ratios in Figure 20
capture the contrast in vowel length more clearly.
Figure 20. Vowel ratios per group in Spanish.
A one-way analysis of variance (ANOVA) determined that there was a statistically significant
difference between groups in Spanish for the content stressed vowel / function vowel ratio (F
0
0.002
0.004
0.006
0.008
0.01
Control
English
Adult Late
Bilinguals
L.A. Born
Bilinguals
Adult Early
Bilinguals
Control
Spanish
content &
stressed
content &
unstressed
function
1.44
1.41
1.20
1.33
1.33 1.31
1.39
1.21
1.36
1.32
1.10
1.02
0.99
0.98
1.01
0
0.5
1
1.5
2
Control
English
Adult Late
Bilinguals
L.A. Born
Bilinguals
Adult Early
Bilinguals
Control
Spanish
content &
stressed /
function
content &
stressed /
content &
unstressed
content &
unstressed /
function
36
(4,40) = 7.922, p < .001). Post-hoc Scheffe tests revealed significant differences between the
English Control Group and the L.A. Born Bilinguals (p = .001) and the Adult Early Bilinguals
and the L.A. Born Bilinguals (p = .003). Another one-way analysis of variance (ANOVA)
showed that there was a statistically significant difference between groups in English for the
content stressed vowel / content unstressed vowel ratio (F (4,40) = 7.077, p < .001). Post-hoc
Scheffe tests revealed significant differences between the Adult Early Bilinguals and the L.A.
Born Bilinguals (p = .001) and the L.A. Born Bilinguals and the Adult Late Bilinguals (p = .013)
and the L.A. Born Bilinguals and the Spanish Control Group (p = 0.46). A final ANOVA
determined that there was a statistically significant difference between groups for the unstressed
vowel / function vowel ratio (F (4,40) = 5,395, p = .001). Post-hoc Scheffe tests revealed
significant differences between the English Control Group and the L.A. Born Bilinguals (p =
.009), the English Control Group and the Adult Late Bilinguals (p = .007) and the English
Control Group and the Spanish Control Group (p = 0.33). Based on these results it seems that the
the context for vowel reduction was not being met by the English Control Group and the Adult
Early Bilinguals.
A linear regression analysis was run to predict the nPVI scores from the ratio of stressed content
vowel / function vowel and significant model emerged as well: R
2
= 0.356; F(1,43) = 23.793; p <
.001 (Figure 21). Notice that this was much lower than in English, however. Although the same
linear trends were found (i.e. higher nPVI – higher ratio), none of the ratios turned out to be good
predictors of the voiceless to voiced ratios in Spanish. It has to be taken into account that voicing
ratios do not depend solely on vowel durations and that other factors are involved (e.g. VOT,
consonant clusters, inherent vowel durations, etc.). Therefore, it is expected that the role of
vowel duration variability is lower than for nPVIs. This was the case for English as well (Figure
17 vs. Figure 18)
37
Figure 21. nPVI values vs. content/unstressed vowels in Spanish.
A crucial factor that may make distinctions in Spanish less clear than in English is the vowels
under consideration. nPVI scores and vowel durations ratios have been calculated with the same
materials and both nPVI values and voiceless to voiced ratios indicate that rhythm is less
syllable-timed for the English Control Group than for native speakers of Spanish in the Spanish
Control Group. Then, it is also a possibility that the reason why speakers in the English Control
Group have only slightly higher length ratios than native speakers of Spanish but different nPVI
scores is that they are reducing, deleting or lengthening vowels that do not always correspond to
the three groups under consideration. Figures 22 and 23 exemplify this possibility. Figure 22
shows the spectrogram for a participant belonging to the English Control Group producing El
Viento Norte sopló – ‘The North Wind blew’ (Capitalized syllables are the stressed ones). Figure
23 corresponds to the same production by a speaker in the Adult Late Bilinguals Group. Notice
the difference in the word norte – ‘north’. While the Adult Late Bilingual shows a longer vowel
in “nor-” (the stressed syllable), the speaker in the English Control Group shows a longer vowel
in “-te” (the unstressed syllable).
20
25
30
35
40
45
50
55
1 1.2 1.4 1.6
nPVI Spanish
Duration of Stressed Vowels in Content Words / Duration of
Unstressed Vowels in Function Words
Group 1
Group 2
Group 3
Group 4
Group 5
Control English
Adult Early
Bilinguals
L.A. Born
Bilinguals
Control Spanish
Adult Late
Bilinguals
38
Figure 22. Spectrogram of ‘el viento norte sopló’ – English Control Group speaker.
Figure 23. Spectrogram of ‘el viento norte sopló’ – Adult Late Bilinguals speaker.
Thus, due to a potential inability to distinguish between kinds of words (and vowels) in Spanish
as expressed by some of the participants after reading the text (especially in the Control English
Group and the Adult Early Bilinguals), reduction and lengthening can be taking place randomly
to still preserve the type of rhythm of their first language. Besides, a closer inspection of the
productions of the English Control Group, the Adult Early Bilinguals and the L.A. Born
Bilinguals revealed the erroneous placement of some stresses throughout the paragraph,
(porfiaBAN instead of porfiAban, aCERto instead of acerTO, lograRA instead of loGRAra,
etc).
23
The following two figures exemplify this finding. Figure 24 represents the production of
abandonó ‘gave up’ by a participant in the Adult Late Bilinguals Group and Figure 25 by a
participant in the Adult Early Bilinguals Group. Notice how in the first case the last syllable (-
Frequency (Hz)
0
5000
el VIEN to NOR te so PLO
Frequency (Hz)
0
5000
el VIEN to NOR te so PLO
39
no) has the longest duration since it is the stressed one. In the other case, in contrast, it is the
syllable -do- the one that has erroneously been assigned stressed and consequently lengthened:
Figure 24. Spectrogram of ‘abandonó’ – Adult Late Bilingual speaker.
Figure 25. Spectrogram of ‘abandonó’ – Adult Early Bilingual speaker.
Thus, stresses in the productions of the English Control Group and the Adult Early Bilinguals are
not always located in the same places as for the Adult Late Bilinguals and the Spanish Control
Group, resulting in lower ratios than the ones expected (i.e. a stressed content vowel in the
Spanish Control Group may be considered an unstressed vowel in the English Control Group
since all the labeling used for the vowels was based on prescriptive pronunciations).
24
VOWEL DURATIONS: CONCLUSIONS
Some observations and conclusions can be made from the current results. It is clear that all the
groups modify the durations of the vowels depending on the language that is being produced. In
Frequency (Hz)
0
5000
a ban do NO
Time (s)
0.512 0
Frequency (Hz)
0
5000
a ban do NO
Time (s)
1.001 0
40
all groups vowel duration ratios tend to be higher (i.e. durations are more different between
vowel classes) in English than in Spanish. Nevertheless, the amount of variation is different for
each group.
In English, the contrast between stressed content vowels and function vowels is larger in the first
two groups than in the last two, demonstrating the tendency of L2 English speakers with Spanish
as their mother tongue to not show enough contrast. These results also predict the nPVI results
and demonstrate that the contrast in vowel durations in stressed syllables and function words
(and between the three kinds of vowels under consideration) plays a very important role in
English rhythm.
Differences in vowel length between classes in Spanish are not as big as those in English;
consequently, statistical differences in the stressed content vowels vs. function vowels ratios
were only found between the L.A. Born Bilinguals vs. the English Controls and Adult Early
Bilinguals. The fact that in some cases stresses fell erroneously in other syllables and that some
speakers did not know some of the words in the text may have resulted in lower vowel durations
in these groups in Spanish. The English Control Group and the Adult Early Bilinguals have
results that clearly indicate a rhythm in Spanish that is less syllable-timed than that of native
speakers. Thus, it is reasonable to assume that the contrast in vowel length is still happening due
to the influence of English, but not necessarily between the three vowel classes that are relevant
to the other groups because of an inability to identify them or the random application of
reduction and lengthening as shown in Figures 22 through 25. As suggested in the previous
section, the phonotactics of the language may have contributed to this asymmetry as well.
2.4. RHYTHM: GENERAL DISCUSSION
Data on the Spanish and English spoken in Los Angeles County reflect the importance of
distinguishing between language groups within the same community in order to make linguistic
generalizations. Language contact is a reality, and Los Angeles County provides an excellent
setting in which to study how two languages belonging to two different linguistic families may
have had an influence on each other. In this section the focus of the study has been rhythm, since
41
as pointed out by O’Rourke (2005), different varieties of the same language may show different
rhythmic properties. This was found to be the case for both English and Spanish in the groups
under study. Given the controversy on which technique quantifies rhythmic differences better,
this research departs from relying on a single measuring method and compares two alternatives:
the normalized pairwise variability index (with manual measurements of vocalic segments) and
the analysis of voiceless to voiced ratios and variability of voiceless intervals (with automatic
measurements of the whole speech stream).
Results rely on the productions of forty-nine speakers and clear trends can already be seen
between groups.
25
Results for English demonstrate that native speakers of English in the English
Control Group and Mexican Spanish speakers in Spanish Control Group have very different
rhythmic properties (see Tables 3 and 4 for significant differences). While native speakers of
English show high voiceless to voiced ratios and high nPVI scores (i.e. more variability in vowel
length), Spanish Control Group has less vowel durational variability and the speech stream is
more voiced. Speakers of Mexican Spanish who moved to Los Angeles at an early age (The
Adult Early Bilinguals) show rhythmic properties very similar to actual native speakers of
English, while speakers of Mexican Spanish who moved to Los Angeles later in their lives (The
Adult Late Bilinguals) show patterns similar to those by Spanish Control Group. Finally,
bilingual children born in Los Angeles (The L.A. Born Bilinguals), show rhythmic properties
similar to those by the English Control Group in English and the Spanish Control Group in
Spanish.
As in Gut (2003) these results have been determined to be the direct result of varying or failing
to vary vowel durations in stressed and unstressed syllables. Spanish does not show as much
reduction in vowel as English does, and stressed syllables are not much longer than unstressed
ones. Adult Late Bilinguals and speakers in the Spanish Control Group transfer these stress
effects to their English productions, resulting in a less stress-timed English than the one
produced by native speakers. The L.A. Born Bilinguals shows more flexibility and less influence
of Spanish and consequently is closer to controls. This variability in the ability to adapt to
English by the last four groups seems to be related to the age when the different groups moved to
Los Angeles and were resultingly exposed to English. The Adult Early Bilingual Group and the
42
L.A. Born Bilingual Group were designed so that the main difference between them was the age
at the time of the recording (The Adult Early Bilinguals were older than the L.A. Born
Bilinguals). The reason for this grouping is to determine if the prosodic characteristics of these
groups have been affected differently due to different amounts of exposure to English. Ideally,
speakers in both groups should have been born in L.A. but most of the speakers in the Adult
Early Bilinguals arrived slightly later than those in the L.A. Born Bilinguals. Crucially, it could
be argued that the L.A. Born Bilinguals could have an advantage over the Adult Early Bilinguals
since they were exposed to English from an early age; Nevertheless, if those extra years outside
L.A. (i.e. in Mexico) had played a major role in the shaping of their rhythm in English and
Spanish, it would be expected that their productions would be more Spanish-like (because they
arrived from Mexico to L.A. later). This is not the case since their rhythms are English-like in
both languages. A good example is speaker g2s5, who arrived in L.A. at the age of seven but
shows an nPVI score in Spanish (41) higher than any of the L.A. Born Bilinguals. Besides, only
two of the L.A. Born Bilinguals show higher voicing ratios scores than him in Spanish (0.41).
Similar observations can be made about g2s6 and g2s7, who arrived in L.A. at the age of five.
Consequently, the slight difference on the age on arrival did not seem to affect the results. Adult
Early Bilinguals moved to L.A. early in their childhood and therefore they have a native-like
English rhythm due to an early exposure to the language. The continuous contact with English
for years has also made their Spanish productions have English-like rhythm. The L.A. Born
Bilinguals, like the Adult Early Bilinguals, have been exposed to English since early in their
childhood and consequently they show stress-timed scores in English. Unlike the Adult Early
Bilinguals, the L.A. Born Bilinguals have lived for a shorter time in contact with English and
have been using Spanish on regular basis; thus, their Spanish has not suffered as much influence
as that of the Adult Early Bilinguals. Finally, Adult Late Bilinguals started being exposed to
English later in their lives, making their English reflect the influence of Spanish rhythm. As with
all the other groups, their Spanish might be acquiring English-like rhythmic properties, however.
These observations are also supported by the analyses of different family members belonging to
different groups.
Carter (2005) points out that the Spanish of bilingual Spanish-English speakers is less syllable-
timed than that of Spanish monolinguals and hypothesizes that it might be due to the influence of
43
English. Data on the Spanish spoken in Los Angeles County confirms this intuition. The results
of this study also agree with Gut (2003) and White and Mattys (2007), since second languages
show clear rhythmic influences of first languages (L1 transfer). The Adult Late Bilinguals and
the Spanish Control Groups exemplify the influence of Spanish with less stress-timed English
and the English Control Group shows how English makes Spanish less syllable-timed. Finally,
the language attrition process is clearly exemplified by the Adult Early Bilinguals. In future
work, a longitudinal study following the productions of non-control groups would shed more
light on the implications for the acquisition rhythm and the influence of English in Spanish in
particular.
Observations at the group level in the Los Angeles County community resemble those in second
language acquisition studies and indicate that two completely different methodologies can lead to
the same conclusions. Although all groups try to accommodate their rhythm to that of the target
language, clear influences of the second language are present. In some cases, even the second
language (English) can influence the first language (Spanish) giving rise to attrition processes.
These are more or less noticeable depending on sociolinguistic factors such as the amount of
exposure or the age. The next step is to determine if other prosodic variables like intonation are
affected in the same way within the same community in Los Angeles.
44
CHAPTER 3: INTONATION
3.1. INTRODUCTION
The rate at which vocal folds vibrate determines how high or low pitch is (the higher the rate the
higher the pitch). Variations in pitch, of which the acoustic correlate is the fundamental
frequency or F0, are used for pragmatic purposes in the discourse in Spanish and English since
different pitch contours convey different pragmatic meanings. This section concentrates on the
intonational properties of the Spanish and English produced in Los Angeles by the five different
groups under study: the English Control Group, the Adult Early Bilinguals, the L.A. Born
Bilinguals, the Adult Late Bilinguals and the Spanish Control Group.
As has been demonstrated in chapter 2, Spanish and English in L.A. differ in their rhythmical
properties; Likewise, a comparison of previous studies on both languages (Ladd 1978, 1980,
Pierrehumbert 1980, 2000, Martín Butragueño 2004, 2006, Girand 2006, Prieto & Roseano
2010) reveal that some of the intonational patterns can differ as well. Previous studies that
concentrate on Spanish intonation in language contact situations include Elordieta (2003) for
Peninsular Spanish and Lekeitio Basque, Colantoni and Gurlekian (2004) for Argentinian
Spanish and Italian, O’Rourke (2005) for two varieties of Peruvian Spanish, Robles-Puente
(2012) for Spanish and Gernika Basque, Kireva (2013) for Olivenza Spanish and Portuguese,
Alvord (2006) for Miami Spanish and English, Thomas and Ericson (2007) for Mexican
American and Anglo English and Miglio and colleagues (2013) for Mexican and Chicano
Spanish. Except for these last three studies, not much attention has been paid to the influence of
English intonation in Spanish (and vice versa) or the potential gradual modification of
intonational patterns, however.
Elordieta (2003), Colantoni and Gurlekian (2004) and O’Rourke (2005) find modifications in the
peak alignments of pitch-accents in Spanish due to Basque, Italian and Quechua influence
respectively. Robles-Puente (2012) demonstrates how Gernika Basque and the Spanish of the
area share intonational patterns in yes/no questions and how younger generations of Gernika
Basque might be losing lexical pitch-accents due to a more generalized use of Spanish. Alvord
(2006) shows how falling intonations in yes/no questions characteristic of Cuban Spanish are
present in certain generations of Miami Spanish but not in others. Thomas and Ericson (2007)
45
show that Mexican Americans show higher proportions of rising pitch-accents than Anglo
speakers. Finally, Miglio and colleagues (2013) find that Mexican Spanish has less pitch
movements than Chicano Spanish. What all these studies have in common is that they
demonstrate that the intonational properties of contact varieties present deviations from the
phonological system of a Spanish monolingual.
The intonational descriptions and particularities presented in the previous lines generally rely on
the Autosegmental-Metrical (AM) model of intonational phonology (Pierrehumbert & Beckman
1986, Ladd 1996, Gussenhoven 2004, Beckman et al. 2005, inter alia). According to this model,
elements from an inventory of discrete tones are associated with stressed syllables and phrase
boundaries. Tones associated with stressed syllables (where ‘*’ denotes the stressed or metrically
strong syllable) are pitch-accents. If a pitch-accent occurs towards the end of a sentence it is
considered a nuclear pitch-accent. Preceding pitch-accents (if any) are pre-nuclear pitch-accents.
Final tones are called boundary tones and the symbol ‘%’ marks the boundary of a prosodic unit.
Not all syllables may have a tone, but tones are varied and can go from low (L* or L% in a
boundary) to high (H* or H% in a boundary). Combinations of these and different affiliations to
stressed syllables are also possible. For example, L*+H indicates that the stressed syllable (*) has
a low tone (L) and it is followed by a syllable with a higher F0 (H - normally in form of a peak).
L+H*, in contrast, indicates that the F0 rise and peak occur within the limits of the stressed
syllable. This methodology gives the possibility of describing and comparing intonational
contours by subdividing them into a string of categorically distinct elements (Ladd 2008).
Based on this theory, some of the most particular intonational characteristics of English and
Spanish (to be described in more detailed in following sections) are the following:
26
1) use of H*
tones in pre-nuclear pitch-accents in English declaratives (vs. L+>H* in Mexican Spanish); 2)
use of H*L% in nuclear pitch-accents in English declaratives (vs. L*L% in Mexican Spanish); 3)
use of sustained boundary tones (H!%) in Mexican Spanish declaratives (vs. L% in English); 4)
use of upstepped nuclear pitch-accents (L+¡H*) in Mexican Spanish declaratives (vs. H*L% in
English) and 5) use of L* HL% tonemes in Mexican Spanish vocatives (vs. L+H* !H% in
English). It is crucial to note that these are just the most salient intonational features but a) some
pragmatic meanings have not been studied so far and b) no study has compared (Mexican)
46
Spanish and (American) English intonations side by side in detail (see Estebas-Vilaplana 2007
for a comparison of Peninsular Spanish and British English declaratives). For these reasons
assorted types of sentences will be covered in both languages. These included declaratives, yes-
no questions, wh-questions, imperatives of different strengths and vocatives (see Appendix C).
Another point that needs consideration is the potential lack of prosodic homogeneity across the
five groups under study speaking the same language. Spanish and English rhythms have been
described as more syllable-timed and more stress-timed respectively; however, this contrast is
reflected to a greater or lesser extent depending on the group that is being studied as reflected in
chapter 2. Consequently, intonational characteristics can exhibit the same kind of variation. The
null hypothesis is that all groups will produce patterns characteristic of English in English and
patterns characteristic of Spanish in Spanish. Nevertheless, based on the observations at the
rhythmic level, the following predictions can be made:
1) Given that their rhythms are more stress-timed in both languages, the English Control
Group and the Adult Early Bilinguals will show English-like intonation in English and in
Spanish.
2) The L.A. Born Bilinguals will accommodate the use of their intonational patterns
depending on the language they are producing like they do with rhythm.
3) Since the Adult Late Bilinguals and the Spanish Control Group have syllable-timed
rhythms in both languages, they will show Spanish-like intonations not only in Spanish
but in English as well.
The validity of these predictions depends on how likely the prosodic features of rhythm and
intonation are to be affected in a similar way.
3.2. GENERAL METHODOLOGY
In order to obtain pragmatic contrasts and contrastive melodies, prompted speech was recorded
in interviews where participants had to give an answer or produce a sentence in a given context.
This methodology was chosen because the same procedure has been satisfactorily used in the
47
study of multiple varieties of Spanish and Romance languages over the world (see Prieto &
Roseano 2010). The experimenter asked participants to imagine themselves in different
situations and to produce a sentence in a natural way. Both situations and answers were visually
presented as well and in the few cases where the situation was not clear to the participants this
was reformulated. Situations were presented in the same order to all participants although
participants who were more fluent in English started producing the English sentences and
participants who preferred Spanish started with the Spanish sentences. In cases where intonation
was clearly forced or sounded unnatural, participants were asked to listen to their productions in
order to evaluate them. In the few cases when participants thought their utterances sounded
unnatural they were repeated. This was the case for utterances with hesitations, unexpected
pauses or background noises as well.
For comparative purposes, most of the elicitation materials are based on the ones used for two
ongoing projects: 1) Atlas Multimedia de la Prosodia del Espacio Románico (AMPER) ‘The
Multimedia Atlas of the Prosody of Romance Space’, whose main goal is to design a prosodic
atlas of all the Romance languages in a digital format so that it is available to the scientific
community via the Internet and 2) Atlas Interactivo de la Entonación del Español ‘The
Interactive Atlas of Spanish Intonation’, which compiles the intonational properties of Spanish
varieties spoken all over the world. Since these projects concentrate on Romance languages and
Spanish respectively, the equivalent elicitation materials were created for English (see Appendix
C).
All productions were analyzed with the speech analysis software Praat (Boersma & Weenink
2009) and labeled following the Autosegmental-Metrical (AM) model of intonational phonology
(Pierrehumbert & Beckman 1988, Ladd 1996, Gussenhoven 2004, Beckman et al. 2005, inter
alia). Since the main purpose of the analysis is to compare the intonational contours produced in
both languages, I am consistent with the tone labeling and follow the latest Spanish ToBI
notations (Prieto & Roseano 2010). I also make references to previous labels used in the
literature in both languages for comparative purposes and propose new ones in cases where a
distinction is not captured by the current labeling system. All these analyses combined provide
an effective way to compare the productions of each speaker not only to the descriptions already
48
present in the literature and in the aforesaid projects but also to those produced by other speakers
in the Los Angeles community.
The following sections present the intonational characteristics of pre-nuclear pitch-accents in
broad and narrow focus in declaratives (3.3. through 3.5.), nuclear pitch-accents in declaratives
(3.6), imperatives of different strengths (3.7.), tonemes in wh- and yes-no questions (3.8. and
3.9.) and vocatives (3.10) in L.A. Spanish and English. Each section contains a short summary of
the previous findings on a given pragmatic meaning (introduction), a brief description of the
sentences produced and the tonal labelings used (methods), a summary of the findings per group
and language in the current study (results) and a final summary with observations (discussion).
27
3.3. DECLARATIVES: PRE-NUCLEAR PITCH-ACCENTS IN BROAD FOCUS
3.3.1 INTRODUCTION
Nuclear pitch-accents are those found in the last stressed syllable of the phrase while pre-nuclear
pitch-accents are those located before. Sosa (1999), Face (2002) and the first Spanish ToBI
proposal (Beckman et al. 2002) label pre-nuclear rising accents with delayed F0 peaks as L*+H
in Spanish. Nevertheless, Prieto and Face (2007) and more recently Prieto and Roseano (2010)
suggest that a three way distinction is necessary: L+>H* (delayed peak) vs. L*+H (late rising)
vs. L+H* (early rising).
L+>H* is used for tones with a rise of F0 from the stressed syllable and a peak in the following
syllable (“>” indicates the peak delay). Apart from Mexican Spanish, this kind of accent is also
found in Castilian, Chilean, Canarian and Ecuadorian Andean Spanish. If the peak is located
within the limits of the stressed syllable (normally at the end), L+H* is used instead. It is
necessary to highlight that this accent is normally found in narrow focus declaratives or in
contrastive focus situations and will be discussed in detail in the following section but it is
introduced here for comparative purposes. Alternatively, the F0 rise can start later, finding a low
tone (L*) in the stressed syllable as is generally the case for Caribbean varieties of Spanish. In
these cases the L*+H label is used. All three configurations are found in the current data on Los
49
Angeles Spanish, although their amounts of use and the pragmatic contexts where they are found
vary.
In English, high tones (H*) are characteristic of pre-nuclear pitch accents (Pierrehumbert 1980,
2000). Nevertheless, the L+H* vs. H* distinction has never been clear (Pitrelli et al. 1994, Xu &
Xu 2005) and both tones are normally equated since they are not considered categorically
distinct (Ladd & Schepman 2003).
28
I will argue that at least a phonetic distinction is desirable in
the variety of English spoken in Los Angeles.
3.3.2 METHODS
The pre-nuclear pitch accents considered were always in sentence initial position to make sure
preceding tones did not affect F0 contours. Also, the syllabic structure of the words under
analysis was controlled so that it contained phonetic material both before and after the stressed
syllable to ensure that tones could be fully produced (e.g. give them time to rise or fall). The
words under consideration were the following (“_” indicates stressed syllable): Amanda, anima
‘cheers up’, el niño ‘the kid’, la señora ‘the woman’ in Spanish and Amanda (twice) and the
woman in English. After removing utterances where the pitch contour was not analyzable, a total
of 100 pre-nuclear pitch accents were analyzed in Spanish and 92 in English. In the present study
and based on the collected data it is necessary to distinguish at least between the tones in Table 5
in initial pre-nuclear position for both languages.
Apart from the basic L*+H configuration, two variants of this were found. In a few instances,
L*+H accents were preceded by a clear H%. The second variation was that some of the subjects
presented phonetic realizations that could not be strictly categorized as L*+H or L+>H* since
the beginning of the F0 rise was not found at the beginning of the post-tonic syllable. This kind
of alignment puts into question the homogeneity in the association of H tones and stressed
syllables and opens the possibility that the anchoring point of the F0 rise within the stressed
syllable may vary. Although these different alignments may not result in a phonological contrast,
their phonetic realization indicates that the coordination of F0 movements and vocal tract
gestures may differ from speaker to speaker even within the same language. This configuration
50
will be labeled as L*+H (intermediate) to capture that the F0 rise is not exactly L*+H nor
L+>H*.
Table 5. Tones found in declaratives in initial pre-nuclear position.
3.3.3 RESULTS
Figures in Table 6 show examples in Spanish and English of the pre-nuclear pitch accents
schematized in Table 5.
29
Tone Representation Description
L*+H
F0 valley on the accented
syllable and a delayed F0
rise normally at the onset of
the post-stressed syllable. It
can also be found combined
with an initial H%.
L+>H*
F0 rise at the onset of the
stressed syllable with a peak
outside the limits of the
stressed syllable.
L+H*
F0 rise at the onset of the
stressed syllable with a peak
within the limits of the
stressed syllable.
(H%) H*
High F0 (plateau) through-
out the stressed syllable.
51
Pre-
nuclear
pitch
accent
Examples Description
L+>H*
Amanda compra
bananas
Amanda is buying
bananas
g1s5 - English Control
F0 rise from /m/ in the
stressed syllable -
MAN- in Amanda
with a peak in the
post-tonic syllable -da.
Anima a Amanda
He cheers Amanda up
g2s5 - Adult Early
Bilingual
F0 rise from /n/ in the
stressed syllable -NI-
in anima with a peak
in the post-tonic
syllable -ma.
Amanda is buying
bananas
g2s6 - Adult Early
Bilingual
F0 rise from /m/ in the
stressed syllable -
MAN- in ‘Amanda’
with a peak in the
post-tonic syllable -da.
Freq (Hz) 0
5000
F0 (Hz)
300
75
Amanda compra bananas
Freq (Hz) 0
5000
F0 (Hz)
200
50
Anima a Amanda
Freq (Hz) 0
5000
F0 (Hz)
200
50
Amanda is buying bananas
52
L*+H
L*+H with
int.
alignment
La señora morena
vende mandarinas
‘The brunette lady
sells tangerines’
g5s10 - Spanish
Control
F0 rise from /ɾ/ in the
post-tonic syllable –ra
in señora.
Amanda is buying
bananas
g3s9 - L.A. Born
Bilingual
F0 rise from /d/ in the
post-tonic syllable -
da- in ‘Amanda’.
La señora morena
vende mandarinas
‘The brunette lady
sells tangerines’
g2s8 - Adult Early
Bilingual
F0 rise from /o/ (not
from /ɲ/) in the tonic
syllable -ÑO- in
señora.
Freq (Hz) 0
5000
F0 (Hz)
350
75
la señora morena vende mandarinas
Freq (Hz) 0
5000
F0 (Hz)
300
75
la señora morena vende mandarinas
Freq (Hz) 0
5000
F0 (Hz)
275
125
Amanda is buying bananas
53
H% L*+H
Amanda is preparing
lemonade
g4s1 - Adult Late
Bilingual
F0 rise from /a/ in the
stressed syllable -
MAN- in ‘Amanda’
with a peak in the
post-tonic syllable -da.
Anima a Amanda
‘He cheers Amanda
up’
g4s2 - Adult Late
Bilingual
F0 rise preceded by a
high boundary tone.
Amanda is preparing
lemonade
g2s2 - the Adult Early
Bilinguals
F0 rise preceded by a
high boundary tone.
Freq (Hz) 0
5000
F0 (Hz)
275
150
Amanda is preparing lemonade
Freq (Hz) 0
5000
F0 (Hz)
300
75
Anima a Amanda
Freq (Hz) 0
5000
F0 (Hz)
300
75
Amanda is preparing lemonade
54
H% H*
Amanda compra
bananas
‘Amanda is buying
bananas’
g1s3 - English Control
H* tone in -man- with
a linear transition
between H* accents
and final fall due to
L%
Amanda is buying
bananas
g1s3 - English Control
H* through-out the
stressed syllable and
an initial H%
Table 6. Examples of pre-nuclear pitch-accents in declarative sentences (red area marks the
limits of the stressed syllable).
Figure 26 shows the percentage of accent instances produced by group and language. Bars with
horizontal patterns indicate that the tone is part of the English inventory; bars with vertical
patterns indicate that the tone is part of the Spanish inventory. Solid bars indicate that tones are
expected to be produced both in English and Mexican Spanish.
Freq (Hz) 0
5000
F0 (Hz)
300
75
Amanda compra bananas
Freq (Hz) 0
5000
F0 (Hz)
200
50
Amanda is buying bananas
55
Figure 26. Percentages of pre-nuclear pitch-accents in declaratives (broad focus).
The most common pattern found in every group and language (except the Adult Late Bilinguals)
was L+>H* followed by the intermediate L*+H pitch-accent. Utterances with initial H%s were
found in every group as well. Interesting contrasts are found in the uses of H* and L*+H pitch-
accents. H*s were exclusively produced by the English Control Group, the Adult Early
Bilinguals and the L.A. Born Bilinguals in both languages, but never by the Adult Late
Bilinguals or the Spanish Control Group. In contrast, L*+H accents (i.e. with a late F0 rise) were
only found in the L.A. Born Bilinguals, the Adult Late Bilinguals and the Spanish Control
Group’s productions. Although the number of instances of each pattern was not high, it indicates
clear group differences and preferences.
3.3.4 DISCUSSION
Girand (2006) after analyzing 1064 sentences in English, finds that 64.3% of the pre-nuclear
pitch accents are H* and 15.8% are L+H* but reports that they may not be intonationally
different and groups them as the same category. Importantly, L*+H pre-nuclear pitch-accents
only represented 1.9% of the total. Similarly, Jun (2003) finds that English uses H* and Spanish
L*+H in focused constructions. Los Angeles data confirm these observations and demonstrate
that there is a clear tendency for L1 English speakers (English Control Group) and early English
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Sp En Sp En Sp En Sp En Sp En
Control
English
Adult early
bilinguals
L.A. born
bilinguals
Adult late
bilinguals
Control
Spanish
Pre-nuclear pitch-accents (declaratives)
H% L*+H
L*+H (int)
L+>H*
L*+H
H*
56
learners (Adult Early Bilinguals) to prefer H* and L+>H* configurations over L*+H as opposed
to Spanish speakers who have the opposite preference.
The large number of what I have called L*+H (intermediate) contours in both languages by all
groups highlights the importance of considering adding a new label the tonal analysis to at least
capture this phonetic difference. Although this alignment may not have phonological
consequences it demonstrates that the affiliation of tones with stressed syllables in pre-nuclear
pitch-accents is phonetically varied and cannot be captured with a simple L*+H vs. L+H*
distinction (see Mücke et al. 2002 to explain this contrast based on gestural coordinations).
Most of the contours were found in different sentences and languages so the materials should not
have played a role in the kinds of tones produced. These findings also complement previous
intonational descriptions in both languages. For example, Estebas-Vilaplana (2007) analyzes the
pre-nuclear pitch-accents produced by a British English speaker and a Peninsular Spanish
speaker in broad focus declaratives. According to her, English and Spanish differ in the
alignment of the F0 peak since the maximum F0 value in English is found within the limits of the
stressed syllable and in Spanish in the post-tonic syllable. The analysis of the data collected in
Los Angeles makes clear that this distinction does not apply to the varieties of English and
Spanish spoken in L.A. County since late peaks were equally found in both languages. Finally,
Hualde and Prieto (forthcoming) hypothesize that L*+H pre-nuclear pitch-accents may be a
dialectal feature of Caribbean varieties and the Spanish spoken in the Canary Islands; however,
these data suggest that, although the use is not as high, Los Angeles Mexican Spanish can also
make use of such configurations in neutral declarative sentences.
3.4. DECLARATIVES: PRE-NUCLEAR PITCH-ACCENTS IN NARROW FOCUS
3.4.1 INTRODUCTION
In the previous section it has been demonstrated that L+>H* patterns are the most common ones
in either language in initial position of broad focus declaratives, but what about utterances with
narrow focus? It has been widely suggested that the main difference between pitch accents in
broad and narrow focus can be captured in terms of peak alignment. While in broad focus
57
declaratives the F0 peak is located outside the boundaries of the stressed syllable (L+>H*), in
narrow focus sentences it is located much earlier and consequently within those boundaries
(L+H*).
3.4.2 METHODS
In order to test if these descriptions apply to L.A. Spanish and English participants produced the
sentence Andrea compra bananas ‘Andrea is buying bananas’ with the subject Andrea in narrow
focus. Two different tones were found in pre-nuclear pitch-accents in narrow focus, namely
L+>H* and L+H*. The main difference between these tones is that in the former the peak is
reached outside the limits of the stressed syllable while in the latter the peak is located within
those limits.
Table 7. Tones found in declaratives in narrow focus in initial pre-nuclear position.
3.4.3 RESULTS
Table 8 shows the results for the sentence ‘Andrea is buying bananas’ (where the subject Andrea
is in narrow focus) in Spanish and English. Figures in Table 9 exemplify both configurations.
Notice the difference between the last two figures in Table 9, where the F0 rises until the end of
the word (L+>H*), and the first two figures in Table 9, where the F0 stays within the boundaries
of the stressed syllable “-dre-” (L+H*). As expected, the latter configuration is widely used in
Tone Representation Description
L+>H*
F0 rise at the onset of the
stressed syllable with a peak
outside the limits of the
stressed syllable.
L+H*
F0 rise at the onset of the
stressed syllable with a peak
within the limits of the
stressed syllable.
58
the narrow focus context but L+>H* is available as well. This observation applies to both
languages.
Pre-nuclear
pitch-accent
Narrow focus
Andrea
compra
bananas
Andrea is
buying
bananas
g1s2 - L+H*
g1s3 L+H* L+H*
g1s4 - L+H*
g1s5 L+H* L+H*
g1s6 - L+H*
g1s9 - L+H*
g1s11 L+>H* L+>H*
g1s12 L+H* L+H*
g2s1 L+H* L+H*
g2s2 L+>H* L+>H*
g2s3 L+>H* L+>H*
g2s4 L+H* L+H*
g2s5 L+>H* L+>H*
g2s6 L+H* L+H*
g2s7 L+H* L+H*
g2s8 L+H* L+H*
g3s1 L+>H* L+>H*
g3s2 L+>H* L+H*
g3s3 L+>H* L+>H*
g3s5 L+>H* L+H*
g3s9 L+>H* L+H*
g3s10 L+>H* L+>H*
g3s11 L+H* L+H*
g4s1 L+>H* L+H*
g4s2 L+H* L+>H*
g4s3 L+H* L+H*
g4s7 L+>H* L+>H*
g5s2 L+H* L+H*
g5s9 L+>H* L+>H*
g5s10 L+>H* L+>H*
g5s11 L+>H* L+H*
Table 8. Types of pre-nuclear pitch-accents in narrow focus declaratives in L.A. Spanish and
English.
59
L+ H*
L+>H*
Andrea compra
bananas – ‘Andrea
is buying bananas’
g1s5 (English
Control)
Andrea is buying
bananas
g4s3 (Adult Late
Bilingual)
Andrea compra
bananas – ‘Andrea
is buying bananas’
g3s3 (L.A. Born
Bilingual)
Andrea is buying
bananas
g5s9 (Spanish
Control)
Table 9. Examples of pre-nuclear pitch-accents in broad focus declaratives in L.A. Spanish and
English (red area marks the limits of the stressed syllable).
Freq (Hz) 0
5000
F0 (Hz)
300
100
Andrea compra bananas
Freq (Hz) 0
5000
F0 (Hz)
175
50
Andrea is buying bananas
Freq (Hz) 0
5000
F0 (Hz)
200
75
Andrea compra bananas
Freq (Hz) 0
5000
F0 (Hz)
175
60
Andrea is buying bananas
60
3.4.4 DISCUSSION
Results show that the same strategy is frequently employed by groups in both languages to
denote narrow focus (L+H*), but that the standard L+>H* configuration is equally available. It is
also relevant that out of the twenty-seven speakers that produced the sentence in both languages,
only six varied their configuration from language to language. This indicates that the use of
L+H* might be dependent on contextual factors and the need of the speaker to indicate the
contrast in each specific situation.
3.5. L+H* VS. L+>H*
3.5.1 INTRODUCTION
The core difference between L+H* and L+>H* is the location of the F0 peak. While analyzing
pre-nuclear pitch-accents in the current L.A. data it was noticed that the exact location of the
peak with respect to the stressed syllable varied depending on its structure. Consequently a more
controlled experiment was run to determine if the location of the F0 peak is dependent on the
length of the syllable.
3.5.2 METHODS
In order to test if the location of F0 peaks in L+H* and L+>H* tones is dependent on syllabic
length, three participants belonging to the English Control Group and three belonging to the
Adult Late Bilinguals took part in the experiment.
30
Their task was to answer questions that
required a neutral declarative (e.g. What happened? – Amanda prepared lemonade) or
declaratives where the subject was in focus. (e.g. Tell your friend that Amanda and not Mary
prepared lemonade – Amanda prepared lemonade). A total of six nouns were analyzed in
subject position. Three of them had a CV stressed sequence (underlined) followed by a
consonant and a vowel or a diphthong: Madonna, Salina, Amelia. The other three names only
differed in that the stressed sequence was CVC and therefore longer: Fernando, Amanda,
Belinda. All proper nouns were followed by verbs that had a CV unstressed syllable and were
stressed in the following syllable. This homogeneity in syllable structure and location of stresses
was used to avoid potential interactions with following pitch accents.
61
Each noun was produced five times for each context. After excluding utterances where pitch
tracks were not usable a total of 168 sentences were analyzed in Spanish (45 CV broad focus, 38
CVC broad focus, 42 CV narrow focus, 43 CVC narrow focus) and 154 in English (40 CV broad
focus, 38 CVC broad focus, 40 CV narrow focus, 36 CVC narrow focus). The beginning of the
F0 rises and the F0 peaks were calculated for each utterance and then their locations and the
length of the syllables under analysis were normalized with Praat (Boersma & Weenink 2009) by
using Xu’s (2013) script.
31
Results were averaged for each speaker for the four separate contexts
(CV broad, CVC broad, CV narrow and CVC narrow).
3.5.3 RESULTS
The following four figures (Figure 27 through Figure 30) capture the peak locations for each of
the contexts (CV broad, CVC broad, CV narrow and CVC narrow). s1 and s2 mark each of the
fifteen portions of the stressed syllable and the post-stressed syllable after the time normalization
respectively. The vertical line indicates the boundary between both syllables and the dots the
highest F0 values (peaks) for each speaker.
Figure 27. F0 peak location in CV sequences with broad focus.
80
90
100
110
120
130
140
150
s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2
Hz
Syllable
F0 peak location (CV broad focus)
en1
en2
en3
sp1
sp2
sp3
62
Figure 28. F0 peak location in CVC sequences with broad focus.
In the case of utterances in broad focus, F0 peaks were always found in the post-stressed syllable
(sp2) regardless of the language or the syllable structure. However it is relevant that for all
Spanish and English speakers except en3, F0 peaks were located earlier in contexts where the
stressed syllable had a CVC structure than in those where it had a CV structure (see Figures 27
and 28).
Figure 29. F0 peak location in CV sequences with narrow focus.
80
90
100
110
120
130
140
150
s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2
Hz
Syllable
F0 peak location (CVC broad focus)
en1
en2
en3
sp1
sp2
sp3
80
90
100
110
120
130
140
150
160
s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2
Hz
Syllable
F0 peak location (CV narrow focus)
en1
en2
en3
sp1
sp2
sp3
63
Figure 30. F0 peak location in CVC sequences with narrow focus.
In narrow focus contexts, F0 peaks are clearly moved towards the stressed syllable. In the case of
stressed CV syllables, although there is a clear contrast with broad focus declaratives, only two
of the English speakers and one of the Spanish speakers presented F0 peaks within the limits of
the stressed syllable. With stressed CVC syllables, all F0 peaks were located in the stressed
syllable, however.
The next step was determining if the duration of the H tones (i.e. from the F0 valley to the F0
peak) were the same in CV and CVC contexts both in broad and narrow focus declaratives.
Given that in CV contexts peaks were found earlier than in CVC contexts, the hypothesis that the
H tone has the same duration in both cases regardless of the segmental material could explain the
differences between peak locations. The following figure shows the peak locations (in ms with
respect to the initial rise) of the four subjects for each context:
80
90
100
110
120
130
140
150
160
170
s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s1 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2 s2
Hz
Syllable
F0 peak location (CVC narrow focus)
en1
en2
en3
sp1
sp2
sp3
64
Figure 31. F0 rise durations in CV and CVC sequences in broad and narrow focus.
As expected, the duration of the F0 rises in narrow focus were always shorter than those in broad
focus. Separate two-way factorial analyses of variance (ANOVA) were run for each subject with
syllable type as the first independent variable and focus type as the second independent variable.
Results showed that there was a main effect for focus type for all subjects (p < .001), but only
sp3 showed a main effect for syllable type (p < .001) and an interaction for focus and syllable
type (p = .048). These results show that differences in the duration of H are mainly the result of
notable contrasts between broad and narrow focus contexts, but not CV vs. CVC sequences.
3.5.4 DISCUSSION
These results suggest that generally the length of a high tone in both languages is independent
from the segmental material it is affiliated to. Figures in Table 10 schematize the contrast
between alignments:
0.1
0.15
0.2
0.25
0.3
0.35
Duration (ms)
Subject
Duration of F0 rise
CV broad
CVC broad
CV narrow
CVC narrow
sp1 sp2 sp3 en1 en2 en3
65
Table 10. Coupling relations in CV and CVC sequences in broad and narrow focus.
According to Articulatory Phonology (Goldstein et al. 2006), speech is formed by articulatory
gestures that are coordinated in time with one another. These coupling relations can be in-phase
(i.e. when gestures are happening synchronously, like in CV sequences) or in anti-phase (i.e.
when gestures are happening in sequentially, like in VC sequences). As captured in Table 10, all
coupling relations between Cs, Vs and tones being equal, the difference between L+H* (narrow)
L+H* in
CV and
CVC
sequences
L+>H* in
CV and
CVC
sequences
L L H
C
V
σ
ˈσ
σ
L L H
C
V
σ ˈσ σ
L L H
C
V
σ ˈσ σ
C
L L H
C
V
σ ˈσ σ
C
66
and L+>H* (broad) would come from the stiffness (parameter that controls movement duration)
of the H gesture. Although it is tentative, this analysis would also capture the differences in peak
alignment between CV and CVC sequences.
Current findings also coincide with Prieto and Torreira (2007). They posit that in pre-nuclear
pitch-accents in Peninsular Spanish the position of the F0 peak is clearly affected by changes in
syllable structure (CV vs. CVC). More specifically they find that in L+H* pitch accents with
closed stressed syllables, F0 peaks are more retracted into the syllable, while in the case of open
syllables peaks are located closer to end of the syllable. Results are relevant not only to
demonstrate that L.A. Spanish and English generally employ the same intonational strategy to
distinguish between broad and narrow focus in pre-nuclear position, but also to highlight the
importance of syllable structures when describing tone alignments. Under ideal circumstances,
the narrow vs. broad focus distinction should be noted as L+H* vs. L+>H*; however, segmental
material may not be enough for earlier peaks to be realized within the limits of the stressed
syllable since the H tone has always a similar length (e.g. like in Figure 29 for speakers en1, sp2
and sp3). Thus, although based on a strict ToBI labeling those cases should be considered
L+>H*, L+H* is more appropriate considering current results.
3.6. DECLARATIVES: NUCLEAR PITCH-ACCENTS
3.6.1 INTRODUCTION
Nuclear pitch-accents in Spanish and English are those found towards the end of an utterance
and they are normally perceived as the most prominent ones within the sentence (Hualde 2005).
Nuclear configurations in Mexican Spanish declaratives can have several shapes and Martín
Butragueño (2006) distinguishes between five different variants. The following table exhibits
those variants with their respective boundary tones, schematic representations and a brief
description of each of them.
67
Toneme -
Declaratives
(nuclear-pitch accent
+ boundary tone)
based on Martín
Butragueño (2006)
Representation Description
H*L%
32
Plateau throughout the stressed syllable
(H*) without any noticeable peak and a
final lowering due to a low boundary tone
(L%)
rise of 1.5 semitones or less
L+H* L%
F0 rise normally at the onset of the
stressed syllable with a peak within the
limits of the stressed syllable (L+H*) and
a final lowering due to a low boundary
tone (L%)
rise between 1.5 and 3 semitones
L+¡H* L%
Same configuration as the previous one
but with a higher F0 rise.
rise of more than 3 semitones
L+H* LH%
F0 rise normally at the onset of the
stressed syllable a peak within the limits
of the stressed syllable (L+H*) followed
by a lowering and a rise result of a
complex low-high boundary tone (LH%)
rise between 1.5 and 3 semitones
L* L%
33
F0 fall throughout the stressed syllable
(L*) followed by a low boundary tone
(L%)
lowering of semitones
Table 11. Declarative tonemes according to Martín Butragueño (2006).
In his study, the proportion of instances for each configuration was the following: H*L% -
17.6%, L+H* L% - 22%, L+¡H* L% - 36.5%, L+H* LH% - 3.7% and L* L% - 20.2%. It is
important to notice that although the so called “circumflex” configurations (i.e. those with a
rising falling movement) are the most characteristic ones in Mexican Spanish (Matluck 1951,
Quilis 1993, Sosa 1999, Martín Butragueño 2004, 2006, Hualde & Prieto, forthcoming), they are
68
not necessarily used by all speakers. In fact, in other studies on Mexican Spanish like De la Mota
and colleagues (2010) or Willis (2005) circumflex and downward configurations are equally
used. Willis (2003), Martín Butragueño (2004) and Hualde and Prieto (forthcoming) also notice
that circumflex patterns can be followed by a sustained F0 (!H% or M%) instead of a final fall.
English declaratives have been generally described as having a high nuclear pitch-accent (H*)
followed by a low boundary tone (L%) (Pierrehumbert 1980, Pierrehumbert & Beckman 1988,
Beckman & Hirschberg 1994). However, based on the figures found in the aforementioned
studies, the H* labeling used in English is similar to the L+H* label proposed for Spanish. In
fact, the same as for Spanish, this final H* is also described as being notably lower than
preceding ones due to downstep and in some cases it is realized as a fall. Finally, high rise
terminals (the so called ‘uptalk’), which are normally labeled as L* LH% (Fletcher et al. 2005,
Ritchart et al. 2014), are also possible in Southern California English .
3.6.2 METHODS
In order to distinguish between nuclear configurations, the rise (or fall) of the F0 in the stressed
syllable was calculated in semitones following Martín Butragueño’s (2006) criteria (Table 11).
Although this classification may not necessarily distinguish between different phonological
pitch-accents I will employ this procedure to compare the uses of the patterns described in the
aforesaid study with those found in L.A. Spanish and English. Participants produced the
following sentences in Spanish: Abre el armario ‘s/he opens the closet’, Amanda compra
bananas ‘Amanda is buying bananas’, anima a Amanda ‘s/he cheers Amanda up’, el niño bueno
come helado ‘the good boy is eating ice-cream’, la señora morena vende mandarinas ‘the
brunette woman is selling tangerines’, prepara limonada ‘s/he is preparing lemonde’; and the
following sentences in English: the woman in blue is selling bananas, Amanda is buying bananas
I buy a banana; for a total of 226 combinations of nuclear pitch-accent and boundary tones
(tonemes); 143 in Spanish and 83 in English.
69
3.6.3 RESULTS
The following figures show actual examples of all the nuclear configurations schematized in
Table 12:
Toneme
(nuclear-pitch
accent +
boundary
tone)
Example Description
H*L%
Amanda compra
bananas
‘Amanda is
buying bananas’
g2s8 - Adult
Early Bilingual
rise of 0.5
semitones
L+H* L%
Amanda compra
bananas
‘Amanda is
buying bananas’
g3s3 - L.A. Born
Bilingual
rise of 2
semitones
L+¡H* L% Anima a
Amanda
‘He cheers
Amanda up’
g2s3 - L.A. Born
Bilingual
rise of 4.3
semitones
Freq (Hz) 0
5000
F0 (Hz)
150
50
Amanda compra bananas
Freq (Hz) 0
5000
F0 (Hz)
200
75
Amanda compra bananas
Freq (Hz) 0
5000
F0 (Hz)
175
50
Anima a Amanda
70
L+H* LH%
I buy a banana
g5s10 - Spanish
Control
F0 rise followed
by a complex
LH boundary
tone.
L* L%
34
Amanda compra
bananas
‘Amanda is
buying bananas’
g4s1 - Adult
Late Bilingual
fall of 1.56
semitones
(L+H*) !H%
I buy a banana
g4s2 - Adult
Late Bilingual
Circumflex
pattern followed
by a final
sustained tone.
Table 12. Examples of nuclear pitch-accents in declarative sentences.
The phonetic contrast between the first and fifth figures in Table 12 reflects the necessity to
distinguish between H* L% and L* L%. While in the first one the F0 is maintained through-out
the stressed syllable until the L boundary tone is reached, in the fifth one there is a constant fall.
These L* L% configurations are also found in Estebas-Vilaplana and Prieto (2010), Gabriel and
colleagues (2010) and Robles-Puente (2011). L+H* L% and L+¡H* L% are phonetically similar
to H* L% but with bigger F0 rises. Another finding in regard to L+¡H* that is worth mentioning
Freq (Hz) 0
5000
F0 (Hz)
320
120
Amanda compra bananas
Freq (Hz) 0
5000
F0 (Hz)
300
125
I buy a banana
71
is that the F0 rise may begin even before the tonic syllable in the last word of the sentence is
reached as exemplified in Table 13:
35
Table 13: Examples of L+¡H* configurations with early F0 rise.
Martín Butragueño (2006) also finds this configuration in Mexican Spanish but he does not
notice a phonological contrast between patterns with earlier or later rising. Consequently he does
not use a specific labeling for such configuration. Figure 32 below summarizes the uses of each
toneme in both languages. Bars with horizontal patterns indicate that the tone is more likely to
happen in English; bars with vertical patterns indicate that the tone is normally part of the
Spanish inventory. Solid bars indicate that those tones can be found in both languages.
Amanda compra
bananas
‘Amanda is buying
bananas’
g4s2 - Adult Late
Bilingual
F0 rise starts in ba- (i.e.
before the stressed
syllable -NA-)
Amanda is buying
bananas
g4s2 - Adult Late
Bilingual
F0 rise starts in ba- (i.e.
before the stressed
syllable -NA-)
Freq (Hz) 0
5000
F0 (Hz)
350
75
Amanda compra ba NA nas
Freq (Hz) 0
5000
F0 (Hz)
350
75
Amanda is buying ba NA nas
72
Figure 32. Percentages of nuclear pitch-accents in declaratives.
In Spanish, H*L% was the most common pitch-accent type for the English Control Group and
the Adult Early Bilinguals (54.5% and 60% respectively), while L*L% was preferred by the L.A.
Born Bilinguals, the Adult Late Bilinguals and the Spanish Control Group (52.9%, 82.6% and
45.8% respectively). The English Control Group also shows a high percentage of L*L% tonemes
with 40.9% of the total; however, H*L% configurations only represent 8.6% of the total in the
Adult Late Bilinguals and 6.6% in the Spanish Control Group due to their preference for
sustained boundary tones (25%). In the Adult Early Bilinguals Group, L*L%, L+!H*L% and
!H% configurations represent 12.5% of the total each. Finally, The L.A. Born Bilinguals’ second
most used configuration is H*L% (23.5%) followed by L+H* L% (14.7%).
In English, H*L% was once again used more frequently by the English Control Group and the
Adult Early Bilinguals (86.3% and 55% respectively) than the rest. The L.A. Born Bilinguals,
like in Spanish, show more uses of L*L% than H*L% (41.1% vs. 29.4%). The Adult Late
Bilinguals do not present uses of H*L% but a high number of L*L% instances (58.3%). Finally,
the Spanish Control Group presents the same amount of instances (25%) for L*L%, H*L% and
!H%. No examples of ‘uptalk’ were found in the current data.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Sp En Sp En Sp En Sp En Sp En
Control
English
Adult early
bilinguals
L.A. born
bilinguals
Adult late
bilinguals
Control
Spanish
Nuclear pitch-accents (declaratives)
L+H* L%
L* L%
!H%
L+¡H* L%
H* L%
73
3.6.4 DISCUSSION
The first observation that has to be made is that, as it is the case for pre-nuclear pitch-accents,
almost all configurations are found in every group. However, the most extreme circumflex
pattern (L+¡H* in Martín Butragueño’s 2006 terminology) was used by the Adult Late Bilinguals
and the Spanish Control Group as expected, but never by the English Control Group. H*L% was
preferred by the English Control Group and the Adult Early Bilinguals while L*L% was more
common in the Adult Late Bilinguals and the Spanish Control Group. The L.A. Born Bilinguals
widely use both configurations with slight preference for L*L%. Relevantly, sustained boundary
tones (H!%), as well as L+¡H*, were used by speakers of all groups except the English Control
Group, indicating that such configuration might be exclusively available to Mexican Spanish
speakers and even marginally retained by early L2 English learners (The Adult Early Bilinguals).
Interestingly, such configurations were more common in their English than in their Spanish
productions. Finally, the only LH% boundary tone, characteristic of Mexican Spanish, was
produced by one the speakers in the Spanish Control Group in English. These two findings
demonstrate that intonational patterns that are exclusive of Mexican Spanish can be transferred
to L2 English. They also agree with Ericson (2007), who notices high uses of sustained boundary
tones in English by Mexican Americans in Texas as opposed to Anglos.
It is crucial to point out that Martín Butragueño (2006) is partially based on Martín Butragueño
(2004) and that his data does not necessarily contain exclusively broad focus declaratives. He
concludes that circumflex patterns in Mexican Spanish are generally related to narrow focus
constructions and answers. It is likely that the lower number of instances of circumflex
configurations in the current results is the result of the elicitation materials, which did not contain
narrow focus but broad focus declaratives. Nevertheless, instances of such configuration were
found in all the groups except the English Control Group.
3.7. IMPERATIVES
3.7.1 INTRODUCTION
This section describes the intonational properties of imperatives in L.A. Spanish and English.
One of the main differences between declaratives and imperatives in Spanish and English is that
74
sentences can be ambiguous in the former language but not in the latter. A sentence like abre el
armario ‘open the closet’ can be interpreted both as a declarative (he/she opens the closet) or as
an imperative (open the closet!). In English this ambiguity is not possible since an overt subject
is always required in the case of the declarative: I open the closet vs. Open the closet.
Nuclear configurations in commands in Mexican Spanish present a L+H* L% configuration
which, depending on the elocutionary force, may be upstepped (L+¡H* L%) according to de la
Mota and colleagues (2010). The same description is found in Peninsular Spanish (Robles-
Puente 2011) or Chilean Spanish (Ortiz et al. 2010). Robles-Puente (2011) shows that in
Peninsular Spanish, declaratives and imperatives can have the same intonational patterns. This
observation agrees with previous studies on the topic like Kvavik (1988) and Willis (2002). In
the same study it is also shown that in contexts where there is an extra need to express
imperativity phonetic modifications take place in order to distinguish between declaratives and
imperatives. These result in a higher F0 in imperatives than in declaratives, the use of earlier F0
peaks in imperatives, the use of H boundary tones in imperatives but not in declaratives and the
use of higher nuclear accents in imperatives than in declaratives.
In Robles-Puente (2011), nine participants produced a total of 103 declaratives sentences in
Peninsular Spanish. Elicitation materials and procedures were similar to the ones that are being
employed in the current experiment but in none of those declarative utterances the nuclear pitch
accent was higher than the pre-nuclear pitch-accent. In imperatives, in contrast, such
configurations emerged, indicating a clear contrast at the intonational level between the two
pragmatic meanings. The following figure captures the different distributions in declaratives and
imperatives after comparing the maximum F0 values of the initial pre-nuclear pitch-accent and
the nuclear pitch-accent:
75
Figure 33. Distributions of F0 Peak differences in Peninsular Spanish declaratives and
imperatives.
As can be noticed here, in the case of declaratives in Peninsular Spanish, the pre-nuclear pitch-
accent was never higher than the nuclear pitch-accent (i.e. the difference was always positive). In
contrast, in imperatives, nuclear pitch-accents were frequently higher than pre-nuclear ones.
Various questions arise; Are the same intonational patterns employed in Peninsular Spanish and
L.A. Mexican Spanish in imperatives?, does the distinction between declaratives and imperatives
found in Peninsular Spanish hold in Mexican Spanish? Also, given that English makes use of
syntactic differences, are intonational implementations required to distinguish between both
kinds of sentences?
-100 -80 -60 -40 -20 0 20 40 60
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
Pre-Nuclear Pitch-Accent F0 - Nuclear Pitch-Accent F0
Density
SP Statement
SP Statement fit
SP Imperative
SP Imperative fit
76
3.7.2 METHODS
Speakers belonging to the five groups in the L.A. community were asked to produce the Spanish
sentence abre el armario ‘open the closet’ in four different pragmatic contexts:
36
a neutral
declarative (st), an imperative (imp1), a repetition of an imperative (imp2) and a second
repetition of an imperative (imp3). The same procedure was used for English with the sentences I
buy a banana and buy a banana!. A total of 98 sentences were produced in Spanish (23 st, 25
imp1, 25 imp2 and 25 imp3) and 119 in English (27 st, 31 imp1, 31 imp2, 30 imp3).
In Robles-Puente (2011) one of the most relevant distinctions between declaratives and
imperatives was the difference in F0 values between pre-nuclear and nuclear pitch-accents.
While in declaratives pre-nuclear pitch-accents were always higher than nuclear pitch-accents, in
imperatives nuclear pitch-accents generally showed higher values than preceding F0 peaks. In
this study utterances with nuclear pitch-accents 5 Hz lower than the pre-nuclear pitch-accents
were labeled as utterances with downstep. Nuclear pitch-accent that were 5 Hz higher than the
pre-nuclear pitch-accent were labeled as utterances with upstep. Finally, utterances in between
(with similar F0 values in the pre-nuclear and nuclear pitch-accents) were labeled as utterances
with F0 maintenance.
37
While Martín Butragueño (2006) bases his description on values related
to the nuclear-pitch accent without taking into account the rest of the sentence, Robles-Puente
(2011) demonstrates the relevance of looking at the global level of the F0 contour when
comparing between declaratives and imperatives.
The following table summarizes the overall F0 contours found in the data combining Robles-
Puente (2011) and Martín Butragueño’s (2006) descriptions with the labels I will employ to refer
to them:
38
77
Label Representation Description
Fall The nuclear pitch-accent is realized as
continuous F0 fall (L*L% in Martín
Butragueño’s terminology) from the pre-
nuclear pitch accent.
Downstep
The nuclear pitch-accent shows a small F0 rise
(typically H*L% or L+H*L% in Martín
Butragueño’s terminology), with a F0 peak
lower (at least -5Hz) than that of the pre-
nuclear pitch-accent.
Maintenance
The nuclear pitch-accent shows a notable F0
rise (typically L+H*L% or L+¡H*L% in
Martín Butragueño’s terminology), with a F0
peak with a similar value (between +/- 5 Hz) to
that of the pre-nuclear pitch-accent.
Upstep
The nuclear pitch-accent shows a notable F0
rise (typically L+¡H*L% in Martín
Butragueño’s terminology), with a F0 peak
with a higher value (at least +5Hz) than that of
the pre-nuclear pitch-accent.
Table 14. Intonational contours in declarative and imperative contexts.
7.7.3 RESULTS
Apart from the four configurations described in the previous section another two modifications at
tone level were detected (L+H* and H%). The following figures summarize the percentage of
contours found per context and group in both languages.
Figure 34. Percentage of intonational contours in declarative context per group and language.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Sp En Sp En Sp En Sp En Sp En
Control
English
Adult early
bilinguals
L.A. born
bilinguals
Adult late
bilinguals
Control
Spanish
Declarative context
upstep
maint.
downstep
fall
78
Figure 35. Percentage of intonational contours in imp1 context per group and language.
Figure 36. Percentage of intonational contours in imp2 context per group and language.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Sp En Sp En Sp En Sp En Sp En
Control
English
Adult early
bilinguals
L.A. born
bilinguals
Adult late
bilinguals
Control
Spanish
Imp1 context
H%
L+H*
upstep
maint.
downstep
fall
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Sp En Sp En Sp En Sp En Sp En
Control
English
Adult early
bilinguals
L.A. born
bilinguals
Adult late
bilinguals
Control
Spanish
Imp2 context
L+H*
upstep
maint.
downstep
fall
79
Figure 37. Percentage of intonational contours in imp3 context per group and language.
Vertical lines in the declarative context indicate that those tonal configurations are part of the
Mexican Spanish intonational system. Notice that in the rest of the contexts (imperatives) these
lines are not present since they are expected in English as well. Tonal variation is found not only
between groups but also between languages and contexts. This is not surprising given that
imperative productions show noticeable tonal variability in other Spanish varieties as well (see
Robles-Puente 2011 for Peninsular Spanish). Generally, while patterns with maintenance and
upstep are scarce or non-existent in declaratives and Imp1, productions with falling patterns are
common. In Imp2 and Imp3, the situation is the opposite since falling patterns are barely used
and the use of maintenance and upstep rises. In the current data many of the imperative
productions had H*, L*+H and L+>H* followed by L* L% and L+H* L% (or !H%) tonemes,
resulting in what I have labeled as fall and downstep configurations in Table 14. These
configurations are frequently found in declaratives as well; however, speakers produced higher
F0 contours in imperatives. The following figures show the average F0 values for all the
productions of abre el armario and I buy a banana / buy a banana! in the four different contexts
by all participants in configurations with downstep (49 in Spanish and 36 in English) or a final
fall (23 in Spanish and 32 in English). b1 marks the beginning of the F0 valley in the first pre-
nuclear pitch-accent, p1 marks the first F0 peak, b2 marks the beginning of the F0 valley in the
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Sp En Sp En Sp En Sp En Sp En
Control
English
Adult early
bilinguals
L.A. born
bilinguals
Adult late
bilinguals
Control
Spanish
Imp3 context
L+H*
upstep
maint.
downstep
fall
80
nuclear pitch-accent and p2 the peak in the nuclear pitch-accent (or in the case of L-L% the
middle of the nuclear syllable since there is no peak but a fall):
Figure 38. F0 means in downsteped configurations (Spanish).
Figure 39. F0 means in configurations with fall (Spanish).
Figure 40. F0 means in downsteped configurations (English).
100
150
200
250
300
b1 p1 b2 p2
F0
Average F0 in configurations with downstep (Spanish)
st
imp1
imp2
imp3
100
150
200
250
300
350
b1 p1 b2 p2
F0
Average F0 in configurations with fall (Spanish)
st
imp1
imp2
imp3
100
150
200
250
b1 p1 b2 p2
F0
Average F0 in configurations with downstep (English)
st
imp1
imp2
imp3
81
Figure 41. F0 means in configurations with fall (English).
Results demonstrate that although intonational contours may resemble those found in
declaratives, imperatives are produced with higher F0 values especially in contexts where an
additional volitional impulse is required (imp2 and imp3).
39
Due to the number of utterances
produced with other configurations it was not possible to divide the productions by group;
however, individual analyses demonstrated that generally all participants produced higher F0
values in imperatives than in declaratives independently from the group they belonged to, the
language they were producing or the overall shape of the intonational contour. Consequently, the
strategy of rising the overall pitch to denote imperativity is shared by both languages. The same
strategy is also noted in Robles-Puente (2011) for Peninsular Spanish.
As discussed above, none of the declaratives in Peninsular Spanish in Robles-Puente (2011)
presented configurations where the nuclear pitch-accent was higher than or as high as the first
pre-nuclear pitch accent. Figure 34 reveals that, in the current data, declaratives can be produced
with maintenance or upstep by speakers in groups other than the English Control Group (i.e. the
English Control Group is the only one where declaratives are always either ‘fall’ or ‘’downstep’
in both languages). A declarative with maintenance was produced by an Adult Early Bilingual
speaker in Spanish, another one by a L.A. Born Bilingual speaker in English and three
declaratives with upstep were produced by the Adult Late Bilinguals and the Spanish Control
Groups in English. In order to provide more data, all the initial pre-nuclear pitch-accents and
nuclear pitch-accents of the declarative sentences analyzed in section 3.3 and 3.6 (with the
exception of those ending in L* L% due to the lack of a nuclear F0 peak) were also measured. A
total of 135 pairs of F0 peaks were compared (79 in Spanish and 56 in English).
100
150
200
250
b1 p1 b2 p2
F0
Average F0 in configurations with fall (English)
st
imp1
imp2
imp3
82
This analysis at the global intonational level of the sentence revealed the production of sixteen
utterances with upstep divided as follows: one utterance in English by an Adult Early Bilingual
speaker, two utterances in Spanish and three in English by three L.A. Born Bilinguals, two
utterances in Spanish and three in English by two Adult Late Bilinguals and three utterances in
Spanish and two in English by two Spanish Control Group speakers. These utterances are
marked in bold in Table D4 in Appendix D. The pair of figures below is an example of the
downstep vs. upstep contrast. Both sentences were produced in a broad focus context but notice
how in the utterance to the left the highest F0 peak in ‘Amanda’ is clearly higher than that in
‘bananas’ due to a gradual lowering of the pitch. In contrast, in the utterance to the right,
‘bananas’ shows a much higher F0 peak than ‘Amanda’:
Figure 42. Downstep vs. upstep contrast in the sentence ‘Amanda is buying bananas’.
The following two figures present boxplots of the differences in F0 in declaratives between the
initial pre-nuclear pitch-accents and the nuclear pitch-accents for each group in each language:
0
F0 (Hz)
200
50
Amanda is buying bananas
0
F0 (Hz)
275
125
Amanda is buying bananas
83
Figure 43. ∆ F0 = nuclear pitch accent – pre-nuclear pitch-accent (English declaratives).
In English (Figure 43), the English Control Group does not present any instances where nuclear
F0 peaks are higher than pre-nuclear peaks. In contrast, the L.A. Born Bilinguals, the Adult Late
Bilinguals and the Spanish Control Group present such configurations, resulting in boxplots that
spread from negative to positive values with means that are closer to 0. Finally, The Adult Early
Bilinguals has an average between the English Control Group and the rest of the groups. The
same observations apply to Spanish:
Figure 44. ∆ F0 = nuclear pitch accent – pre-nuclear pitch-accent (Spanish declaratives).
The following figure presents the distributions of the five groups after collapsing all the
productions (Spanish and English):
84
Figure 45. ∆ F0 = nuclear pitch accent – pre-nuclear pitch-accent (English and Spanish
declaratives).
The same procedure was applied to imperative productions. As captured in figures 46 and 47
(collapsed in Figure 48) all groups behaved similarly and produced nuclear pitch-accents that
were lower than pre-nuclear pitch-accents in some cases and higher in other cases; however, the
number of instances with negative values was much higher in imperatives than declaratives.
Thus, although the number of productions with maintenance and upstep grows in imperative
contexts, the intonational constraint that regulates the differences in F0 between pitch-accents
depending on the pragmatic meaning applies to the English Control Group and English but not to
Mexican Spanish.
Figure 46. ∆ F0 = nuclear pitch accent – pre-nuclear pitch-accent (English imperatives).
85
Figure 47. ∆ F0 = nuclear pitch accent – pre-nuclear pitch-accent (Spanish imperatives).
Figure 48. ∆ F0 = nuclear pitch accent – pre-nuclear pitch-accent (English and Spanish
imperatives).
3.7.4 DISCUSSION
Although Spanish means are slightly higher than English ones, all groups behaved similarly
when producing imperatives and multiple instances of nuclear pitch-accents that were higher
than pre-nuclear pitch-accent were found. In the case of the English Control Group’s
declaratives, there are no examples where the nuclear pitch-accent is higher than the first pre-
nuclear pitch-accent, however. Like in Peninsular Spanish, this results in a distribution where 0
(no difference in height between pitch-accents) can be considered an intonational barrier. A
similar behavior but not as absolute is found for the Adult Early Bilinguals. The remaining three
86
groups exhibit productions more similar to imperatives in Peninsular Spanish, demonstrating that
declaratives in L.A. Mexican Spanish have intonational properties different not only from L.A.
American English but also from Peninsular Spanish. These results also show that the L.A. Born
Bilinguals, the Adult Late Bilinguals and the Spanish Control Group transfer the lack of
restrictions on pitch-accent differences to English, while the Adult Early Bilinguals show an
intermediate behavior closer to the English Control Group indicating an intonation closer to that
of English.
It is important to remember that the intonational constraint that prevents nuclear pitch-accents
from being higher than pre-nuclear ones is active in declaratives not only in English but in other
Spanish varieties like Peninsular Spanish (Robles-Puente 2011) creating three different ways to
distinguish between declaratives and imperatives as summarized in Table 15:
Declaratives vs.
imperatives
Grammatical
Difference
Lower F0 in nuclear
pitch-accents
Higher F0 in nuclear
pitch-accents
English Yes declaratives or
imperatives
imperatives
Mexican Spanish No declaratives or
imperatives
declaratives or
imperatives
Peninsular Spanish No declaratives or
imperatives
imperatives
Table 15: Distinctions between declaratives and imperatives.
In English the grammatical difference (subject vs. no subject) suffices to make the distinction but
there is an option to add an intonational cue with high nuclear pitch-accents to mark imperatives.
The same option is used in Peninsular Spanish. Finally, in Mexican Spanish, no distinction is
made between the two kinds of sentences in this respect.
Apart from these widely used strategies, where pre-nuclear pitch-accent structures in imperatives
did not differ from the ones presented for declaratives in section 3.3, the tonal alignment
characteristic of narrow focus (L+H*) was used in six cases. This strategy was also noted in
Robles-Puente (2011) for Peninsular Spanish and Willis (2002) for Mexican Spanish. Finally,
like in Robles-Puente (2011), high boundary tones (H%) were found in five imperative contexts
87
as well. These two configurations were scarcely used by different participants belonging to
different groups in both languages.
Willis (2002) posits that there is undeniable intonational connection between declaratives and
imperatives in Spanish. Data on L.A. productions confirm this observation because even
configurations that have been identified as characteristic of Mexican Spanish declaratives (i.e.
circumflex contours and !H% boundary tones) can be found in imperative contexts as well.
3.8. WH- QUESTIONS
3.8.1 INTRODUCTION
The intonational configuration of wh- questions in Spanish resembles that of declaratives
(Navarro Tomás 1944, Quilis 1993, Sosa 2003, Hualde & Prieto forthcoming). According to
Sosa (2003), the default nuclear configuration of wh- questions in Mexican Spanish (and other
varieties like Venezuelan or Colombian) shows a final fall (L* L%). De la Mota and colleagues
(2010) find a final low boundary tone as well and categorize the nuclear pitch accent as L+H*
HL%. Apart from falling contours, high boundary tones (H%) have been attested in wh-
questions in Spanish as well (Quilis 1993, Orozco 2008, Prieto & Roseano 2010, Torreira et al.
2012, Hualde & Prieto forthcoming). According to Quillis (1993) and Sosa (2003), the contrast
between the low boundary tone (L%) and the high one (H%) is that the latter may indicate a
higher level of politeness. Hualde and Prieto (forthcoming) do not rule out this subtle contrast in
meaning between L% and H% but suggest that they might be in free variation. Similarly, Kohler
(2004) hypothesizes that in English the same variation between low and high boundary tones is
found in wh- questions.
3.8.2 METHODS
Participants produced a total of four wh- question each: ¿dónde venden mermelada?/where do
they sell marmalade?, ¿qué hora es?/what time is it? for a total of 115 analyzable tonemes (53 in
Spanish and 62 in English). Two different pitch-accents were found in initial pre-nuclear
position; namely L+>H* and H*. The variety of tonemes was richer and showed different
88
combinations of nuclear pitch-accents and boundary tones: L* L%, L* H% (or HH%) and L+H*
L%. Table 16 describes each of the tonemes and Table 17 shows some examples:
Toneme – Wh-
questions
(nuclear-pitch accent
+ boundary tone)
Schematic
representation
Description
L* L%
F0 fall throughout the stressed syllable
followed by a low boundary tone.
L* H% Low tone thorough-out the last stressed
syllable of the utterance and an F0 rise at
the end of it.
L+H* L%
F0 rise in the last stressed syllable with
peak within its limits followed by a fall.
Table 16. Tonemes found in Wh-questions.
3.8.3 RESULTS
Although there were no instances of L* L% by the English Control Group in English and the
Spanish Control Group in Spanish, Figure 49 shows how all groups used low (L%) and high
(H%) boundary tones in similar amounts to produce wh- questions in both languages.
ˈσ
ˈσ
ˈσ
89
Toneme Example Description
L* L%
g2s8 – F0 fall
throughout the
stressed syllable
-la-
L+>H* L*
H%
g3s11 – Low
tone in es
followed by a
final F0 rise.
L+>H*
L+¡H*
L%
g4s2 – F0 rise
in the stressed
syllable -la- (in
this case with
upstep)
followed by a
low boundary
tone.
Table 17. Examples of tonemes found in Wh-questions.
Freq (Hz) 0
5000
F0 (Hz)
200
75
donde venden mermelada
Freq (Hz) 0
5000
F0 (Hz)
300
150
que hora es
Freq (Hz) 0
5000
F0 (Hz)
350
200
donde venden mermelada
90
Figure 49. Percentage of tonemes in Wh- questions per group and language.
3.8.4 DISCUSSION
Previous studies on both languages as well as current results strengthen the idea that the two
boundary tones (L% and H%) might be in free variation not only in English but in Spanish and
show that there are no language or group preferences in this pragmatic context.
3.9. YES/NO QUESTIONS
3.9.1 INTRODUCTION
Hualde (2005), Robles-Puente (2011) and Hualde and Prieto (forthcoming) highlight the
importance of intonation to distinguish between yes/no questions (also known as absolute
questions) and declaratives in Spanish since both kinds of sentences can be identical otherwise
(e.g. Tienes mermelada – ‘You have marmalade’ vs. ¿tienes mermedala? – ‘Do you have
marmalade?’)
Spanish yes/no questions are frequently characterized by a final F0 rising (Sosa 1999, Prieto &
Roseano 2010, O’Rourke 2010, Ortiz-Lira et al. 2010, De la Mota et al. 2010); however, L%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Sp En Sp En Sp En Sp En Sp En
Control
English
Adult early
bilinguals
L.A. born
bilinguals
Adult late
bilinguals
Control
Spanish
Wh- questions
L+H* L%
L* H%
L* L%
91
boundary tones have also been attested in multiple varieties, mainly concentrated in the
Caribbean and the Northern part of Spain (Prieto & Roseano 2010, López-Bobo & Cuevas-
Alonso 2010, Robles-Puente 2011, inter alia).
Sosa (1999), Avila (2003) and de-la-Mota and colleagues (2010) notice that Mexican Spanish
makes use of a final high boundary tone (H%) preceded by a low tone (L*). According to the
latter, this variety is characterized for a later rise in information-seeking yes-no questions and
consequently they label the boundary tone as LH%. They note that contrastive HH% boundary
tone with an earlier rise is found in invitation yes/no questions; however, the authors conclude
that more data is necessary to elucidate if these two boundary tones are actually marking a
categorical phonological contrast.
40
A third toneme discussed by the authors is L+H*, but
according to them is not found in their samples in Mexican Spanish. For English, according to
Levis (1999), yes/no questions in American English are normally produced with a high boundary
tones, nevertheless low boundary tones are attested as well and speakers do not distinguish
between them. Finally, Hedberg and colleagues (2004), find that 81% of positive yes/no
questions in English are produced with a high boundary tone, 62% of which correspond to
L*HH% patterns. Thus, a priori, both languages should show a large number of H% boundary
tones in this pragmatic context.
3.9.2 METHODS
In the current study speakers produced four kinds of yes/no questions in Spanish and English: A
confirmation yes/no question (¿tienes frío? / are you cold?), an information seeking yes/no
question (¿tienen mermelada? / do you have lemonade?), an invitation yes/no question (¿quieren
caramelos? / do you want some candy?) and an imperative yes/no question (¿se pueden callar? /
could you lower the volume?). Four main tonemes (summarized in Table 18) were attested in
their 232 productions (108 in Spanish and 124 in English): L+H*H%, L*H%, H+L*L% and H*
H%.
41
92
Toneme – Yes/no
questions
(nuclear-pitch
accent + boundary
tone)
Schematic
representation
Description
L+H* H%
F0 rise from the last stressed syllable of
the utterance until the end of the sentence.
L* H% Low tone thorough-out the last stressed
syllable of the utterance and an F0 rise at
the end of it.
H+L* L%
F0 fall from the beginning of the stressed
syllable followed by a low boundary tone.
H* H%
Sustained F0 thought out the stressed
syllable to the end of the utterance.
Table 18. Tonemes found in Yes/no questions.
3.9.3 RESULTS
Results confirm that high boundary tones are widely used in all kinds of yes/no questions both in
Spanish and English. However, differences in the alignment of the F0 rise showed variation and
seemed to be more of phonetic than of phonological nature. All tonemes were found in at least
two of the pragmatic contexts, L* H% being the most commonly used in all most of them as
reflected in Table 20. The only exception was found with the sentence ‘are you cold?’, where the
last word of the sentence was monosyllabic and the F0 rise could not happen later (resulting in
L+H*H% instead of L* H%) unless lengthening occurred. Other sentences and its Spanish
ˈσ
ˈσ
ˈσ
ˈσ
93
counterpart suggest that L* H% would have been more frequently attested otherwise. L+H* was
the second most commonly used configuration. Finally, H* H% was sporadically used for
imperative yes/no questions while H+L* L% was used in all four pragmatic contexts in a few
cases.
Context and
toneme
Example Description
Invitation
L+H* H%
g2s4 - F0 rise from
the stressed syllable
-me-
Invitation
L* HH%
g4s3 - F0 rise from
the post-stressed
syllable –los
confirmation
H+L*L%
g5s9 – F0 fall from
the stressed syllable
fri-
Freq (Hz) 0
5000
F0 (Hz)
200
75
quieren caramelos
Freq (Hz) 0
5000
F0 (Hz)
200
75
quieren caramelos
Freq (Hz) 0
5000
F0 (Hz)
200
75
tienes frio
94
confirmation
L+H* H%
g3s2 - F0 rise from
the stressed syllable
fri-
confirmation
H*H%
g2s6 - High F0
throughout the
stressed syllable to
the end of the
utterance
imperative
H+L* L%
g4s3 - F0 fall from
the stressed syllable
-llar
Freq (Hz) 0
5000
F0 (Hz)
350
75
tienes frio
Freq (Hz) 0
5000
F0 (Hz)
250
75
are you cold
Freq (Hz) 0
5000
F0 (Hz)
200
75
se pueden callar
95
imperative
L+H*HH%
g2s6 - F0 rise from
the stressed syllable
–llar.
information
seeking
H+L* L%
g3s2 - F0 fall from
the stressed syllable
le-
information
seeking
L* H%
g1s11 - F0 rise
from the end of the
stressed syllable le-
Table 19. Examples of tonemes found in Yes/no questions.
Freq (Hz) 0
5000
F0 (Hz)
300
75
se pueden callar
Freq (Hz) 0
5000
F0 (Hz)
200
75
do you have lemonade
Freq (Hz) 0
5000
F0 (Hz)
300
100
do you have lemonade
96
Figure 50. Percentage of tonemes in Wh- questions per group and language.
Table 20. Number of tonemes in Wh- questions per question type.
3.9.4 DISCUSSION
Recordings were done using very similar materials and procedures to de-la-Mota and colleagues
(2010) so it is unlikely that differences are the result of the elicitation technique. As is the case
for wh- questions, no language or group preferences were found and variation within speakers
and types of yes/no questions also support a case of free variation with an overall preference for
L*+(H)H% tonemes.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Sp En Sp En Sp En Sp En Sp En
Control
English
Adult early
bilinguals
L.A. born
bilinguals
Adult late
bilinguals
Control
Spanish
Yes/No - questions
H* H%
H+L* L%
L* H%
L+H* H%
tonemes
Yes/No-
questions Sp
L+H*H
%
L* H% H+L*
L%
H*
H%
tonemes
Yes/No
questions En
L+H*H
%
L* H% H+L*
L%
H*H%
confirmation 9/27
33.3%
16/27
59.2%
2/27
7.4%
0 confirmation 25/31
80.6%
2/31
6.5%
0 4/31
12.9%
absolute 11/27
40.8%
16/27
59.2%
0 0 absolute 3/31
9.7%
26/31
83.8%
1/31
3.2%
1/31
3.2%
invitation 10/27
37%
16/27
59.2%
1/27
3.7%
0 invitation 10/31
32.2%
20/31
64.5%
1/31
3.2%
0
imperative 9/27
33.3%
15/27
55.6%
3/27
11.1%
0 imperative 4/31
12.9%
25/31
80.6%
2/31
6.5%
0
97
3.10. VOCATIVES
3.10. 1 INTRODUCTION
The last pragmatic context under consideration is vocatives. According to Prieto and Roseano
(2010) vocatives can be used for various grammatical purposes (calling somebody, ordering,
etc.) and the most common configuration is Spanish is L+H* !H (or L+H* M%). This is
considered the most neutral form of vocative to call someone. In cases where there is a
repetition, insistence or some level of imperativeness, L+H* HL% configurations are found as
well. Ladd (1978, 1980) and Pierrehumbert (1980) describe English vocatives as a rise-fall
followed by a sustained plateau. This configuration is analyzed as H*+L H-L% which
corresponds to the label L+H* !H% in the Spanish ToBI system.
3.10. 2 METHODS
In order to obtain instances of vocatives with different strengths, participants were asked to call a
friend twice (initial vocative and repetition). In the Spanish and English productions of the L.A.
community, four different tonemes (nuclear + boundary tone) were found after producing a total
of 112 vocatives (50 in Spanish and 62 in English). These are schematized in Table 21 and
exemplified in Table 22 in the next subsection.
Toneme – Vocatives
(nuclear-pitch accent
+ boundary tone)
Schematic
representation
Description
L+H* (H)H%
F0 rise from the last stressed syllable of
the utterance until the end of the sentence.
L+H* !H% F0 rise and peak in stressed syllable
followed by a sustained boundary tone.
ˈσ
ˈσ
98
L+H* L%
F0 rise from the beginning of the stressed
syllable with a peak within its limits and
followed by a low boundary tone.
L* HL%
F0 rise from the beginning of the stressed
syllable with a peak outside its limits and
followed by a low boundary tone.
Table 21. Tonemes found in vocatives.
3.10.3 RESULTS
Toneme Example Description
L+H* HH*
g5s9 - F0 rise
from the stressed
syllable -man- to
the end of the
utterance
g1s5 - F0 rise
from the stressed
syllable -ri- to
the end of the
utterance.
Freq (Hz) 0
5000
F0 (Hz)
250
75
Amanda
Freq (Hz) 0
5000
F0 (Hz)
350
150
Marina
ˈσ
ˈσ
99
L+H* !H%
g1s12 – F0 rise
from the stressed
syllable -man-
followed by a
sustained F0
until the end of
the utterance.
g4s3 - F0 rise
from the stressed
syllable -ri-
followed by a
sustained F0
until the end of
the utterance.
L+H* L%
g1s9 - F0 rise
from the stressed
syllable -man-
with a peak
within its limits
followed by a
fall.
g2s4 - F0 rise
from the stressed
syllable -ri- with
a peak within its
limits followed
by a fall.
Freq (Hz) 0
5000
F0 (Hz)
400
200
Amanda
Freq (Hz) 0
5000
F0 (Hz)
200
75
Marina
Freq (Hz) 0
5000
F0 (Hz)
250
75
Amanda
Freq (Hz) 0
5000
F0 (Hz)
150
50
Marina
100
L* HL%
g5s2 - F0 rise
from the stressed
syllable -man-
with a late peak
followed by a
fall.
g3s1 - F0 rise
from the stressed
syllable -ri- with
a late peak
followed by a
fall.
Table 22. Examples of tonemes found in vocatives.
Figure 51. Percentage of tonemes in vocatives per group and language.
Freq (Hz) 0
5000
F0 (Hz)
250
100
Amanda
Freq (Hz) 0
5000
F0 (Hz)
350
125
Marina
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Sp En Sp En Sp En Sp En Sp En
Control
English
Adult early
bilinguals
L.A. born
bilinguals
Adult late
bilinguals
Control
Spanish
Tonemes (vocatives)
L* HL%
L+H*L%
L+H* !H%
L+H*H%
101
The English Control Group, agreeing with previous descriptions of English, showed a clear
preference for L+H* !H% in both languages (75% in Spanish and 68.75% in English). The most
interesting finding is that none of the utterances by this group presented the commonly used
Spanish L* HL% pattern (marked with vertical patterns in Figure 51). This contour was found in
all other groups in both languages, however. Most of the Adult Early Bilinguals’ productions
were either L+H* !H% (33.3% in Spanish or 56.25% in English) or L* HL% (50% in Spanish
and 25% in English), indicating a certain level of accommodation depending on the language.
The L.A. Born Bilinguals opted for a more neutral configuration in both languages (64.3% in
Spanish and 71.4% in English) since L+H* L% corresponds to a toneme characteristic of
declaratives as well as presented in section 3.6. As explained in Brown and Levinson (1987) and
more recently in Borràs-Comes and colleagues (2013), the use of certain contours in vocatives as
opposed to others can be used to indicate social distance or power. Given that members of the
L.A. Born Bilinguals Group are children and adolescents it is possible that the L+H* L% contour
is an indicator of their younger age. Finally, the last two groups used mainly L+H* !H% and L*
HL% contours with no clear preference in their vocatives.
Table 23 summarizes the number of tonemes found in each vocative type (initial vocative or
repetition). The most interesting finding is that the use of L* HL% rises for the second vocatives
in both languages (from 20% to 44% in Spanish and from 13% to 22.5% in English). This result
agrees with Prieto and Roseano (2010). In the case of L+H* !H% and L+H* L%, numbers are
roughly maintained or lowered in deference of L* HL%, indicating that they are frequently used
for both kinds of vocatives. Finally, L+H* H% configurations were not as common and only
three instances were found in each language in both the first and the second vocative.
Spanish L+H*H% L+H*!H% L+H*L% L* HL% English L+H*H% L+H* !H% L+H*L% L* HL%
1
st
vocative
1/25
4%
10/25
40%
9/25
36%
5/25
20%
1
st
vocative
3/31
9.5%
13/31
42%
11/31
35.5%
4/31
13%
2
nd
vocative
2/25
8%
9/25
36%
3/25
12%
11/25
44%
2
nd
vocative
0 15/31
48.5%
9/31
29%
7/31
22.5%
Table 23: Number of tonemes in vocatives per vocative type.
102
3.10.4 DISCUSSION
The characteristic L* HL% pattern of Mexican Spanish vocatives was produced by all groups
except the English Control Group, which showed a clear preference for the English L+H* !H%
contour in both languages. Interestingly, even the Adult Early Bilinguals showed multiple
instances with such configuration. The analysis of the productions based on the kind of vocative
confirms Prieto and Roseano (2010) observations since the use of L* HL% rises for the second
vocatives in both languages.
3.11. INTONATION: OVERALL RESULTS
Given the amount of variability in the use of tonal patterns found in all contexts as seen in
sections 3.3. through 3.10., exact contingency table analyses were run in order to establish
differences between groups and languages. The percentages of all tones where variation was
expected (H*, L+H*, H*L%, L* HL%, H!% and L+¡H*, maintenance and upstep) were
considered for the comparisons. The goal was to determine a) if there is a statistical difference in
the use of tones in Spanish as opposed to English within each group and b) if there is a statistical
difference in the use of tones in English and Spanish between all the groups and the controls.
The comparison of the tones produced in Spanish and English within each group demonstrates
that none of the first three groups accommodate their intonation since differences between
languages were not significant (English Control Group: p = 0.461; Adult Early Bilinguals: p =
0.397; L.A. Born Bilinguals: p = 0.117). Interestingly, significant differences were found in the
Adult Late Bilinguals (p = 0.009) and the Spanish Control Group (p = 0.006). A closer
inspection indicates that what could look like an accommodation goes in the opposite direction
than would be expected though. As mentioned in previous sections, many Mexican Spanish
tones were found more frequently in English than in Spanish.
The second step was to statistically determine whether groups behave like (quasi) monolinguals
(i.e. control groups) or not when it comes to intonation. The number of tones of each type
produced by each group in English was compared to those produced by the English Control
Group and the amount of tones produced in Spanish to those produced by the Spanish Control
103
Group. For English, all groups show statistically significant differences from the control (Adult
Early Bilinguals: p = 0.004; L.A. Born Bilinguals: p < .001; Adult Late Bilinguals: p < .001;
Spanish Control Group: p < .001), indicating that Mexican Spanish intonation is present in the
English productions of these speakers. For Spanish, both the L.A. Born Bilinguals (p = 0.184)
and the Adult Late Bilinguals (p = 0.287) show no differences with the Spanish Control Group.
In contrast, the English Control Group (p < .001) and the Adult Early Bilinguals (p = 0.003) do
show significant differences. It is particularly interesting that the Adult Early Bilinguals are
significantly different from both controls, indicating an intermediate intonational behavior.
3.12. INTONATION: GENERAL DISCUSSION
In the previous sections the English and Spanish productions of thirty one subjects have been
analyzed following the Autosegmental-Metrical (AM) model of intonational phonology
(Pierrehumbert & Beckman 1986, Ladd 1996, Gussenhoven 2004, Beckman et al. 2005, inter
alia). Their productions covered some of the most common pragmatic contexts: declaratives
(with broad focus and narrow focus), imperatives (with three different strengths), wh-questions,
yes-no questions and vocatives (with two different strengths). It can be argued that elicitation
materials may have affected the results; however they are very similar to those used in other
studies where differences between other Spanish varieties were found (Atlas Multimedia de la
Prosodia del Espacio Románico (AMPER) ‘The Multimedia Atlas of the Prosody of Romance
Space’ & Atlas Interactivo de la Entonación del Español ‘The Interactive Atlas of Spanish
Intonation’). The idea behind using prompted speech is getting comparable results among all the
speakers and languages and it is expected that by using natural productions, differences would
have been even clearer.
The first observation to be made is that although the two languages under consideration belong
to different linguistic families (Spanish-Romance vs. English-Germanic), a priori intonational
differences are not as salient as rhythmic ones. According to Bolinger (1986), pitch is used to
signal emotion and consequently intonational systems across languages are commonly shared.
Therefore, it is expected that in many cases intonational contours for a given pragmatic meaning
can coincide in both languages. This is the case for pre-nuclear pitch-accents in declaratives
104
where L+>H* and intermediate L*+H patterns are widely attested across groups and in
declaratives with narrow focus in the subject where both L+H* and L+>H* are frequent
regardless the group or the language. The same happens in wh- questions, where control groups
as well as the other three groups make use of the tonemes L*L%, L*H%, L+H* L% generally
proceeded by L+>H* pre-nuclear tones. In yes/no questions all groups show a preference for L*
H% tonemes, followed by L+H* H% and to a much lesser extent H+L* L% or H* H%. The
language they are speaking does not make a difference in this case either. Finally, imperatives in
both languages show higher F0 contours and configurations with higher nuclear accents.
More interesting are the findings in some broad focus declaratives, imperatives (when compared
to declaratives) and vocatives. In these pragmatic contexts we can find contours that are shared
by both languages but with clear preferences for a pattern over another depending on the group.
At the same time, cases where contours from a language are produced in the other language and
cases where certain groups never produced a given contour are also attested. All these findings
can shed light in the processes of acquisition, retention and attrition of pragmatic features in
general and of intonation in particular.
In the case of pre-nuclear pitch accents in declaratives, H*s are exclusively found in the English
Control Group, the Adult Early Bilinguals and the L.A. Born Bilinguals in both languages, but
never in the Adult Late Bilinguals or the Spanish Control Group. In contrast, L*+H accents (i.e.
with a late F0 rise) are only attested in the L.A. Born Bilinguals, the Adult Late Bilinguals and
the Spanish Control Group. These results demonstrate that L1 English speakers and early English
learners prefer H* over L*+H as opposed to L1 Spanish speakers indicating a) the attrition (or
incomplete acquisition according to Montrul 2008) of L*+H in the Adult Early Bilinguals and b)
the lack of acquisition of H* in the Adult Late Bilinguals and the Spanish Control Group. Very
interestingly, the L.A. Born Bilinguals speakers produce both kinds of pitch-accents (although
not frequently) independently of the language they are speaking. This suggests the existence of a
more varied tone inventory in younger speakers which is available in both languages.
For nuclear accents and boundary tones (tonemes) in declaratives, the English Control Group and
the Adult Early Bilinguals show a large number of productions in Spanish and English with
105
H*L% (English Control in Spanish: 54.5% and Adult Early Bilinguals in Spanish: 86.3%) and
L*L% (English Control in English: 60% and Adult Early Bilinguals in English: 55%). In
contrast, the Adult Late Bilinguals and the Spanish Control Group use L*L% more frequently
(Adult Late Bilinguals Sp 82.6%, En 58.3% & Spanish Control Group Sp 45.8%, En 25%) than
H*L% (Adult Late Bilinguals Sp 8.6%, En 0% & Spanish Control Group Sp 16.6%, En 25%).
In the case of the L.A. Born Bilinguals, L* L% configurations (Sp 52.9%, En 41.1%) show
slightly higher levels of use in both languages than H*L% (Sp 23.5%, En 29.4%). These
preferences indicate that the English Control Group and the Adult Early Bilinguals have a similar
behavior that is different from that of the Adult Late Bilinguals, the Spanish Control Group or
the L.A. Born Bilinguals. As discussed before, these last three groups turn out to be statistically
different from the English Control Group in English but the Adult Early Bilinguals and the L.A.
Born Bilinguals show no significant differences when compared to the Spanish Control Group in
Spanish. There is also evidence that the Adult Early Bilinguals do not always behave like the
English Control Group and consequently this is the only group that is significantly different from
both controls. Sustained boundary tones (H!%) and upstepped nuclear accents (L+¡H*) were
used by all groups except the English Control Group, indicating that configurations typical of
Mexican Spanish, are available not only to the L.A. Born Bilinguals, the Adult Late Bilinguals
and the Spanish Control Group, but also to heritage speakers whose Spanish has been gradually
lost in favor of English (i.e. the Adult Early Bilinguals). A similar conclusion can be reached if
the overall F0 contour of the sentence is analyzed. All groups show a similar number of
imperative productions where the nuclear pitch-accent is higher than the pre-nuclear one;
however, this is not the case for declaratives. It has been well documented that circumflex
contours are characteristic of Mexican Spanish but not of American English. A detailed analysis
of all the declaratives produced by all groups revealed that in none of the utterances recorded in
the English Control Group the nuclear pitch-accent showed higher values than the nuclear ones.
This was not the case for the L.A. Born Bilinguals, the Adult Late Bilinguals and Spanish
Control Group. Finally, the Adult Early Bilinguals also produced instances where the pre-nuclear
pitch-accent is lower than the nuclear one, however their frequencies in both Spanish and
English show an intermediate behavior that do not exactly coincide with those by the English
Control Group or those of the rest of the groups. Vocatives are another clear instance where
heritage speakers in the Adult Early Bilinguals Group show pitch configurations never attested in
106
the English Control Group. All groups present vocatives with L+H* !H% and L+H* L%.
However, L* HL% patterns, which are characteristic of Spanish but not English, are found in all
groups except the English Control Group, indicating once again the retention of the intonation of
the mother tongue even by adult early learners in the Adult Early Bilinguals Group. As pointed
out before, these observations were supported statistically (section 3.11.).
In the case of English, all groups showed significant differences when compared to the English
Control Group. The overall lack of English tonal patterns by the Adult Late Bilinguals and the
Spanish Control Group and the use of Mexican Spanish patterns by the Adult Early Bilinguals
and the L.A. Born Bilinguals resulted in these differences. In the case of Spanish, the L.A. Born
Bilinguals and the Adult Late Bilinguals behaved like the Spanish Control Group. The English
Control Groups and the Adult Early bilinguals showed significant differences when compared to
the Spanish Control Group due the higher use of English patterns, however.
There are interesting findings when it comes to a possible development of two separate prosodies
in bilingual speakers. A priori, especially for cases where intonational contours are clearly
different, bilingual speakers could be using different contours depending on the language they
were producing. This behavior was even more likely to happen in the L.A. Born Bilinguals due
to their ability to accommodate their rhythms. However, there are no clear cases where groups
have a Spanish intonation for Spanish and an English intonation for English. This observation
was supported statistically since differences between languages were not significant for the
English Control Group, the Adult Early Bilinguals or the L.A. Born Bilinguals. In the case of the
Adult Late Bilinguals and the Spanish Control Group, significant differences between languages
were found; however, these were not the result of accommodation since Mexican Spanish
patterns were being used more in English than in Spanish. This leads to a situation where any
available tone or contour in the grammar is used regardless of the language that is being
produced. According to Kroll and Hermans (2011), bilinguals have two grammars available
when speaking a given language; however, the grammatical system of the language that is not in
use is suppressed. In the case of bilinguals in the L.A. community, it seems that, probably due to
the vast inventory of tonal movements available in both languages, the suppression of given
tones is not possible. A good example is the aforementioned use of both H* and L*+H pre-
107
nuclear pitch-accents by the L.A. Born Bilinguals. Another example is the use of sustained
boundary tones (H!%) and L+¡H* nuclear tones by the Adult Early Bilinguals, the L.A. Born
Bilinguals, the Adult Late Bilinguals and the Spanish Control Group. These tones, which are a
priori exclusively used in Mexican Spanish, are more common in English productions. Similarly,
the only Mexican Spanish LH% boundary tone was found in an English utterance. These
findings lead to the conclusion that intonation can be transferred from the L1 to the L2 in the
form of pre-nuclear pitch accents (e.g. L+H*), nuclear pitch accents (e.g. L+¡H*), boundary
tones (e.g.!H%) or contours at the global level of the utterance (e.g. upstepped declaratives).
108
CHAPTER 4: DISCUSSION AND CONCLUSION: RHYTHM AND INTONATION COMPARED
In chapters 2 and 3 the rhythmic and intonational properties of the five groups under study have
been presented. This chapter summarizes the findings of both prosodic properties and compares
them hand in hand in order to establish similarities and differences between them both within
and between groups.
If rhythmic results are considered the base line to draw similarities between groups and all
prosodic properties undergo similar changes, it can be hypothesized that a) the English Control
Group and the Adult Early Bilinguals should present similar intonations in both languages, b) the
Adult Late Bilinguals and the Spanish Control Group should present similar intonations which
are different from the English Control Group and the Adult Early Bilinguals and c) the L.A. Born
Bilinguals should perform like the English Control Group and the Adult Early Bilinguals in
English and the Adult Late Bilinguals and the Spanish Control Group in Spanish. As has been
demonstrated in the previous section, only b) is a correct hypothesis. There is no case where the
Adult Late Bilinguals and the Spanish Control Group differ consistently from each other in either
language and they generally show similar amounts of use for each intonational pattern. It is also
clear that in cases where Spanish and English intonations differ, the Adult Late Bilinguals and
the Spanish Control Group make use of those belonging to their native language while the
English Control Group does not and vice versa. For example, the Spanish Control Group and the
Adult Late Bilinguals use L*+H, !H% or upstepped declaratives but never H*, while the opposite
behavior applies to the English Control Group. As is the case for rhythm, the intonation of their
mother tongue is not only preserved in Spanish but also transferred to English. Although their
rhythm switches, the L.A. Born Bilinguals do not really accommodate their intonations
depending on the language they are speaking. For example, they sporadically produced H* pre-
nuclear tones in Spanish and L*+H ones in English or !H% boundary tones and L* HL%
tonemes in English. The clearest example was the production of various upstepped declaratives
characteristic of Mexican Spanish in English, however. The low number of instances in some
pragmatic contexts might be an indicator that only some of the L.A. Born Bilingual speakers
make use of certain patterns. Another possibility is that these speakers have intonational patterns
from both languages in their grammar but they only use more neutral configurations if available.
109
It is nevertheless clear that their behavior is different from that in the English Control Group and
partially the Adult Early Bilinguals.
Considering the similarities in rhythm that the English Control Group and the Adult Early
Bilinguals have, it is surprising that in many cases the Adult Early Bilinguals perform more like
the Spanish Control Group than the English Control Group in intonation. They behave similarly
in pre-nuclear pitch-accents in declaratives since both of them produce H* but never L*+H. In
the case of nuclear pitch-accents differences are found, however. The Adult Early Bilingual
speakers produced various instances with L+¡H and !H%, while in the English Control Group
these two tones were never attested. In the case of global configurations, the Adult Early
Bilinguals’ productions were not identical to those by the English Control Group either. The
L.A. Born Bilinguals, the Adult Late Bilinguals and the Spanish Control Groups presented
several declaratives with nuclear pitch-accents higher than pre-nuclear ones; while in the English
Control Group these kinds of configurations were not found, Adult Early Bilingual speakers
produced them in various occasions, resulting in frequencies half way between control groups. A
final example that the Adult Early Bilinguals heritage speakers are still retaining some of the
intonational properties from Spanish is the use of L* HL% in vocatives, a pattern never used by
the English Control Group but by Mexican Spanish speakers. This intermediate behavior was
captured statistically since the productions of the Adult Early Bilinguals were different from both
the English Control Group and the Spanish Control Group.
The following four figures capture the asymmetries between rhythm and intonation. The group
nPVI scores are plotted against the percentages of Mexican Spanish and English tonal patterns
found both in English and Spanish. These percentages represent uses of tones that are expected
to be found in a given language but not in the other. Tones that are shared by both languages are
not considered. To show clearer results, percentages are based only on contexts where
differences in tonal patterns were expected. Thus, based on previous sections, 100% represents
the pre-nuclear and nuclear pitch-accents in declaratives, the global configurations in
declaratives and all the vocatives (American English tones include H* and H*L% in declaratives.
Mexican Spanish tones include L*+H, L+¡H L%, H%, maintenance and upstep in declaratives
and L* HL% in vocatives). So, for example, if group shows 40% of Mexican Spanish tones, it
110
means that 40% of their productions were tones that are expected in Mexican Spanish but not
American English (i.e. L*+H, L+¡H L%, H%, maintenance, upstep and L* HL%). The other
60% would be tones that are part of English or both the Mexican and English inventories.
Figure 52. English nPVI scores vs. % of English tones in English.
Figure 53. English nPVI scores vs. % of Mexican Spanish tones in English.
30
35
40
45
50
55
60
0 20 40 60
nPVI (English)
% of Am. English tones in English
Control English
Adult early bilinguals
L.A. born bilinguals
Adult late bilinguals
Control Spanish
30
35
40
45
50
55
60
0 10 20 30 40
nPVI (English)
% of Mexican Spanish tones in English
Control English
Adult early bilinguals
L.A. born bilinguals
Adult late bilinguals
Control Spanish
111
In English, the English Control Group, the Adult Early Bilinguals and the L.A. Born Bilinguals
show the highest nPVI scores indicating more stress-timed productions in the continuum of
rhythm. They also show higher percentages of English tonal patters than the Adult Late
Bilinguals and the Spanish Control Groups. It is important to highlight that English tones are
found in these last two groups because although the configuration H*L% is more common in
English, it is still possible in Spanish declaratives (section 3.6). These last two groups also show
higher percentages of Mexican tones in English than any of the other groups; however, it is
interesting that while the English Control Group has no traces of Mexican Spanish intonation,
both the Adult Early Bilinguals and the L.A. Born Bilinguals do, indicating a transfer process.
Figure 54. Spanish nPVI scores vs. % of English tones in Spanish.
Figure 55. Spanish nPVI scores vs. % of Mexican Spanish tones in Spanish.
30
32
34
36
38
40
42
44
46
0 10 20 30 40
nPVI (Spanish)
% of Am. English tones in Spanish
Control English
Adult early bilinguals
L.A. born bilinguals
Adult late bilinguals
Control Spanish
25
30
35
40
45
50
0 10 20 30
nPVI (Spanish)
% of Mexican Spanish tones in Spanish
Control English
Adult early bilinguals
L.A. born bilinguals
Adult late bilinguals
Control Spanish
112
In Spanish, we can see that the Spanish Control Group, the Adult Late Bilinguals and the L.A.
Born Bilingual avoid English tones in general. This is not the case for the English Control Group
or the Adult Early Bilinguals, which are precisely those groups with more stress-time rhythm in
Spanish. The percentage of Spanish tones in Spanish shows once again the asymmetry between
rhythm and intonation. Although the Adult Early Bilinguals have a rhythm close to that of the
English Control Group (which was statistically similar as pointed out in section 2), they show
traces of Spanish prosody when it comes to intonation.
Thomas and Ericson (2007) look at Mexican American intonation in Texas and North Carolina
in English from an acoustic point of view and posit that these varieties show higher proportions
of rising pitch-accents than those produced by Anglo speakers. They also compare the rhythmic
patterns of all participants in English and conclude that there is a clear different between Anglos
(with more stress-timed rhythm) and Mexican Americans (with more syllable-timed rhythm).
Their main conclusion is that intonation can definitely be used as a sociolinguistic variable,
however they point out that results showed more complex patterns with intonation than with
rhythm possibly due to factors like assimilation or identity. In the current study a similar trend
has been found; while rhythmic differences are quite well established, intonational uses show
more variation.
Our observations lead to the same conclusion as Thomas and Ericson (2007) since rhythm and
intonation do not necessarily change in the same fashion. If prosody was modified as a whole,
the rhythm and the intonation of a group should reflect the same characteristics and preferences.
These could be 100% Spanish-like, 100% English-like or something in between indicating a
mixed prosody. However, the prosodic analysis of some of the groups under study reveals that
prosodic changes can affect a feature more than the other and consequently that the two are not
equally modified. For example, the English and Spanish rhythms of the Adult Early Bilinguals
speakers are on the stress-timed side indicating a shift to English prosody; nevertheless, their
intonation in both languages still shows traces of contours characteristic of Spanish. Similarly,
speakers in the L.A. Born Bilinguals accommodate their rhythm to the language they are
producing (i.e. more syllable-timed in Spanish and more stress-timed in English) but still show
intonational configurations of English in Spanish and vice versa. This contrast can be linked to
113
the inherent properties of both prosodic features we referred to in the introduction (section 1).
We can also conclude that rhythm and intonation should be considered independent building
blocks of language prosody with their own characteristics.
This line of thought has already been captured by the hierarchy of domains proposed by the
theory of Prosodic Phonology (Selkirk 1980, 1984, Nespor & Vogel 1986, Hayes 1989,
Schiering et al. 2007, Prieto 2006) presented in the introduction and reproduced below again
(Figure 56).
the intonational phrase
|
the phonological phrase
|
the phonological word
|
the foot
|
the syllable
Figure 56. Hierarchy in Prosodic Phonology.
Based on the results presented in chapter 2, the contrasting relevance of the syllable and the foot
domains in Spanish and English may be the reason why different rhythms are found and clear
transfers in the L2 occur. As pointed out in the aforesaid chapter, English is constrained by the
necessity to avoid stress clashes and distributing stresses regularly. In the case of Spanish, this
kind of distribution is not found and syllables are considered the rhythmic units. The following
figures (Figure 57 and Figure 58) exemplify this contrast:
X X X
Figure 57. Spanish rhythm: 10 syllables with similar lengths and stresses (X) found at irregular
intervals.
114
X X X
Figure 58. English rhythm: 10 syllables of different lengths, but 3 feet with similar lengths and
stresses (X) found at regular intervals.
This contrast raises some questions regarding the prosodic hierarchy. For example, is the foot
domain necessary in Spanish at all? According to Harris (1983, 1991, 1992) and Roca (1988,
1997) penultimate stresses in Spanish are more common because there is a preference for
trochaic feet that are formed from right to left. Nevertheless, the role of the foot in Spanish to
create rhythm is questionable (Quilis 1975, Harris 1983, 1992, Roca 1988). Although further
research is necessary to confirm that quantitative differences in rhythm are the direct
consequence of the prosodic organization (syllable vs. foot), data presented in chapter 2 (i.e. the
similarities in vowel length) can lead to the idea that L1 Spanish speakers do not rely on feet to
measure rhythm in L.A. Spanish (or English in the case of the Adult Late Bilinguals and the
Spanish Control Groups). Similarly, L1 English speakers and heritage speakers who have lost
their Spanish in deference to English rely on the foot and would not consider the domain of the
syllable in their grammar at least for rhythmical purposes.
The assumption that the foot domain does not exist in Spanish would go against Nespor and
Vogel (1986) and support Selkirk’s (1980) proposal. According to the former all domains in the
prosodic hierarchy must be present in all languages and according to the latter all domains are
universal but languages may choose their own domains. An alternative explanation is that the
foot domain is actually part of the prosodic hierarchy in Spanish as well but it is not used or it is
skipped. This option would go against the Strict Succession Hypothesis, which states that none
of the levels of the hierarchy can be skipped. Although according to Nespor and Vogel (1986) all
the domains are universal and have to interact, there has been an ongoing debate on this respect.
For example, Hyman (1982) and Auer (1983) present Gokana and !Xóõ as languages where the
syllable domain does not play a role and Selkirk (1980) puts into question the universality of the
intonational phrase as it seems not to be a relevant domain in Sanskrit. Finally, Schiering et al.
115
(2007) provide multiple languages where the status of the phonological word is questionable as
well. Although this position would go against the universality of the domains or the Strict
Succession Hypothesis, it seems to be the case for the foot domain in Spanish as well. Based on
current findings, phonological processes might be taking place at different levels in English and
Spanish. If rhythmic phonological patterns happen at two different domains in these languages,
the L2 transfers would be satisfactorily explained as well (Figure 59):
the foot (English / L2 Spanish rhythm)
|
the syllable (Spanish / L2 English rhythm)
Figure 59. Foot and syllable domains.
Therefore it is plausible that bilingual speakers can give preference to one domain over another
for phonological processes to happen; the syllable in the case of the Adult Late Bilinguals and
the Spanish Control Group and the foot in the case of the English Control Group and the Adult
Early Bilinguals. The L.A. Born Bilinguals would still consider both domains; one for each
separate language.
42
In order to confirm these possibilities, further research that looks at the foot
organization of both languages in a more detailed and controlled way is necessary. Evidence for
foot structure could be found by analyzing polysyllabic shortening since the duration of
individual syllables in a foot decreases with the number of syllables (Kim & Cole 2006). The
five groups under study could differ in the amount of shortening/compression (if any), explaining
the differences in nPVI as well.
Apart from the lowest levels (syllable and foot), other prosodic domains like the intonational
phrase and the phonological phrase/word can be affected in language contact situations as well as
demonstrated in chapter 3. However, it is necessary to make a distinction. It has been argued that
the foot may not be a rhythmic unit in Spanish (and L2 English) and consequently its domain or
its relevance can be put into question if favor of the syllable. However, the domain of the
intonational phrase (and other intermediate domains) holds in both Spanish and English. Thus,
we can talk about two different ways in which prosodic domains can be affected. The first one
would be a difference in the relevance of certain domains (in this case the foot in L1 Spanish/L2
116
English for rhythm – Figure 59) while the second one would be a difference of phonological
patterns within the same domain. The use of a given pitch contour in Spanish and English would
be determined by its availability within the intonational phrase in the language. This would also
explain why the Adult Early Bilinguals still retains some Spanish contours in English and why
the L.A. Born Bilinguals speakers use configurations belonging to Spanish and English in either
language. While speakers are capable of switching between domains in their prosodic grammar
(e.g. Adult Early Bilinguals from syllable to foot due to extended contact with English / L.A.
Born Bilinguals syllable and foot depending on the language), they would encounter more
difficulties when distinguishing between alternative inventories within the same domain. It is
difficult to determine why this might be, but there is evidence that both Mexican Americans and
speakers of Spanish with English as their L2 (or speakers of English with Spanish as their L2)
also have difficulties to create two different phonological inventories at the segmental level
(Weinreich 1953, MacDonald 1989, Zampini 1998, Quilis & Fernández 1996, Lipski 2008,
Escudero 2006, inter alia). A possible explanation based on Kroll and Hermans (2011), is that
bilinguals do have both grammars available simultaneously and are failing to suppress tonal or
segmental choices from the language that is not being used. Although the two systems are active,
bilinguals suppress the grammar of the language they are not producing in order to facilitate
communication. It can be the case that bilinguals in the current study fail to suppress some of the
tones of the other language given the multiple competing tonal movements that are available in
both Mexican Spanish and English.
Another point that needs consideration is whether results arise from differences in proficiency,
exposure or identity. At the time of the formation of the groups the time of exposure,
proficiency, amounts of use of each language and the confidence each speaker has speaking each
language were considered. Based on the questionnaires speakers had to fill out before being
recorded, the amount of use, their confidence speaking the language and the amount of exposure
were normally correlated (i.e. if a participant had been exposed to English for a long time both
his/her confidence and the amount of English use were high as opposed to somebody who had
not been in contact with it). The few speakers that differed in these from the rest of the group
were not considered for the study to avoid having groups that were not homogeneous. It can also
be argued that cases where Spanish prosody is used instead of the English one reflect the desire
117
of the speakers to somehow preserve their identity and not necessarily a lack of proficiency.
Although this is definitely a possibility and may be the case with a few participants, various
studies have shown that Spanish per se is not a requirement to have Latino or Hispanic identity
(Rivera-Mills 2000, Potowsky et al. 2008, Phinney et al. 2010, inter alia). Additionally,
speakers’ attitudes towards Spanish and English in the current study were also revealing of the
strength the L1 prosody has. For example, some heritage speakers pointed out that they were
aware of their “latino accent” and that they try to avoid using it in English as much as possible in
order to sound “correct” or “proper”. However, current data demonstrates that the transfer of the
L1 intonation to the L2 is a fact even when speakers are trying to avoid it. But why are there
differences in the performance of the groups in the current study at the prosodic level? Potowski
(2012) highlights that according to the 2000 census, 75% of Hispanics speak Spanish at home
but that there is a notorious shift towards English monolingualism. This behavior has been
noticed in California in Rivera-Mills (2001) and Hurtado and Vega (2004) and it is compatible
with the results found in current study. Montrul (2008) also points out that the earlier the
acquisition of the second language by heritage speakers and concomitant disuse of the L1, the
more severe the loss of the L1 is. Based on the information obtained in the language
questionnaires, it is likely that younger speakers (The L.A. Born Bilinguals) still maintain the
distinction between languages thanks to the Spanish input they receive at home and the English
input they receive at school. In contrast, adult heritage speakers (The Adult Early Bilinguals)
may have lost the Spanish input from their families and have been more in contact with English,
resulting in a gradual loss of Spanish prosody (Potowski 2012). This line of thought is
corroborated by relationship between the length of exposure and the age on arrival and the
rhythms of different groups. In the case of the English Control Group, the Adult Late Bilinguals
and the Spanish Control Group, their lack of acquisition of the L2 (Spanish for the English
Control Group and English for the last two groups) would be linked to a lack of necessity to
communicate, learn and be exposed to the target language.
The kind of research pursued in this study by combining different linguistics subfields also has
its limitations as has been pointed out in previous chapters. First, although participants were
asked to fill out detailed questionnaires it is very difficult to form groups that are 100%
homogenous. Similarly, sociolinguistic variables like the exact Mexican variety speakers use,
118
their gender or the inclusion of more varied socioeconomic statuses should be considered in
future research. Although using laboratory materials provides us with comparable data and our
results are compatible with those found in other studies, it would also be desirable to compare
them with more spontaneous productions.
Recordings were made in the same formal and experimental setting for all the participants and
differences were still found at the group level. It is expected that with more spontaneous data,
differences would be even clearer. The same applies to the socioeconomic status of the
participants. In order to limit the influence of other sociolinguistic variables like class, most
speakers belonged to what can be considered the mid-class. In this class, more formal and
standard productions are expected than from speakers in a lower class due to the necessity to
communicate in “proper English” on daily basis. Another reason why data from speakers from a
lower class was not included was because they either did not know how to read properly or
because they felt uncomfortable being recorded due to what they considered a bad Spanish or a
bad English. Nevertheless, available productions still show enough variation to demonstrate the
prosodic variation within the L.A. community. Future studies on other Spanish varieties in
contact and longitudinal research should complement current findings to get a better
understanding of the transfer and the potential simplification of grammatical system in
bilinguals. Finally, although impressionistically differences between English and Mexican
Spanish prosodies are clearly noticeable, future research should also address how salient
prosodic modifications are to L1 and L2 speakers through a perception study.
The main goal of this work was to study the rhythmic and intonational properties of Mexican
Spanish and English in Los Angeles to determine if both components remain the same or present
deviations from the original grammars due to language contact. Very recent works have shown
that this is the case for other Spanish varieties in contact. Miglio et al. (2013) study the
intonation of Mexican and Chicano Spanish and conclude that the prosody of the latter shows
more plasticity and uses more pitch movements than the former due to English prosody. Kireva
(2013) looks at the prosody of Olivenza Spanish, a variety of Peninsular Spanish that has been in
extensive contact with Portuguese and concludes that its prosodic properties show a clear
convergence of both grammars. Their findings corroborate the idea that a competing prosody can
119
have an effect on the Spanish one. The current study goes one step further and considers the
prosodic properties of rhythm and intonation simultaneously in bilinguals taking into account
different social groups. Thanks to this sociolinguistic component it has been demonstrated that
the Mexican Spanish speaking community of Los Angeles is not homogeneous and that rhythm
and intonation can be affected in different fashions in language contact situations. The study also
shows how these building blocks of prosody can undergo more or less salient transfer and
attrition processes depending on social variables like the age or the exposure to both languages.
Thus, as a whole this work contributes to a better understanding of the phonological and prosodic
systems of Spanish and English with a special emphasis on the prosodic level. It also highlights
the importance of social variables in the study of phonological systems in general and of prosody
in particular. Finally it sheds light on the interaction of multiple competing prosodies in
bilinguals, opening an innovative line of research in second language acquisition and potential
applications to second language teaching.
120
APPENDIX A
*Age at the time of the first visit **Total time counting all visits
Group (Group 1 = English Control Group; Group 2 = Adult Early Bilinguals; Group 3 = L.A. Born Bilinguals;
Group 4 = Adult Late Bilinguals; Group 5 = Spanish Control Group), Subject (g=group; s=subject number), Gender
(M=male; F=female), Age at recording (age when participants were recorded), Age on arrival (age when speakers
arrived in L.A.), Parents (Group of the parents), %Sp & %En (percentages showing the use of Spanish and English
at different ages), NPVI (normalized pairwise variability index scores), voicing ratio (voicing ratio results).
Group Subject Gender A g e
at
r eco r di ng
A g e on
a rriv al
Tim e
i n L . A .
(y ears )
P a re nt s
(m ot he r/
f at he r)
% Sp
vs.
% En
1-14
years
% Sp
vs.
% En
15-25
years
% Sp
vs.
% En
26-50
years
nP VI
EN
nP VI
S P
Vo ici ng
r at io
EN
Vo ici ng
r at io
S P
Group1 g1s1 F 45 0 45 G1-G1 0-100 0-100 1-99 63 46 0.45 0.44
Group1 g1s2 M 22 0 22 G1-G1 5-95 5-95 NA 54 46 0.42 0.37
Group1 g1s3 F 29 0 29 G1-G1 0-100 2-98 25-75 53 38 0.49 0.42
Group1 g1s4 M 23 0 23 G1-G1 2-98 2-98 NA 56 NA 0.41 NA
Group1 g1s5 F 50 0 50 G1-G1 0-100 30-70 66-36 52 41 0.43 0.43
Group1 g1s6 M 20 0 20 G1-G1 0-100 0-100 NA 56 NA 0.38 NA
Group1 g1s7 M 20 0 20 G1-G1 2-98 5-95 NA 52 51 0.55 0.38
Group1 g1s8 M 29 0 29 G1-G1 0-100 0-100 0-100 59 NA 0.48 NA
Group1 g1s9 M 30 0 30 G1-G1 0-100 0-100 0-100 49 NA 0.42 NA
Group1 g1s10 F 36 0 36 G1-G1 0-100 10-90 10-90 49 41 0.45 0.33
Group1 g1s11 F 21 0 21 G1-G1 2-98 5-95 NA 46 42 0.42 0.36
Group1 g1s12 F 19 0 19 G1-G1 1-99 4-96 NA 50 48 0.48 0.43
Group2 g2s1 F 50 1 49 G4-G4 25-75 27-73 65-35 54 38 0.32 0.3
Group2 g2s2 F 48 4 44 G4-G4 35-65 17-83 20-80 53 42 0.45 0.45
Group2 g2s3 F 48 4 44 G4-G4 11-89 10-90 13-87 58 49 0.43 0.41
Group2 g2s4 M 24 0 24 G4-G4 40-60 17-83 NA 52 38 0.37 0.43
Group2 g2s5 M 32 7 25 G4-G4 66-34 30-70 35-75 52 41 0.4 0.41
Group2 g2s6 M 23 5 18 G4-G2 75-25 66-33 NA 53 42 0.47 0.49
Group2 g2s7 F 28 5 23 G4-G4 50-50 30-70 30-70 61 40 0.39 0.4
Group2 g2s8 M 22 2 20 G4-G4 60-40 35-65 NA 53 42 0.38 0.42
Group3 g3s1 F 18 0 18 G4-G4 50-50 15-85 NA 60 35 0.48 0.49
Group3 g3s2 M 20 0 20 G4-G4 20-80 15-85 NA 51 36 0.41 0.36
Group3 g3s3 M 15 0 15 G4-G1 12-88 12-88 NA 51 34 0.43 0.37
Group3 g3s4 F 20 0 20 G4-G2 1-99 1-99 NA 56 39 0.56 0.42
Group3 g3s5 M 18 0 18 G4-G2 26-74 26-74 NA 56 34 0.41 0.33
Group3 g3s6 F 14 0 14 G4-G4 25-75 NA NA 46 33 0.36 0.29
Group3 g3s7 M 12 0 12 G4-G4 30-70 NA NA 40 29 0.28 0.23
Group3 g3s8 F 9 0 9 G1-G4 4-96 NA NA 53 36 - -
Group3 g3s9 F 19 0 19 G1-G4 30-70 15-85 NA 49 32 0.39 0.30
Group3 g3s10 F 19 0 19 G4-G4 33-66 40-60 NA 47 32 0.39 0.33
Group3 g3s11 F 19 0 19 G4-NA 40-60 25-75 NA 60 34 0.41 0.36
Group4 g4s1 F 47 21 26 G5-G5 100-0 93-7 48-52 35 36 0.29 0.33
Group4 g4s2 F 37 17 20 G5-G5 100-0 65-35 85-15 45 33 0.36 0.41
Group4 g4s3 M 39 22 17 G5-G5 99-1 80-20 70-30 45 28 0.32 0.31
Group4 g4s4 F 48 25 23 G5-G5 100-0 90-10 50-50 36 36 0.23 0.33
Group4 g4s5 M 49 26 23 G5-G5 87-13 65-35 50-50 43 30 - -
Group4 g4s6 M 51 21 30 G5-G5 100-0 85-15 75-25 38 36 - -
Group4 g4s7 F 35 20 15 G5-G5 70-30 55-45 55-45 36 27 0.37 0.36
Group5 g5s1 M 24 24 0.2 G5-G5 100-0 99-1 NA 40 33 0.19 0.19
Group5 g5s2 M 29 27 2 G5-G5 93-7 75-25 32-68 44 34 0.34 0.3
Group5 g5s3 M 28 NA 0 G5-G5 87-13 85-15 75-25 34 33 0.3 0.29
Group5 g5s4 M 49 NA 0 G5-G5 100-0 97-3 100-0 NA 27 NA 0.25
Group5 g5s5 F 47 NA 0 G5-G5 100-0 95-5 100-0 NA 33 NA 0.18
Group5 g5s6 M 38 NA 0 G5-G5 99-1 99-1 99-1 40 28 0.32 0.3
Group5 g5s7 F 68 60* 2** G5-G5 100-0 100-0 100-0 32 31 0.27 0.24
Group5 g5s8 F 35 28* 0.9** G5-G5 100-0 100-0 100-0 34 33 0.18 0.14
Group5 g5s9 M 26 25 1 G5-G5 90-10 80-20 NA 43 31 0.26 0.38
Group5 g5s10 F 27 25 2 G5-G5 90-10 90-10 25-75 52 35 0.32 0.25
Group5 g5s11 F 22 20 2 G5-G5 94-6 54-46 NA 53 34 0.33 0.37
121
APPENDIX B
El viento norte y el sol
El viento norte y el sol porfiaban sobre cuál de ellos era el más fuerte, cuando acertó a pasar un
viajero envuelto en ancha capa. Convinieron en que quien antes lograra obligar al viajero a
quitarse la capa sería considerado más poderoso. El viento norte sopló con gran furia, pero
cuanto más soplaba, más se arrebujaba en su capa el viajero; por fin el viento norte abandonó
la empresa. Entonces brilló el sol con ardor, e inmediatamente se despojó de su capa el viajero;
por lo que el viento norte hubo de reconocer la superioridad del sol.
The North Wind and the Sun
The North Wind and the Sun were disputing which was the stronger, when a traveler came along
wrapped in a warm cloak. They agreed that the one who first succeeded in making the traveler
take his cloak off should be considered stronger than the other. Then the North Wind blew as
hard as he could, but the more he blew the more closely did the traveler fold his cloak around
him; and at last the North Wind gave up the attempt. Then the Sun shined out warmly, and
immediately the traveler took off his cloak. And so the North Wind was obliged to confess that
the Sun was the stronger of the two.
122
APPENDIX C
ENGLISH
DECLARATIVES
What is Amanda doing?
Amanda is preparing lemonade.
What is Amanda doing?
Amanda is buying bananas.
What is the good boy doing?
The good boy is eating ice-cream
What is the woman in blue doing?
The woman in blue is selling bananas.
What do you do?
I buy a banana
Your friend thinks that Mary is buying bananas. Tell your friend that Andrea (not Mary)
is the one buying bananas:
Andrea is buying bananas
YES/NO QUESTIONS
You go to the supermarket and ask if they have lemonade.
Do you have lemonade?
It is very cold outside and your friend comes into your apartment wearing a coat and a
hat.
Are you cold?
You are trying to sleep but TV is really loud. You ask your friend:
Could you lower the volume?
Ask your friends if they want candy:
Do you want some candy?
123
WH- QUESTIONS
Ask what time it is to somebody on the street.
What time is it?
You need to know where they sell marmalade. Ask your friend.
Where do they sell marmalade?
VOCATIVES:
You go to your friend’s apartment but you don’t see her. Call her:
¡Amanda!
After 10 seconds she does not show up. Call her again:
¡¡Amanda!!
IMPERATIVES
You want your friend to buy a banana:
Buy a banana
He ignores you:
Buy a banana!
He is still ignoring you. Insist:
¡¡Buy a banana!!
SPANISH
DECLARATIVES
¿Qué hace Amanda?
Prepara limonada
¿Qué hace Amanda?
Amanda compra bananas
¿Qué hace el niño bueno?
El niño bueno come helado.
¿Qué hace la señora morena?
La señora morena vende mandarinas.
¿Qué hace Pedro?
124
Abre el armario
¿Qué hace Pedro?
Anima a Amanda
Tu amigo piensa que María compra bananas. Di a tu amigo que es Andrea (y no María) la
que compra bananas:
Andrea compra bananas
YES/NO QUESTIONS
Entras en una tienda y le preguntas a un empleado si tiene mermelada.
¿Tiene mermelada?
Sabes que afuera hace mucho frío. Entra alguien bien abrigado y le preguntas si tiene frío
¿Tienes frío?
Tus nietos hacen mucho ruido y no te dejan oír las noticias (en la televisión/radio). Les
pides que se callen.
¿Se pueden callar?
Pregunta a tus sobrinos si quieren caramelos.
¿Quieren caramelos?
WH-QUESTIONS
Pregunta qué hora es.
¿Qué hora es?
Quieres saber dónde venden mermelada.
¿Dónde venden mermelada?
VOCATIVES
Entras en la casa de una amiga tuya, Marina, pero al entrar no la ves. Llámala.
¡Marina!
Pasan diez segundos y no sale nadie. Vuelve a llamarla.
¡¡Marina!!
125
IMPERATIVES
Quieres que tu amigo abra el armario:
Abre el armario
Tu amigo no te hace caso pero sigues queriendo que abra el armario:
¡Abre el armario!
Es la tercera vez que se lo repites pero tu amigo sigue ignorándote:
¡¡Abre el armario!!
126
APPENDIX D
43
Table D1. Number of tones in pre-nuclear pitch-accents in declarative sentences.
Table D2. Types of pre-nuclear pitch-accents in broad focus declaratives in L.A. Spanish and
English.
SP
pre-nuclear
H* L*+H L*+H
int.
L+>H* H% L*+H EN
pre-nuclear
H* L*+H L*+H
int.
L+>H* H% L*+H
g1 3/14
21.4%
0 2/14
14.2%
9/14
64.2%
0 g1 4/24
16.7%
0 2/24
8.3%
17/24
70.8%
1/24
4.1%
g2 4/30
13.3%
0 4/30
13.3%
19/30
63.3%
3/30
10%
g2 3/23
13%
0 2/23
8.7%
13/23
56.5%
5/23
21.7%
g3 1/26
3.8%
4/26
15.3%
5/26
19.2%
15/26
57.7%
1/26
3.8%
g3 2/21
9.5%
2/21
9.5%
3/21
14.2%
10/21
47.6%
4/21
19%
g4 0 3/15
20%
5/15
33.3%
3/15
20%
4/15
26.6%
g4 0 1/12
8.3%
2/12
16.7%
3/12
25%
6/12
50%
g5 0 2/15
13.3%
4/15
26.6%
9/15
60%
0 g5 0 1/12
8.3%
2/12
16.7%
6/12
50%
3/12
25%
pre-nuclear
pitch-
accent
broad focus
Amanda compra
bananas
anima a
amanda
el niño
bueno
come
helado
la señora
morena vende
mandarinas
Amanda is
buying bananas
Amanda is
preparing
lemonade
the
woman in
blue is
selling
bananas
g1s2 - - - - L+>H* L+>H* L+>H*
g1s3 H* - H* H* H* H* L+>H*
g1s4 - - - - L+>H* L+>H* L+>H*
g1s5 L+>H* L+>H* L+>H* L+>H* L+>H* L+>H* L+>H*
g1s6 - - - - H* H* L+>H*
g1s9 - - - - L*+H (int) L+>H* L+>H*
g1s11 L*+H (int) L+>H* L+>H* L*+H (int) L*+H (int) H% L*+H L+>H*
g1s12 L+>H* L+>H* L+>H* - L+>H* L+>H* L+>H*
g2s1 L+>H* L+>H* L+>H* - - L+>H* L+>H*
g2s2 H% L*+H - L+>H* H% L*+H H% L*+H H% L*+H H% L*+H
g2s3 H% L*+H L+>H* L+>H* L*+H (int) H% L*+H H% L*+H L+>H*
g2s4 L+>H* L+>H* L+>H* L+>H* L+>H* L+>H* L+>H*
g2s5 L+>H* L+>H* L+>H* L+>H* L+>H* L+>H* L+>H*
g2s6 L+>H* L+>H* L+>H* L+>H* L+>H* L+>H* L+>H*
g2s7 H* H* H* H* H* H* H*
g2s8 L*+H (int) L*+H (int) L+>H* L*+H (int) L*+H (int) L*+H (int) L+>H*
g3s1 L+>H* L+>H* L+>H* L+>H* L+>H* L+>H* L+>H*
g3s2 L+>H* H* L+>H* L*+H (int) H% L*+H H% L*+H L*+H
(int)
g3s3 L+>H* L+>H* L+>H* L*+H (int) L+>H* L+>H* L+>H*
g3s5 L*+H (int) L+>H* L+>H* - H* L+>H* L+>H*
g3s9 L+>H* L*+H (int) L+>H* L*+H L*+H H% L*+H H% L*+H
g3s11 H% L*+H L*+H L*+H (int) L+>H* H* L*+H (int) L+>H*
g3s10 L*+H L+>H* L*+H - L*+H (int) L*+H L+>H*
g4s1 L*+H (int) L*+H (int) L+>H* L*+H (int) L*+H (int) L*+H (int) L*+H
g4s2 H% L*+H H% L*+H L*+H - H% L*+H H% L*+H H% L*+H
g4s3 H% L*+H H% L*+H L+>H* L*+H (int) H% L*+H L+>H* L+>H*
g4s7 L*+H (int) L+>H* L*+H L*+H H% L*+H H% L*+H L+>H*
g5s2 L+>H* L+>H* L*+H (int) L+>H* H% L*+H L*+H (int) L+>H*
g5s9 L+>H* L*+H (int) L+>H* L*+H (int) L+>H* L+>H* L+>H*
g5s10 - L*+H (int) L*+H L*+H H% L*+H L*+H H% L*+H
g5s11 L+>H* L+>H* L+>H* L+>H* L*+H (int) L+>H* L+>H*
127
SP
nuclear
L*L% H*L%
L+H*L%
L+¡H* L% !H% EN
nuclear
L*L% H*L%
L+H*L%
L+¡H* L% !H%
g1 9/22
40.9%
12/22
54.5%
1/22
4.5%
0/22
0/22 g1 2/22
9%
19/22
86.3%
1/22
4.5%
0/22 0/22
g2 5/40
12.5%
24/40
60%
5/40
12.5%
1/40
2.5%
5/40
12.5%
g2 6/20
30%
11/20
55%
0/20 1/20
5%
2/20
10%
g3 18/34
52.9%
8/34
23.5%
5/34
14.7%
3/34
8.8%
0/34 g3 7/17
41.1%
5/17
29.4%
1/17
5.8%
3/17
17.6%
1/17
5.8%
g4 19/23
82.6%
2/23
8.6%
0/23 2/23
8.6%
0/23 g4 7/12
58.3%
0/12 0/12 3/12
25%
2/12
16.6%
g5 11/24
45.8%
4/24
16.6%
0/24 3/24
12.5%
6/24
25%
g5 3/12
25%
3/12
25%
1/12
8.3%
2/12 (1
with LH%)
16.6%
3/12
25%
Table D3. Number of tones in nuclear pitch-accents in declarative sentences per group.
Nuclear
accents
and
semitones
44
abre el
armario
Amanda
compra
bananas
anima a
Amanda
el niño
bueno
come
helado
la señora
morena
vende
mandarinas
prepara
limonada
the woman
in blue is
selling
bananas
Amanda is
buying
bananas
I buy a
banana
g1s2 NA NA NA NA NA NA H*L%
0.3
L*L%
-1.5
H*L%
0.3
g1s3 H*L%
0.3
H*L%
0.9
- L*L%
-0.2
H*L%
1.4
L*L%
0.3
L*L%
-0.7
L+H*L%
2.7
-
g1s4 NA NA NA NA NA NA H*L%
0.7
H*L%
0.2
H*L%
0.3
g1s5 H*L%
0.6
L*L%
-1.8
H*L%
0.4
L*L%
-1.6
H*L%
1.1
L*L%
-2.2
H*L%
0.4
H*L%
0.1
H*L%
1
g1s6 NA NA NA NA NA NA H*L%
0.8
H*L%
1.1
H*L%
0.3
g1s9 NA NA NA NA NA NA H*L%
1.4
H*L%
1
H*L%
0.2
g1s11 L*L%
-0.8
H*L%
0.2
H*L%
1.4
- H*L%
0.1
H*L%
0.2
H*L%
0.2
H*L%
0.2
H*L%
0.2
g1s12 H*L%
0.3
L*L%
-1.6
L+H*
1.4
L*L%
-0.8
L*L%
-0.9
H*L%
0.4
H*L%
1.3
H*L%
0.2
-
g2s1 L+H*L
%
1.6
H*L%
0.6
H*M%
-0.2
H*L%
0.1
L+H*L%
1.9
H*L%
0.2
H*L%
0.9
H*L%
0.6
H*L%
0.3
g2s2 - - - L*L%
-3.0
- L*L%
-1.7
L*L%
-1
- L*L%
-4.3
g2s3 H*L%
0.3
H*!H%
0.6
H*L%
0.7
H*!H%
0.7
H*!H%
0.4
H*!H%
0.1
H*L%
0.3
H*!H%
0.1
H*!H%
0.2
g2s4 L+H*L
%
1.6
H*L%
0.2
L+¡H*
L% early
4.3
H*L%
0.2
H* L%
1.4
H* L%
0.9
H* L%
0.7
H* L%
0.2
H*L%
0.2
g2s5 L+H*L
%
1.5
H*L%
0.1
H*L%
0.1
L*L%
-0.2
H*L%
0.1
H*L%
0.1
L*L%
-0.5
H*L%
0.6
L*L%
-0.8
g2s6 L*L%
-1.6
H*L%
0.1
H*L%
0.7
H*L%
0.1
H*L%
0.3
H*L%
0.3
L*L%
-1.9
H* L%
0.2
L*L%
-0.3
g2s7 - - - H*L%
0.1
H*L%
0.8
- - L+¡H*
L%
3.6
-
g2s8 H*L%
0.3
H*L%
0.5
L+H*L%
2.3
L*L%
0.2
H*L%
0.2
H*L%
0.3
- H*L%
0.7
H*L%
0.2
g3s1 L*L%
-4
L*L%
-2.8
L*L%
-4.4
L*L%
-2.8
L*L%
-1.4
L*L%
-4
- L+¡H*L%
3.5
L*L%
-1.6
g3s2 H*L%
0.1
L*L%
-2.6
H*L%
0.1
L*L%
-0.5
L*L%
-1
L*L%
-2.8
L*L%
-6
H*L%
1.4
H*!H%
0.1
128
Table D4. Number of nuclear pitch-accents in declarative sentences per group.
SP
st
fall downstep maint upstep L+H* H% EN
st
fall downstep maint upstep L+H* H%
g1 1/4
25%
3/4
75%
0 0 0 0 g1 0 6/6
100%
0 0 0 0
g2 1/6
16.7%
4/6
66.7%
1/6
16.7%
0 0 0 g2 3/7
43%
4/7
57% (1 with !H%)
0 0 0 0
g3 2/5
40%
3/5
60%
0 0 0 0 g3 3/6
50%
2/6
33.3%(1 with !H%)
1/6
16.7%
0 0 0
g4 4/4
100%
0 0 0 0 0 g4 1/4
25%
2/4
50% (2 with!H%)
0 1/4
25%
0 0
g5 2/4
50%
2/4
50%(1 with
!H%)
0 0 0 0 g5 0 2/4
50%(1 with !H%)
0 2/4 (1
with
LH%)
50%
0 0
Table D5. Number of intonational contours in declarative context per group.
g3s3 L+H*L
%
2.3
L+H*L%
2
L+H*L%
2
H* L%
0.8
L+H*L%
2
L+H*L%
2
L*L%
-1.3
H*L%
0.3
L*L%
-0.8
g3s5 - - - H*L%
0.1
- L*L%
-0.7
- H*L%
0.1
-
g3s9 H*L%
-0.2
L*L%
-0.6
- L*L%
-0.3
H*L%
0.1
H*L
0.2
H*L%
0.1
L+¡H* L%
4.9
H*L%
0.5
g3s10 L*L%
-2
L*L%
-2.8
L*L%
-0.5
L*L%
-0.8
- L*L%
-2
L*L%
-2.1
L*L%
-3
L*L%
-1.1
g3s11 - L+¡H*
L%
3
L+¡H*
L%
7.1
H*L%
0.1
L+¡H* L%
3.1
- - L+¡H* L%
3.9
L+H*
L%
1.5
g4s1 L*L%
-2.3
L*L%
-1.5
L*L%
-2.2
L*L%
-1.6
L*L%
-1.3
- L*L%
-2.8
L*L%
-1.8
L*L%
-0.8
g4s2 L*L%
-3
L+¡H*
L%
5.1
L+¡H*
L%
early
7.3
L*L%
-1.6
L*L%
-1.5
L*L%
-2.3
L+¡H*L%
2.7
L+¡H* L%
early
5
L+H*!
H%
2.1
g4s3 L*L%
-0.9
L*L%
-1.1
L*L%
-1.5
L*L%
-1.9
L*L%
-1.6
L*L%
-0.8
L*L%
-0.8
L*L%
-0.9
H*!H%
0.1
g4s7 L*L%
-1.1
L*L%
-2.4
H*L%
0.1
L*L%
-5.7
H*L%
0.1
L*L%
-1.9
L*L%
-3.5
L*L%
-3.1
L+¡H*
L%
13.7
g5s2 H*!H%
0.9
L*L%
-1.3
H*L%
0.4
H*!H%
0.8
H*!H%
0.1
L*L%
-0.4
H*L%
0.1
H*L%
1.2
H*L%
0.1
g5s9 H*!H%
0.9
H* L%
0.9
L*L%
-0.8
H*!H%
0.9
H*!H%
0.2
H*L%
1.2
H*!H%
0.1
H*!H%
0.2
H*!H%
0.2
g5s10 L*L%
-2.7
L*L%
-1.4
L*L%
-0.8
L*L%
-4.7
L*L%
-1.2
L*L%
-3.5
L*L%
-0.8
L*L%
-1.3
L+¡H*
LH%
5.6
g5s11 L*L%
-0.7
L+¡H*L%
3.6
L*L%
-3.1
L+¡H*
L%
early
3.8
L+¡H*
L%
3.8
H*L%
0.2
L*L%
-7.2
L+H* L%
2.1
L+¡H*
L%
3.1
129
SP
imp1
fall downstep maint up L+H* H% EN
imp1
fall downstep maint up L+H* H%
g1 2/4
50%
1/4
25%
0 0 1/4
25%
g1 5/8
62.5%
1/8
12.5%
0 1/8
12.5%
1/8
12.5%
0
g2 5/6 83.3% 0 0 0 1/6
16.6
%
g2 2/8
25%
2/8
25%
2/8
25%
2/8
25%
0 0
g3 2/7
28.6
%
3/7 42.8% 1/7
14.3%
0 0 1/7
14.3
%
g3 6/7
85.7%
0 1/7
14.3%
0 0
g4 3/4
75%
0 0 0 0 1/4
25%
g4 3/4
75%
(1 with
L*+H)
0 0 1/4
25%
0 0
g5 2/4
50%
1/4
25%
0 0 0 1/4
25%
g5 2/4
50%
0 0 1/4
25%
0 1/4
25%
Table D6. Number of intonational contours in imp1 context per group.
SP
imp2
fall downstep maint up L+H* EN
imp2
fall downstep maint up L+H*
g1 0 2/4
50%
1/4
25%
1/4
25%
0 g1 0 2/8
25%
0 5/8
62.5%
1/8
12.5%
g2 0 4/6
66.7%
1/6
16.6%
1/6
16.6%
0 g2 2/8
25%
2/8
25%
1/8
12.5%
3/8
37.5%
0
g3 0 4/7
57.1%
1/7
14.3%
2/7
28.6%
0 g3 2/7
28.6%
2/7
28.6%
1/7
14.3%
2/7
28.6%
0
g4 2/4
50%
0 1/4
25%
0 1/4
25%
g4 1/4
25%
1/4
25%
0 2/4
50%
0
g5 0 2/4
50% (1 with !H%)
2/4
50%
0 0 g5 0 0 2/4
50%
(with !H%)
2/4
50%
0
Table D7. Number of intonational contours in imp2 context per group.
SP
imp3
fall downstep maint up L+H* En
imp3
fall downstep maint up L+H*
g1 1/4
25%
3/4
75%
0 0 0 g1 0 2/8
25%
4/8
50%
1/8
12.5%
1/8
12.5%
(with up)
g2 0 4/6
66.7%
0 1/6
16.6%
0 g2 0 4/8
50%
2/8
25%
2/8
25%
0
g3 0 5/7
71.4%
(1 with
!H%)
0 1/7
14.3%
1/7
14.3%
g3 1/6
16.6%
3/6
50%
1/6
16.6%
1/6
16.6%
0
g4 1/4
25%
1/4
25%
1/4
25%
1/4
25%
0 g4 1/4
25%
1/4
25%
1/4
25%
1/4
25% (with
L*+H)
0
g5 0 2/4
50%
1/4
25%
1/4
25%
0 g5 0 0 1/4
25%
3/4
75%
(1 with
initial H%
and two with
final !H%)
0
Table D8. Number of intonational contours in imp3 context per group.
130
tonemes Wh-questions
Sp
L*L% L*H% L+H* L% tonemes Wh-questions
EN
L*L% L*H% L+H*
L%
g1 3/8
37.5%
4/8
50%
1/8
12.5%
g1 6/16
37.5%
10/16
62.5%
g2 4/16
25%
8/16
50%
4/16
25%
g2 2/16
12.5%
10/16
62.5%
4/16
25%
g3 3/13
23%
6/13
46.1%
4/13
30.7%
g3 5/14
35.7%
7/14
50%
2/14
14.3%
g4 4/8
50%
3/8
37.5%
1/8 (L+¡H*
L%)
12.5%
g4 1/8
12.5%
6/8
75%
1/8
12.5%
g5 3/8
37.5%
5/8
62.5%
g5 3/8
37.5%
3/8
37.5%
2/8
25%
Table D9. Number of tonemes in Wh- questions per group.
¿dónde venden
mermelada?
¿qué hora es? where do they sell
marmalade?
what time is it?
g1s2 NA NA L+>H* L+H* L% L+>H* L+H* L%
g1s3 L+>H* L* L% L+>H* L* H% L+>H* L+H* L% L+>H* L+H* L%
g1s4 NA NA L+>H* L* H% L+>H* L+H* L%
g1s5 L+>H* L* H% L+>H* L* H% L+>H* L* H% L+>H* L* H%
g1s6 NA NA L+>H* L+H* L% L+>H* L* H%
g1s9 NA NA L+>H* L* H% L+>H* L+H* L%
g1s11 L+>H* L+H* L% L+>H* L* L% L+>H* L+H* L% L+>H* L+H* L%
g1s12 L+>H* L* H% H* L* L% L+>H* L+H* L% L+>H* L* H%
g2s1 L+>H* L* H% L+>H* L* H% L+>H* L+H* L% L+>H* L* H%
g2s2 L+>H* L* L% H* L* H% L+>H* L+H* L% L+>H* L* H%
g2s3 L+>H* L* H% H* L* H% L+>H* L* H% L+>H* L* H%
g2s4 L+>H* L+H* L% H* L* H% L+>H* L* H% L+>H* L* L%
g2s5 L+>H* L+H* L% L+>H* L* H% L+>H* L* H% L+>H* L* H%
g2s6 L+>H* L+H* L% L+>H* L+H* L% L+>H* L* L% L+>H* L* H%
g2s7 L+>H* L* H% L+>H* L* H% L+>H* L* H% L+>H* L* H%
g2s8 L+>H* L* L% L+>H* L* L% L+>H* L+H* L% L+>H* L+H* L%
g3s1 L+>H* L* L% L+>H* L* H% L+>H* L+H* L% L+>H* L* L%
g3s2 L+>H* L* L% H* L* H% L+>H* L* H% L+>H* L* H%
g3s3 L+>H* L+H* L% H* L+H* L% L+>H* L* H% L+>H* L* L%
g3s5 - H* L*L% L+>H* L* L% L+>H* L* L%
g3s9 L+>H* L* H% L+>H* L* H% L+>H* L* H% H* L* H%
g3s10 L+>H* L+H* L% L+>H* L* H% H* L+H* L% L+>H* L* L%
g3s11 L+>H* L+H* L% L+>H* L* H% L+>H* L* H% L+>H* L* H%
g4s1 L+>H* L* H% L+>H* L* H% L+>H* L* H% L+>H* L* H%
g4s2 L+>H* L+¡H* L% H* L* H% L+>H* L+H* L% H* L* H%
g4s3 L+>H* L* L% H* L* L% L+>H* L* H% L+>H* L* L%
g4s7 L+>H* L* L% H* L*L% L+>H* L* H% L+>H* L* H%
g5s2 L+>H* L+H* L% L+>H* L+H* L% L+>H* L* H% L+>H* L+H* L%
g5s9 L+>H* L+H* L% H* L* H% L+>H* L+H* L% L+>H* L* H%
g5s10 L+>H* L* H% L+>H* L* H% L+>H* L* L% L+>H* L* L%
g5s11 L+>H* L+H* L% L+>H* L+H* L% L+>H* L+H* L% L+>H* L* H%
Table D10. Types of tonemes in Wh-questions in L.A. Spanish and English.
131
tonemes Yes/No-
questions Sp
L+H*H% L*
H%
H+L*
L%
H*H% tonemes Yes/No
questions EN
L+H*H% L* H% H+L*
L%
H*H%
g1 3/16
18.8%
12/16
75%
1/16
6.2%
0 g1 12/32
37.5%
20/32
62.5%
0 0
g2 15/32
46.9%
16/32
50%
1/32
3.1%
0 g2 15/32
46.9%
14/32
43.8%
1/32
3.1%
2/32
6.2%
g3 13/28
46.4%
14/28
50%
1/28
3.6%
0 g3 6/28
21.4%
18/28
64.2%
1/28
35.7%
3/28
10.7%
g4 4/16
25%
10/16
62.5%
2/16
12.5%
0 g4 3/16
18.8%
12/16
75%
1/16
6.2%
0
g5 4/16
25%
11/16
68.8%
1/16
6.2%
0 g5 6/16
37.5%
10/16
62.5%
0 0
Table D11. Number of tonemes in Yes/no questions per group.
¿tienes
frío?
¿tienen
mermelada?
¿quieren
caramelos?
¿se pueden
callar?
are you
cold?
do you have
lemonade?
do you want
some candy
could you
lower the
volume?
g1s2 NA NA NA NA L+H* H% L*H% L+H* H% L*H%
g1s3 L*H% L*H% L+H* H% L*H% L+H* H% L*H% L*H% L*H%
g1s4 NA NA NA NA L+H* H% L*H% L+H* H% L*H%
g1s5 L*H% L+H* H% L*H% L*H% L+H* H% L*H% L*H% L*H%
g1s6 NA NA NA NA L+H* H% L*H% L+H* H% L*H%
g1s9 NA NA NA NA L+H* H% L*H% L+H* H% L*H%
g1s11 L+H* H% L*H% L*H% L*H% L+H* H% L*H% L*H% L*H%
g1s12 L*H% L*H% L*H% H+L*L% L+H* H% L*H% L*H% L*H%
g2s1 L+H* H% L*H% L+H* H% L*H% L+H* H% L*H% L+H* H% L*H%
g2s2 L+H* H% L*H% L+H* H% L+H* H% L+H* H% L*H% L*H% L*H%
g2s3 L*H% L*H% L*H% L*H% L+H* H% L*H% L*H% L*H%
g2s4 L+H* H% L+H* H% L+H* H% L+H* H% L+H* H% L+H* H% L+H* H% L+H* H%
g2s5 L+H* H% L+H* H% L+H* H% L+H* H% L+H* H% L*H% L+H* H% L+H* H%
g2s6 L*H% L*H% L*H% L+H* H% H*H% L+H* H% L*H% H+L*L%
g2s7 H+L*L% L*H% L*H% L+H* H% L+H* H% H*H% H+L*L% L*H%
g2s8 L*H% L*H% L*H% L*H% L+H* H% L*H% L+H* H% L*H%
g3s1 L*H% L*H% L*H% L*H% L+H* H% L*H% L*H% L*H%
g3s2 L+H* H% L+H* H% L+H* H% L+H* H% L+H* H% H+L*L% L*H% L+H* H%
g3s3 L*H% L+H* H% L+H* H% L+H* H% L+H* H% L*H% L*H% L*H%
g3s5 L+H* H% L+H* H% L+H* H% L*H% H*H% L*H% L+H* H% L*H%
g3s9 L*H% L*H% L*H% L*H% H*H% L*H% L+H* H% L*H%
g3s10 L*H% L*H% L*H% L*H% L*H% L*H% L*H% L*H%
g3s11 L+H* H% L+H* H% H+L*L% L+H* H% H*H% L*H% L*H% L*H%
g4s1 L*H% L*H% L*H% L*H% L+H* H% L*H% L*H% L*H%
g4s2 L+H* H% L+H* H% L+H* H% L*H% L*H% L*H% L*H% L*H%
g4s3 L*H% L*H% L*H% H+L*L% L+H* H% L*H% L*H% H+L*L%
g4s7 L*H% L+H* H% L*H% H+L*L% L+H* H% L*H% L*H% L*H%
g5s2 L*H% L+H* H% L*H% L*H% L+H* H% L+H* H% L*H% L*H%
g5s9 H+L*L% L+H* H% L+H* H% L+H* H% L+H* H% L*H% L*H% L+H* H%
g5s10 L*H% L*H% L*H% L*H% L+H* H% L*H% L*H% L*H%
g5s11 L*H% L*H% L*H% L*H% L+H* H% L*H% L*H% L*H%
Table D12. Types of tonemes in Yes/no questions in L.A. Spanish and English.
132
Tonemes -
Vocatives SP
L+H*H% L+H*
!H%
L+H*L% L*
HL%
Tonemes -
Vocatives EN
L+H*H% L+H*
!H%
L+H*L% L* HL%
g1 2/8
25%
6/8
75%
0 0 g1 0 11/16
68.75%
5/16
31.25%
0
g2 0 4/12
33.3%
2/12
16.6%
6/12
50%
g2 1/16
6.25%
9/16
56.25%
2/16
12.5%
4/16
25%
g3 0 0 9/14
64.3%
5/14
35.7%
g3 0 2/14
14.3%
10/14
71.4%
2/14
14.3%
g4 1/8
12.5%
5/8
62.5%
1/8
12.5%
1/8
12.5%
g4 1/8
12.5%
3/8
37.5%
2/8
25%
2/8
25%
g5 0 4/8
50%
0 4/8
50%
g5 1/8
12.5%
3/8
37.5%
1/8
12.5%
3/8
37.5%
Table D13. Number of tonemes in vocatives per group.
Table D14. Types of tonemes in vocatives in L.A. Spanish and English.
Marina 1 (Sp) Marina 2 (Sp) Amanda 1
(En)
Amanda 2
(En)
g1s2 NA NA L+H* L% L+H* L%
g1s3 L+H* !H% L+H* !H% L+H* !H% L+H* !H%
g1s4 NA NA L+H* !H% L+H* L%
g1s5 L+H* HH* L+H* HH* L+H* !H% L+H* !H%
g1s6 NA NA L+H* !H% L+H* !H%
g1s9 NA NA L+H* L% L+H* L%
g1s11 L+H* !H% L+H* !H% L+H* !H% L+H* !H%
g1s12 L+H* !H% L+H* !H% L+H* !H% L+H* !H%
g2s1 L+H* !H% L* HL% L+H* !H% L+H* !H%
g2s2 NA NA L+H* HH* L+H* L%
g2s3 L* HL% L* HL% L* HL% L* HL%
g2s4 L+H* L% L+H* !H% L+H* L% L+H* !H%
g2s5 L+H* !H% L* HL% L+H* !H% L+H* !H%
g2s6 L* HL% L* HL% L* HL% L* HL%
g2s7 NA NA L+H* !H% L+H* !H%
g2s8 L+H* L% L+H* !H% L+H* !H% L+H* !H%
g3s1 L+H* L% L* HL% L+H* L% L+H* L%
g3s2 L+H* L% L* HL% L+H* L% L+H* L%
g3s3 L+H* L% L* HL% L+H* L% L* HL%
g3s5 L+H* L% L+H* L% L+H* L% L+H* L%
g3s9 L+H* L% L+H* L% L+H* L% L+H* L%
g3s10 L* HL% L* HL% L* HL% L+H* L%
g3s11 L+H* L% L+H* L% L+H* !H% L+H* !H%
g4s1 L+H* L% L* HL% L+H* L% L* HL%
g4s2 L+H* !H% L+H* HH% L+H* HH% L+H* !H%
g4s3 L+H* !H% L+H* !H% L+H* !H% L+H* !H%
g4s7 L+H* !H% L+H* !H% L+H* L% L* HL%
g5s2 L* HL% L* HL% L+H* L% L* HL%
g5s9 L* HL% L* HL% L+H* HH% L+H* !H%
g5s10 L+H* !H% L+H* !H% L* HL% L* HL%
g5s11 L+H* !H% L+H* !H% L+H* !H% L+H* !H%
133
REFERENCES
Abercrombie, David. 1967. Elements of general phonetics. Chicago: Aldine.
Adams, Corinne. 1979. English speech rhythm and the foreign learner. The Hague, Mouton.
Aguilar, L. Machuca. 1995. Intentionality in the speech act and phonetic reduction phenomena.
Proceedings of XIII International Congress of Phonetic Sciences, Stockholm.
Alvord, Scott. 2006. Spanish intonation in contact: the case of Miami Cuban bilinguals.
University of Minnesota dissertation.
Atterer, Michaela and Robert Ladd. 2004. On the phonetics and phonology of ‘‘segmental
anchoring’’ of F0: evidence from German. Journal of Phonetics, 32(2). 177-197.
Ávila, Sylvia. 2003. La entonación del enunciado interrogativo en el español de la ciudad
de México. La tonía. Dimensiones fonéticas y fonológicas, ed. by Esther Herrera
Zendejas and Pedro Martín Butragueño, 331-355. México: El Colegio de México.
Beckman. Mary E. 1992. Evidence for speech rhythms across languages. Speech Perception,
Production and Linguistic Structure, ed. by Y. Tohura, E. Vatikiotis-Bateson and Y.
Sagisaka, 457-463. Tokyo: Omsha and Amsterdam: IOS Press.
Beckman, Mary E. and Janet Pierrehumbert. 1986. Intonational Structure in Japanese and
English. Phonology Yearbook 3. 255-309.
Beckman, Mary. E. and Julia Hirschberg. 1994. The ToBI Annotation Conventions. Online:
http://www.ling.ohio-state.edu/~tobi/ame_tobi/annotation_conventions.html.
Beckman, Mary. E., Manuel Díaz-Campos, Julia McGory and Terence Morgan. 2002. Intonation
Across Spanish, in the Tones and Break Indices Framework. Probus 14. 9-36.
134
Beckman, Mary E., Julia Hirschberg and Stefanie Shattuck Hufnagel. 2005. The original ToBI
system and the evolution of the ToBI framework. Prosodic typology: The phonology of
intonation and phrasing, ed. by Sun-Ah Jun, 9-54. New York: Oxford University Press.
Boersma, Paul and David Weenink. 2009. Praat: doing phonetics by computer (Version 5.1.17)
[Computer program]. Retrieved October 9, 2009, from http://www.praat.org/
Bolinger, Dwight. 1986. Intonation and its parts: melody in spoken English. Stanford, Stanford
University Press.
Borràs-Comes, Joan, Rafeu Sichel-Bazin and Pilar Prieto. 2013. Vocative intonation in Central
Catalan: Social & contextual factors. Paper presented at the Phonetics and Phonology in
Iberia, Lisbon.
Brookes, Mike. 2003. VOICEBOX - MATLAB toolbox for speech processing. Retrieved from
http://www.ee.ic.ac.uk/hp/staff/dmb/voicebox/voicebox.html
Brown, Penelope and Stephen Levinson. 1987. Politeness. Cambridge: Cambridge University
Press.
Carter, Phillip M. 2005. Quantifying rhythmic differences between Spanish, English, and
Hispanic English. Theoretical and experimental approaches to romance linguistics:
Selected papers from the 34th linguistic symposium on Romance languages (Current
issues in linguistic theory 272), ed. by Randall Gess, and Edward J. Rubin, 63-75.
Amsterdam, Philadelphia: John Benjamins.
Colantoni, Laura and Jorge Gurlekian. 2004. Convergence and intonation: historical evidence
from Buenos Aires Spanish. Bilingualism: Language and Cognition 7. 107-119.
135
de-la-Mota, Carme, Martín Butragueño, Pedro, and Prieto, Pilar. 2010. Mexican Spanish
Intonation. Transcription of Intonation of the Spanish Language, ed. by Pilar Prieto and
Paolo Roseano, 319-350. München: Lincom Europa.
Dasher, Richard and Dwight Bolinger. 1982. On pre-accentual lengthening. Journal of the
International Phonetic Association, 12. 58-69.
Dauer, Richard. 1983. Stress-timing and syllable-timing reanalyzed. Journal of Phonetics 11. 51-
62.
Dellwo, Volher, Adrian Fourcin and Evelyn Abberton. 2007. Rhythmical classification based on
voice parameters. International Conference of Phonetic Sciences (ICPhS).
Deterding, David. 2001. The measurement of rhythm: a comparison of Singapore and British
English. Journal of Phonetics 29. 217-230.
Elordieta, Gorka . 2003. The Spanish intonation of speakers of a Basque pitch-accent dialect.
Catalan Journal of Linguistics 2. 67-95.
Ericson, Holly A. 2007. An intonational analysis of Mexican American English in comparison to
Anglo American English. Master’s Thesis, North Carolina State Univeristy.
Escudero, Paola. 2006. The phonological and phonetic development of new vowel contrasts in
Spanish learners of English. English with a Latin beat: Studies in Portuguese/Spanish
English interphonology, ed. by B. O. Baptista, and M. A. Watkins, 149-161.
Amsterdam: John Benjamins.
Estebas-Vilaplana, Eva. 2007. The phonological status of English and Spanish prenuclear F0
peaks. Atlantis 29, 2. 39-57.
136
Estebas-Vilaplana, Eva and Pilar Prieto. 2010. Castilian Spanish Intonation. Transcription of
Intonation of the Spanish Language, ed. by Pilar Prieto and Paolo Roseano 17-48,
München: Lincom Europa.
Face, Timothy L. 2002. Intonational Marking of Contrastive Focus in Madrid Spanish. Munich:
Lincom Europa.
Fletcher, Janet, Esther Grabe and Paul Warren. 2005. Intonational variation in four dialects of
English: the high rising tune. Prosodic Typology: The Phonology of Intonation and
Phrasing, ed. by Sun-Ah Jun. 390-409. Oxford: Oxford University Press
Frota, Sonia and Marina Vigário. 2001. On the correlates of rhythmic distinctions: The
European/Brazilian Portuguese case. Probus 13. 247-275.
Gabriel, Christoph, Ingo Feldhausen, Andrea Pesková, Laura Colantoni, Su-Ar Lee, Valeria
Arana and Leopoldo Labastía. 2010. Argentinian Spanish intonation. Transcription of
intonation of the Spanish language, ed. by Pilar Prieto and Paolo Roseano, 285-317.
München: Lincom Europa.
Gao, Man. 2008. Gestural Coordination among Vowel, Consonant and Tone Gestures in
Mandarin Chinese. Invited presentation at the 8th Phonetic Conference of China and the
International Symposium on Phonetic Frontiers. Beijing, China.
Girand, Cynthia. 2006. Prenuclear Accentual Structure in Conversational English. University of
Colorado dissertation.
Goldstein, Louis, Dani Byrd, and Elliot Saltzman. 2006. The role of vocal tract gestural action
units in understanding the evolution of phonology. Action to Language via the Mirror
Neuron System, ed. by Michael Arbib, 215-249. Cambridge University Press.
137
Grabe, Esther and Ee Low. 2002. Durational Variability in Speech and the Rhythm Class
Hypothesis. Papers in Laboratory Phonology 7, ed. by. Carlos. Gussenhoven and
Natasha Warner, 377-401. Berlin: Mouton de Gruyter.
Grabe, Esther, Brechtje Post and Ian Watson. 1999 . The acquisition of rhythmic patterns in
English and French. Proceedings of the International Congress of Phonetic Sciences,
1201-1204. San Francisco.
Gussenhoven, Carlos. 2004. The phonology of tone and intonation. Cambridge, CUP.
Gut, Ulrike. 2003. Prosody in second language speech production: the role of the native
language. Fremdsprachen Lehren und Lernen 32. 133-152.
Harris, Michael J. and Stefan Th. Gries. 2011. Measures of speech rhythm and the role of
corpus-based word frequency: a multifactorial comparison of Spanish(-English) speakers.
International Journal of English Studies 11(2). 1-22.
Hayes, Bruce. 1989. The prosodic hierarchy in meter. Rhythm and Meter, ed. by Paul Kiparsky
and Gilbert Youmans, 201-260. Phonetics and Phonology 1, Academic Press, New York.
Hedberg, Nancy, Juan M. Sosa, and Lorna Fadden. 2004. Meanings and configurations of
questions in English. Proceedings of the 2
nd
International Conference on Speech
Prosody, Nara. Japan.
Hoequist, Charles. 1983. Syllable Duration in Stress-, Syllable- and Mora-Timed Languages.
Phonetica 40. 203-237.
Hualde, Jose Ignacio. 2005. The Sounds of Spanish. Cambridge, UK/NY: Cambridge University
Press.
138
Hualde, Jose Ignacio and Pilar Prieto. 2013. Intonational variation in Spanish: European and
American varieties. Intonational Variation in Romance, ed. by Sónia Frota and Pilar
Prieto, Oxford University Press, to appear.
Hurtado, Aida and Luis A. Vega. 2004. Shift Happens: Spanish and English Transmission
between Parents and Their Children. Journal of Social Issues 60. 137-55.
Jun, Sun-Ah. 2003. Prosodic Phrasing and Attachment Preferences. Journal of Psycholinguistic
Research. 32(2). 219-249.
Kim, Heejin and Jennifer Cole. 2006. Evidence for rhythmic shortening in American English as
conditioned by prosodic phase structure. Proceedings of 42nd annual Meeting of the
Chicago Linguistic Society.
Kireva, Elena. 2013. Prosodic transfer in a contact variety: The case of Olivenza Spanish. Poster
presented at the conference Phonetics and Phonology in Iberia (PaPI 2013), University
of Lisbon, Lisbon, Portugal.
Kohler, Klaus J. 2004. Pragmatic and Attitudinal Meanings of Pitch Patterns in German
Syntactically Marked Questions From Traditional Phonology to Modern Speech
Processing, ed. by G. Fant, H. Fujisaki, J. Cao, and Y. Xu, 205-214. Beijing: Foreign
Language Teaching and Research Press.
Kroll, Judith and Daan Hermans. 2011. Psycholinguistic perspectives on language processing in
bilinguals. Modeling Bilingualism. From Structure to Chaos: In Honor of Kees de Bot,
ed. by Monika Schmid and Wander Lowie, 15-36. Studies in Bilingualism, 43.
Amsterdam: John Benjamins Publishers.
Kvavik, Karen. 1988. Is there a Spanish imperative intonation? Studies in Caribbean Spanish
Dialectology, ed. by Robert M. Hammond and Melvyn C. Resnick, 35-49. Washington,
D.C.: Georgetown University Press.
139
Ladd, D. Robert. 1978. Stylized intonation. Language 54. 517‐540.
Ladd, D. Robert. 1980. The Structure of Intonational Meaning. Indiana University Press,
Bloomington and London.
Ladd, D. Robert. 1996, 2008. Intonational Phonology. Cambridge: Cambridge University Press.
Ladd, D. Robert and Astrid Schepman. 2003. Sagging transitions' between high pitch accents in
English: experimental evidence. Journal of Phonetics 31. 81–112.
Labov, William. 1966. The Social Stratification of English in New York City. Cambridge, U.K.:
CUP.
Levis, John M. 1999. The intonation and meaning of normal yes/no questions. World Englishes,
18. 373–380.
Lleó, Conxita, Martin Rakow, and Margaret Kehoe. 2004. Acquisition of language specific pitch
accent by Spanish and German monolingual and bilingual children. Spanish Laboratory
Phonology, ed. by Timothy Face, 3-27. Berlin/New York: Mouton de Gryter.
Lipski, John M. 2008. Varieties of Spanish in the United States. Georgetown University Press.
Lope Blanch, Juan. 1963. Sobre las vocales caedizas del español mexicano. Nueva Revista de
Filología Hispánica 17. 1-20.
López-Bobo, M.ª Jesús and Miguel Cuevas-Alonso. 2010. Cantabrian Spanish Intonation.
Transcription of Intonation of the Spanish Language, ed. by Pilar Prieto and Paolo
Roseano, 49-58. München: Lincom Europa.
Low, Ee. 1998. Prosodic Prominence in Singapore English. Unpublished doctoral dissertation.
University of Cambridge.
140
Low, Ee and Esther Grabe. 1995. Prosodic patterns in Singapore English. Proceedings of the
XIIIth International Congress of Phonetic Sciences 3. 636-639, Stockholm.
Low, Eee, Esther Grabe and Francis Nolan. 2000. Quantitative characterizations of speech
rhythm: Syllable-timing in Singapore English. Language and Speech 43. 377-401.
MacDonald, Marguerite. 1989. The influence of Spanish phonology on the English spoken by
United States Hispanics. American Spanish pronunciation: Theoretical and applied
perspectives, ed by. Bjarkman, Peter and Robert M. Hammond, 215-236. Washington,
DC: Georgetown University Press.
Martín Butragueño, Pedro. 2004. Configuraciones circunflejas en la entonación del español
mexicano. Revista de Filología Española 84, 2. 347-373.
Martín Butragueño, Pedro. 2006. Proyección sintáctico-discursiva de la entonación circunfleja
mexicana. El español en América. Diatopía, diacronía e historiografía. Homenaje a
José G. Moreno de Alba en su 65 aniversario, ed. by Company Company, 35-63.
UNAM, Mexico.
Matluck, Joseph H. 1951. La pronunciación en el español del Valle de México. UNAM
dissertation.
Miglio, Viola, Gries, Stefan and Michael Harris. 2013. New/given information in Mexican &
Chicano Spanish intonation. Poster presented at the 88th Annual Meeting of the
Linguistic Society of America, Minneapolis, Minnesota. January 2-5, 2013.
Montrul, Silvina. 2008. Incomplete Acquisition in Bilingualism. Re-examining the Age Factor.
Amsterdam: John Benjamins.
Mücke, Doris, Hosung Nam, Anne Hermes and Louis Goldstein. 2012. Coupling of tone and
constriction gestures in pitch accents. Consonant clusters and structural complexity, ed.
141
by Phi Hoole, Lasse Bombien, Marianne Pouplier, Christine Mooshammer and Barbara
Kühnert, 205-230. De Gruyter.
Nava, Emily. 2010. Connecting Phrasal and Rhythmic Events: Evidence from Second Language
Speech. University of Southern California dissertation.
Nava, Emily and Maria Luisa Zubizarreta. 2009. Order of Acquisition of Prosodic Prominence
Patterns: Evidence from L1Spanish/L2English Speech. Proceedings of the 3rd
Generative Approaches to Language Acquisition North America (GALANA 2008).
Somerville MA: Cascadilla Proceedings Project.
Navarro Tomás, Tomás. 1974. Manual de Pronunciación Española. Madrid: Consejo Superior
de Investigaciones Científicas.
Nespor, Marina and Irene Vogel. 1986. Prosodic Phonology. Dordrecht: Foris.
Oller, Kimbrough D. 1979. Syllable timing is Spanish, English and Finnish. Current issues in the
Phonetic Sciences: proceedings of the IPS-77 Congress, ed. by Hollien, Harry and
Patricia Hollien, 331-343. John Benjamin’s Publishing Company.
O’Rourke, Erin .2005. Intonation and language contact: A case study of two varieties of
Peruvian Spanish. University of Illinois at Urbana-Champaign dissertation.
O’Rourke, Erin. 2008. Correlating speech rhythm: Evidence from two Peruvian dialects.
Selected Proceedings of the Hispanic Linguistic Symposium. University of Western
Ontario, London, Ontario (Canada): October 19-22, 2006. Somerville, MA: Cascadilla
Proceedings Project.
O’Rourke, Erin. 2010. Ecuadorian Andean Spanish Intonation. Transcription of Intonation of the
Spanish Language, ed. by Pilar Prieto and Paolo Roseano, 227-253. München: Lincom
Europa.
142
Orozco, Leonor. 2008. Peticiones corteses y factores prosódicos. Fonología Instrumental,
ed. by Esther Herrera Zendejas and Pedro Martín Butragueño, 335-355. México: El
Colegio de México.
Ortiz, Héctor, Marcela Fuentes and Lluïsa Astruc. 2010. Chilean Spanish Intonation.
Transcription of Intonation of the Spanish Language, ed. by Pilar Prieto and Paolo
Roseano 255-283. München: Lincom Europa.
Otheguy, Ricardo and Ofelia García. 1993. Convergent conceptualizations as predictors of
degree of contact in U.S. Spanish. Spanish in the United States. Linguistic Contact and
Diversity ed. by Ana Roca and John Lipski, 135-151. Berlin and New York: Mouton de
Gruyter.
Otheguy, Ricardo, García, Ofelia and Fernández, Mariela. 1989. Transferring, switching, and
modeling in West New York Spanish: An intergenerational study. International Journal
of Social Language, 41-52. New York.
Perissinotto, Giorgio. 1975. Fonología del español hablado en la ciudad de México. Guanajuato,
Mexico: El Colegio de Mexico.
Phinney, Jean S, Irma Romero, Monica Nava, and Dan Huang. 2001. The Role of Language,
Parents, and Peers in Ethnic Identity Among Adolescents in Immigrant Families. Journal
of Youth and Adolescence 30. 135-53.
Pierrehumbert, Janet. 1980. The phonology and phonetics of English intonation. Indiana
University dissertation. Bloomington: Indiana University Press.
Pierrehumbert, Janet. 2000. Tonal elements and their alignment. Prosody: Theory and
Experiment. Studies. ed. by Merle Horne, 11-36. Dordrecht/Boston/London. Kluwer.
143
Pike, Kenneth L. 1945. The intonation of American English. Ann Arbor, MI: University of
Michigan Press.
Pitrelli, John, Beckman, Mary. E. and Julia Hirschberg. 1994. Evaluation of prosodic
transcription labeling reliability in the ToBI framework. ICSLP94 1. 123-126.
Potowski, Kim. 2012. Identity and heritage learners: Moving beyond essentializations. Spanish
as a Heritage Language in the US: State of the Science, ed. by Sara Beaudrie and Marta
Fairclough, 283-304. Georgetown University Press.
Potowski, Kim and Janine Matts. 2008. Interethnic language and identity: MexiRicans in
Chicago. Journal of Language, Identity and Education 7, 2. 137-160.
Prieto, Pilar Solé, Maria Josep and Joan Mascaró. 2006. Segmental and prosodic issues in
Romance linguistics. Amsterdam: John Benjamins.
Prieto, Pilar and Timothy Face. 2007. Rising accents in Castilian Spanish: a revision of Sp-ToBI.
Journal of Portuguese Linguistics 6, 1 (special issue on Prosody of Iberian Languages,
ed. by Gorka Elordieta and Marina Vigário), 117-146.
Prieto, Pilar and Francisco Torreira. 2007. The segmental anchoring hypothesis revisited.
Syllable structure and speech rate effects on peak timing in Spanish. Journal of
Phonetics 35, 4. 473-500.
Prieto, Pilar and Paolo Roseano. 2010. Transcription of Intonation of the Spanish Language.
Lincom Europa: München.
Prieto, Pilar, Maria del Mar Vanrell, Lluisa Astruc, Elinor Payne and Brechtje Post. 2010.
Speech rhythm as durational marking of prosodic heads and edges. Evidence from
Catalan, English, and Spanish. Speech Prosody 2010, Chicago, Illinois.
144
Quilis, Antonio. 1993. Tratado de fonología y fonética españolas. Madrid: Gredos.
Quilis, Antonio and Joseph Fernández. 1996. Curso de fonética y fonología españolas para
estudiantes angloamericanos. Consejo Superior de Investigaciones Científicas. Madrid.
Ramus, Franck. 2002. Language discrimination by newborns. Teasing apart phonotactic,
rhythmic, and intonational cues. Annual Review of Language Acquisition 2. 85-115.
Ramus, Franck, Marina Nespor and Jaques Mehler. 1999. Correlates of linguistic rhythm in the
speech signal. Cognition 73, 265-292.
Ritchart, Amanad and Amalia Arvaniti. 2014. The use of high rising terminals in Southern
Californian English. POMA 20.
Rivera-Mills, Susana. 2000. Un análisis comparativo del voseo salvadoreño. Paper presented at
82nd Annual International Conference of the American Association of Teachers of
Spanish and Portuguese (AATSP), San Juan, Puerto Rico.
Rivera-Mills, Susana. 2001. Acculturation and communicative need: Language shift in an
ethnically diverse Hispanic community. Southwest Journal of Linguistics, 20, 2. 211-
221.
Roach, Peter. 1982. On the distinction between “stress-timed” and “syllable-timed” languages.
Linguistic controversies: essays in linguistic theory and practice in honour of F.R.
Palmer, ed. by David Crystal, 73-79. London: Edward Arnold.
Robles-Puente, Sergio. 2011a. Looking for the Spanish Imperative Intonation: Combination of
Global and Pitch-Accent Level Strategies. Selected Proceedings of the 5th Conference
on Laboratory Approaches to Romance Phonology, ed. by Scott M. Alvord, 153-164.
Somerville, MA: Cascadilla Proceedings Project.
145
Robles-Puente, Sergio. 2011b. Absolute Questions Do Not Always Have a Rising Pattern:
Evidence from Bilbao Spanish. Selected Proceedings of the 5th Conference on
Laboratory Approaches to Romance Phonology, ed. by Scott M. Alvord, 98-107.
Somerville, MA: Cascadilla Proceedings Project.
Robles-Puente, Sergio. 2012. Two Languages, two intonations? Statements and Yes/No
Questions in Spanish and Basque. International Journal of Basque Linguistics and
Philology (ASJU), ed. by Campos-Astorkiza Rebeka and Jon A. Franco, to appear.
Schiering, René, Hildebrandt, Kristine and Balthasar Bickel. 2007. Crosslinguistic challenges for
the prosodic hierarchy: Evidence from word domains. The 14th Manchester Phonology
Meeting, Manchester, U.K..
Selkirk, Elisabeth O. 1980. The role of prosodic categories in English word stress. Linguistic
Inquiry 11, 563-605.
Selkirk, Elisabeth O. 1984. Phonology and Syntax: The Relation between Sound and Structure.
Cambridge: MIT Press.
Silva-Corvalán, Carmen. 1994. Language contact and change: Spanish in Los Angeles. (Oxford
studies in language contact). Oxford: Clarendon.
Silva-Corvalán, Carmen. 2001. Sociolingüística y pragmática del español. Washington, DC:
Georgetown University Press.
Sosa, Juan Manuel. 1999. La entonación del español: su estructura fónica, variabilidad y
dialectología. Madrid: Cátedra.
Sosa, Juan Manuel. 2003. Wh-questions in Spanish: Meanings and Configuration Variability.
Catalan Journal of Linguistics 2. 229-247.
146
Thomas, Erik and Holly A. Ericson. 2007. Intonational distinctiveness of Mexican American
English. University of Pennsylvania Working Papers in Linguistics 13.2: 193-205.
Thomas, Erik and Phillip M. Carter. 2003. A First Look at Rhythm in Southern African
American and European American English. New Ways of Analyzing Variation
(NWAV,32): Philadelphia, PA.
Thomason, Sarah G. and Terrence Kaufman. 1988. Language contact, creolization, and genetic
linguistics. Berkeley: University of California Press.
Torreira, Francisco and Simeon Floyd. 2012. Intonational Meaning: The Case of Spanish Yes-
No Questions. Paper presented at the Fifth European Conference on Tone and
Intonation, Oxford.
Vann, Robert E. 2009. On the importance of spontaneous speech innovations in language contact
situations. Convergence and Divergence in Language Contact Situations. ed. by Kurt
Braunmüller and Juliane House, 153–182. John Benjamins Publishing.
Weinreich, Uriel. 1953. Languages in contact: Findings and problems. Linguistic Circle of New
York. New York.
White, Laurence and Sven L. Mattys. 2007. Calibrating rhythm: First language and second
language studies. Journal of Phonetics 35. 501-522.
Williams, Briony and Steve M. Hiller. 1994. The question of randomness in English foot timing:
A control experiment. Journal of Phonetics 22. 423-429.
Willis, Erik W. 2002. Is there a Spanish imperative intonation revisited: Local considerations.
Linguistics 40. 347-74.
147
Xu, Yi. 2013. ProsodyPro - A Tool for Large-scale Systematic Prosody Analysis. Proceedings of
Tools and Resources for the Analysis of Speech Prosody (TRASP 2013), 7-10. Aix-en-
Provence, France.
Xu, Yi and Ching X. Xu. 2005. Phonetic realization of focus in English declarative intonation.
Journal of Phonetics 33. 159-197.
Zampini, Mary. 1998. L2 Spanish spirantization: A prosodic analysis, and pedagogical
implications. Hispanic Linguistics 10, 1. 154-188.
Zentella, Ana Celia. 1990. Defining social and linguistic variables in the study of subject
pronouns in U.S. Spanish. Paper presented at the XI congress Spanish in the United
States, Chicago, IL.
148
NOTES
1
Figure 1 is adapted from Prieto (2006) but there are different opinions on whether the utterance,
the clitic group or other domains under the syllable like the mora should be included in the
hierarchy.
2
The possible outcomes presented for rhythm and intonation can be more complex if bilinguals
can accommodate their prosody to the language they are producing. This point will be explored
in chapters 2 and 3.
3
English speakers were European Americans in Carter’s (2005) terminology. Speakers that have
been never been to the U.S. were recorded in Mexico.
4
Individual information on the participants can be found in Appendix A.
5
Although they were slightly older than other speakers in the same group, g3s2 and g3s4 were
considered L.A. Born Bilinguals since they were born in L.A. g2s4 was not considered a L.A.
Born Bilingual due to her age and because her family moved to L.A. a few weeks after she was
born.
6
Speakers g5s2 and g5s10 show high percentages of English in the last years (three and one
based on their ages) since they were taking classes in this language.
7
Given the complexity adding more variables would entail, multiple varieties within Mexico
were not considered. Nevertheless, having prosodic properties from different varieties is not
expected to interfere with the results since they are still distinct from English.
8
Although ideal control groups should be formed by monolingual speakers, it was not possible
to find enough participants who had null knowledge of English or Spanish. Results show that the
rhythms of these two groups should be close enough to those of English and Spanish
monolinguals, however.
9
Two of the speakers had visited the States in the past for short periods of time as indicated in
Appendix A.
10
These speakers reported that they learned English from other L2 English speakers.
11
Texts can be found in Appendix B.
12
A potential drawback of using readings is that participants’ productions may not always
resemble natural speech; however, results were similar to those reported by Carter (2005) with
149
natural data. See Harris and Gries (2011) for the interaction between rhythm and lemma
frequencies in natural speech.
13
In order to collect data from the large number of subjects needed for this study, recordings
were made in locations convenient to the participants. Thus, conditions were not always optimal.
One of the audio files belonging to the L.A. Born Bilinguals, two belonging to the Adult Late
Bilinguals and one belonging to Spanish Control Group were challenging to analyze with
vocDetect due to noise in the background. Since results were not reliable (i.e. some voiceless
sequences were being classified as voiced and vice versa), they were excluded from the analysis.
14
After all the analyses were made, one hundred and fifty syllables were re-measured randomly
to ensure consistency in the measurements both in English and in Spanish. None of the re-
measurements exceeded a 15ms difference.
15
Perissinotto (1975) and Lope Blanch (1963) notice that in Mexico City and the Valle de
Mexico, speakers tend to weaken unstressed vowels or vowels in certain consonantal contexts
(especially before /s/). None of the speakers that took part in the experiment showed consistent
weakening in Spanish.
16
Four of the speakers belonging to English Control Group did not complete the reading task for
the Spanish passage since they considered themselves English monolinguals. Two of the
Mexican Spanish speakers in Spanish Control Group did not read the English passage for the
same reason. Each subjects contributed to one number.
17
Nava (2010) does not report the scale used for the SD. Also notice that a higher voicing ratio
(voiceless/voiced) corresponds to a lower voicing %.
18
After removing the five most extreme outliers the predictability of the model was of 50% (R
2
= 0.4982).
19
The scores provided in Carter (2005) are 0.532, 0.426 and 0.279 since the author did not
multiply the resulting scores by 100.
20
The same classification is used in Nava (2010). I follow the labeling used for the same text in
Stresses from International Phonetic Association (1999). Handbook of the International Phonetic
Association. Cambridge University Press. Vowels with secondary stress were considered
unstressed vowels.
150
21
Since reading paces varied slightly, durations are normalized to capture differences between
groups better. The normalized values are the result of dividing the duration of each vowel by the
total duration of all vowels. After calculating the means for each subject, each subject
contributed one data point for the confidence intervals.
22
Linear regressions were run with the nPVI scores and the voicing ratios and the sum of all the
vowel type ratios. A significant model emerged for the nPVI: R
2
= 0.726; F(1,45) = 119.513; p <
.001 and another one for voiceless to voiced ratio: R
2
= 0.378; F(1,41) = 24.933; p < .001.
23
Adams (1979) finds the same behavior in L2 English speakers.
24
These findings were corroborated by Yolanda Congosto (personal communication) while
recording L.A. speakers for the ongoing project “Características Prosódicas del Español de los
Estados Unidos: los Ángeles, California”.
25
Although they did not seem to differ from monolinguals, bilinguals in the control groups were
not excluded and clear results were still obtained. It is expected that if these groups were formed
just by monolinguals, results should be even clearer.
26
These descriptions are based on the Autosegmental-Metrical (AM) model of intonational
phonology (Pierrehumbert and Beckman 1986, Ladd 1996, Gussenhoven 2004, Beckman et al.
2005, inter alia) and will be described in detail in the following sections.
27
Tables with the all the percentages per context and individual productions can be found in
Appendix D. Given the low number of tonal instances found in some contexts and in order to
capture general trends, statistical analyses were run considering several pragmatic meanings at
once. These analyses can be found in section 3.11.
28
It has to be taken into account that the L+>H* notation is quite recent and to the best of my
knowledge has not been used in English intonation. Therefore, like in Spanish, the distinction
between L+H* and L+>H* may not have been captured in previous works.
29
Micro prosodic variation caused by certain phonemes (e.g. small dips in the F0 due to voiced
obstruents) should not be taken into account.
30
Two of the English Control Group speakers and one of the Adult Late Bilinguals speakers only
took part in this experiment.
31
Each syllable was divided in fifteen identical sections.
151
32
Hualde and Prieto (forthcoming) label pitch-accents with no pitch excursions as L* or simply
* to indicate an accent without a tonal correlate. I follow Martín Butragueño (2006) and reserve
the L* notation for configurations with clear F0 falling movements.
33
This configuration could also be considered H*+L.
34
This configuration could also be considered H*+L.
35
Martín Butragueño (personal communication).
36
One speaker in English Control Group and one speaker in the Adult Early Bilinguals Group
did not know what armario ‘closet’ meant. In those cases the word ventana ‘window’ was used
instead.
37
A similar three way distinction is used in Ortiz and colleagues (2010) to distinguish levels of
prominence in biased statements in Chilean Spanish. This three way classification was supported
by perception results.
38
In some cases !H% and LH% boundary tones characteristic of Mexican Spanish statements
were found together with the four configurations. Those instances are marked in parentheses in
Tables D5-D8 (Appendix D).
39
A Wilcoxon signed-rank test revealed that significant differences between declarative and
imperative F0 values in configurations with downstep were found only for Imp2 and Imp3
contexts in both Spanish and English. Spanish: statement p2 vs. imp2 p2 (p = .022), statement p1
vs. imp3 p1 (p = .05), statement b2 vs. imp3 p2 (p = .021) and statement p2 vs. imp3 p2 (p =
.026). English: statement b1 vs. imp3 b1 (p = .05), statement p1 vs. imp3 p1 (p = .011),
statement b2 vs. imp3 p2 (p = .028) and statement p2 vs. imp3 p2 (p = .028). These results agree
with those found in Robles-Puente (2011).
40
Based on the F0 contours presented in de-la-Mota and colleagues (2010), some of the
configurations that are labeled as LH% may correspond to HH% since the F0 rise technically
starts at the beginning of the stressed syllable. Since there are no clear instances of LH% in the
current data, all similar configurations have been labeled as H%.
41
H% boundary tones refer both to H% and HH%.
42
Since measurements in the current study were calculated over an entire text, it could be the
case that some bilinguals were producing English-like rhythm in some parts and Spanish-like
152
rhythm in others (i.e. by using two different grammars).This possibility was ruled out by
quantifying their rhythms with different portions of the text and finding similar results.
43
g1 = English Control Group; g2 = Adult Early Bilinguals; g3 = L.A. Born Bilinguals; g4 =
Adult Late Bilinguals; g5 = Spanish Control Group.
44
Nuclear accent and semitones in bold mark utterances where the nuclear pitch-accent was
higher than the initial pre-nuclear pitch-accent. This information is explained in detail in section
3.7. (Imperatives).
Abstract (if available)
Abstract
Spanish and English rhythm and intonation have been described in detail
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
The phonology and phonetics of Turkish intonation
PDF
The Spanish feminine el at the syntax-phonology interface
PDF
Connecting phrasal and rhythmic events: evidence from second language speech
PDF
The locative alternation: unaccusative constructions and subject position
PDF
A reduplicative analysis of sentence modal adverbs in Spanish
PDF
Adding and subtracting alternation: resumption and prepositional phrase chopping in Spanish relative clauses
PDF
Prosodic recursion and syntactic cyclicity inside the word
PDF
Tone gestures and constraint interaction in Sierra Juarez Zapotec
PDF
Asymetrical discourse in a computer-mediated environment
PDF
Dynamics of consonant reduction
PDF
Building adjectival meaning without adjectives
PDF
Register and style variation in speakers of Spanish as a heritage and as a second language
PDF
Minimal contrast and the phonology-phonetics interaction
PDF
Superlative ambiguities: a comparative perspective
PDF
Copy theory of movement and PF conditions on spell-out
PDF
Agreement on the left edge: the syntax of left dislocation in Spanish
PDF
Asymmetry of scope taking in wh -questions
PDF
Sources of non-conformity in phonology: variation and exceptionality in Modern Hebrew spirantization
PDF
Investigating the production and perception of reduced speech: a cross-linguistic look at articulatory coproduction and compensation for coarticulation
PDF
Sound sequence adaptation in loanword phonology
Asset Metadata
Creator
Robles-Puente, Sergio
(author)
Core Title
Prosody in contact: Spanish in Los Angeles
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Linguistics (Hispanic Linguistics)
Publication Date
08/18/2014
Defense Date
05/09/2014
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
bilingualism,English,intonation,OAI-PMH Harvest,prosody,rhythm,sociolinguistics,Spanish
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Goldstein, Louis (
committee chair
), Saltarelli, Mario (
committee member
), Walker, Rachel (
committee member
), Zubizarreta, Maria Luisa (
committee member
)
Creator Email
roblespu@gmail.com,vipergt@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-461488
Unique identifier
UC11287091
Identifier
etd-RoblesPuen-2823.pdf (filename),usctheses-c3-461488 (legacy record id)
Legacy Identifier
etd-RoblesPuen-2823.pdf
Dmrecord
461488
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Robles-Puente, Sergio
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
intonation
prosody
rhythm
sociolinguistics