Indo-Trinidadian speech: An investigation into a ... .edu

June 15, 2018 | Author: Anonymous | Category: Documents
Share Embed

Short Description

Historical, social, and linguistic background of the Indo-Trinidadian population Ater the abolition of slavery in the Br...


John Benjamins Publishing Company

This is a contribution from English in the Indian Diaspora. Edited by Marianne Hundt and Devyani Sharma. © 2014. John Benjamins Publishing Company This electronic file may not be altered in any way. The author(s) of this article is/are permitted to use this PDF file to generate printed copies to be used by way of offprints, for their personal use only. Permission is granted by the publishers to post this file on a closed server which is accessible to members (students and staff) only of the author’s/s’ institute, it is not permitted to post this PDF on the open internet. For any other use of this material prior written permission should be obtained from the publishers or through the Copyright Clearance Center (for USA: Please contact [email protected] or consult our website: Tables of Contents, abstracts and guidelines are available at

Chapter 2

Indo-Trinidadian speech An investigation into a popular stereotype surrounding pitch Glenda Alicia Leung and Dagmar Deuber University of Freiburg / University of Münster

This paper examines the extent to which fundamental frequency (F0) contri­ butes to the stereotype that Indo-Trinidadians have a distinctive way of speak­ ing, namely that they have high-pitched voices in contrast to Afro-Trinidadians, who are perceived as having voices with a low tone. We report the results of an experiment in which the F0 of voice samples of both Afro- and Indo-Trinidad­ ians was acoustically modified to investigate this stereotype. Listeners were pre­ sented with unmodified as well as modified samples and were asked to identify the speaker’s ethnicity. Our results reveal that F0 is indeed one of the salient cues which Trinidadians rely on to distinguish ethnicity. In addition, phonation emerges as a potential ethnicity cue. Keywords: Indo- versus Afro-Trinidadians, ethnicity, perception, F0 manipulation

1. Introduction At the southernmost point of the Caribbean islands, Trinidad forms a two-island state together with the much smaller island of Tobago. While Tobago has not been significantly affected by Indian immigration, Indo-Trinidadians are now one of the two major ethnic groups in Trinidad, the other one being Afro-Trinidadians. In the 2000 census of Trinidad and Tobago, 40 percent of the country’s popula­ tion was recorded to be of Indian descent, 37.5 percent of African descent, 20.5 percent mixed and the remainder of unknown or other ethnicity (Central Intel­ ligence Agency 2013). The main languages of Trinidad are English, the official language, and a mesolectal English-lexifier creole. Indo-Trinidadians have shifted to English/Creole over time. According to a popular stereotype, their speech is © 2014. John Benjamins Publishing Company All rights reserved


Glenda Alicia Leung and Dagmar Deuber

characterized by a distinctive type of pitch, but there is little empirical evidence for this so far. After an outline of the historical, social, and linguistic background to the Indo-Trinidadian population (Section 2) and a brief sketch of related previous research (Section 3), this paper will present in its main part (Sections 4–7) an experiment designed to test the extent to which fundamental frequency (F0) – or pitch – is an ethnicity marker in Indo-Trinidadian speech. Conclusions and an outlook for further research will be given in the final Section 8.

2. Historical, social, and linguistic background of the Indo-Trinidadian population

After the abolition of slavery in the British West Indies in 1834–1838, a new source of labour was needed for the sugar plantations. Between 1845 and 1917, approximately 144,000 Indians came to Trinidad under an indentureship scheme (see Shepherd 2006, 306 for detailed figures of Indian migration to the West In­ dies). Only about 30,000 of these immigrants returned to India between 1850 and 1917 and some re-emigrated as a result of disappointment with what they found there (Haraksingh 2006, 280). The majority of Indian immigrants to Trinidad came from the Bhojpuri-­ speaking areas in the present-day states of Uttar Pradesh and Bihar (Mohan & Zador 1986, 294; Mohan 1990, 22). Although there were also speakers of other Indian languages among the immigrants, Bhojpuri established itself as the lan­ guage of the Indian community in Trinidad and came to represent the new Indo-­ Trinidadian identity (Mohan & Zador 1986, 294; Mohan 1990, 23).1 It was locally known as “plantation Hindi” rather than Bhojpuri, however (Mohan & Zador 1986, 294). Having been transported from a diglossic situation in which Hindi was the superordinate language (Mohan 1990, 25), it was generally considered a “broken” form of that language in Trinidad, which, it has been argued, contribut­ ed to a reluctance to maintain it (Mohan 1990, 21, 25). The first Indian indentured workers generally lived in barracks on the plan­ tations, while after 1870 village settlements became more common (Haraksingh 2006, 280). After the 1920s the life of Indians in Trinidad changed significant­ ly (Haraksingh 2006, 283). Increasing numbers moved away from agricultural environments, a development which heightened after the 1940s (Haraksingh 2006, 280). Opportunities for education, at first provided mainly by Canadian 1. According to Mohan & Zador (1986, 294), the only Indian language that survived in Trin­ idad beyond the immigrant generation was Tamil.

© 2014. John Benjamins Publishing Company All rights reserved

Chapter 2.  Indo-Trinidadian speech

Presbyterian missionaries (Samaroo 1975), were increasingly made use of and Indians began to enter professional occupations (Haraksingh 2006, 283). Since the 1940s, successive generations of Indians have achieved increasing econom­ ic and educational success and they have become progressively integrated into the larger Trinidad society (Haraksingh 2006, 284–286).2 In 1995–2001 the of­ fice of prime minister was held for the first time by a person of Indian ethnicity, Basdeo Panday. The country’s current Prime Minister, Kamla Persad-Bissessar, in office since 2010, is also of Indian ethnicity. Haraksingh (2006, 286) observes that many of today’s concerns of Indians in Trinidad are shared by Trinidadians of other ethnicities and that specific concerns are not so large as to create a major division in the society. “This”, he writes “is an indication of the depth of integra­ tion of Indians into Trinidad society, but more telling is the overriding tenor of life in Trinidad, which is more embracing and celebrates the country’s diversity” (Haraksingh 2006, 286). Bhojpuri has given way to English/Creole as the Indian population has be­ come assimilated. Mohan and Zador (1986) diagnosed language death in a study based on 1979 data from rural Caroni, where Bhojpuri had survived longer than anywhere else (Mohan & Zador 1986, 295; see also Bhatia 1988). The oldest native speaker of the language in their sample had been born in 1930. Among young­ er members of the community they were able to find only a few speakers, who, moreover, had only a limited, non-native knowledge of the language. The SIL Ethnologue gives the number 15,600 remaining speakers of Trinidad Bhojpuri in 1996, with the addition that those who use the language are older adults and that 90 percent or more of them reportedly have English/Creole as their first language (Lewis 2009).3 In another study based on the 1979 Caroni data, Mohan (1990, 30) concludes: “The linguistic future, then, would appear to be essentially identical to the socioeconomic future: an Indian community now completely transplanted in Trinidad, and essentially integrated into its new home”.

3. Previous studies of Indo-Trinidadian speech Although Indians are linguistically well-integrated into society, it is not implausi­ ble that an Indo-Trinidadian ethnolect may exist. Clyne (2000, 86) broadly defines ethnolects as “varieties of a language that mark speakers as members of ethnic 2. Transnational ties to India are not generally maintained by Indo-Trinidadians today. 3. Despite the dwindling number of speakers, it is noteworthy that there has been renewed interest in research on Trinidadian Bhojpuri, most notably in the research agendas of Jennifer De Silva (2012) and Kofi Yakpo (2012).

© 2014. John Benjamins Publishing Company All rights reserved



Glenda Alicia Leung and Dagmar Deuber

groups who originally used another language or distinctive variety”. A natural concern which arises in such language scenarios is the degree to which elements of the language originally spoken by the ethnic group are reflected in the new lan­ guage the group has acquired (be it in lexicon, grammar, phonology, or prosody). As usage of the original language decreases over time, “in the second generation and beyond, its symbolic significance as an identity marker is transferred to a variety of the majority language, the variety which is employed by the minority group either generally or in in-group situations” (Clyne 2000, 86). Very few studies have focused exclusively on ethnolects in Trinidad despite the lay belief that Indo- and Afro-Trinidadians “talk differently”. In a macro-so­ ciolinguistic study carried out in 1992, a majority of respondents thought that there was a difference between Indo- and Afro-Trinidadian speech; pitch, “ac­ cent”, and pronunciation were cited most often in this connection, followed by vocabulary (Mühleisen 2001, 56). Topics which have received some attention as regards Indo-Trinidadians and their English/Creole language usage include In­ dic loan words (Winer 2007), phonology (Winford 1978; Leung 2013), and pitch accent (Gooden, Drayton & Beckman 2009). Though indirectly related, work has also been done on the use of Indic lexicon by bi-racial Trinidadians of Indian and African descent (Regis 2012), locally known as douglas.4 To date, no study has investigated whether the grammar of Bhojpuri has had any effect on Trinidadian English/Creole. As regards lexicon, Winer (2007) has identified 1,844 words of Indian (main­ ly Bhojpuri) origin that can be considered part of the lexicon of Trinidad English/ Creole. Most of these words are from various specific domains such as religion, cooking/food, kinship, and agriculture. Some are also known and used by non-In­ do-Trinidadians, but most are associated with the language use of Indo-Trinidad­ ians, who, however, have varying knowledge of them. In terms of phonology, Winford’s major sociolinguistic survey conducted in 1970 indicated that old Indian speakers for whom Creole was a second language had a distinctive vowel system but that younger, first language speakers con­ verged with Afro-Trinidadians (Winford 1978). In a similar vein, Leung’s (2013) recent work on vowel variation in contemporary Trinidadian English indicates that at a segmental level there is little difference between the vowel realizations of Indo- and Afro-Trinidadians. That is not to say that differences do not exist at a suprasegmental level. In fact Gooden et al.’s (2009) study of Trinidadian Creole speech reveals that Indo-Trinidadians tend to have a low tone on stressed syl­ lables (L*) while Afro-Trinidadians use a long rise to peak on stressed syllables 4. The term dougla is derived from Hindi dogalā which means “hybrid: mixture; mongrel animal; person of mixed descent” (Winer 2008, 311).

© 2014. John Benjamins Publishing Company All rights reserved

Chapter 2.  Indo-Trinidadian speech

(L+H*). Furthermore, Drayton (2006) notes that “other Afro-Trinidadians who interact with Indo-Trinidadians also have L* accents instead of L+H* accents on stressed syllables” (qtd. in Gooden et al. 2009). 4. An investigation into a popular stereotype surrounding pitch As mentioned before, there is a commonly held folk belief that Indo-Trinidadians have a distinctive speech pattern. Indo-Trinidadian voices are often said to be light, soft, fine, and high in pitch, in contrast to Afro-Trinidadian voices, which are commonly described as strong, heavy, deep and low in tone.5 This stereotype can be well observed in a short cartoon called “Vendor Rivalry” (Sattar 2009) which won the Animae Caribe 2009 award for Best Animation. The video high­ lights ethnic pitch stereotyping by showing two vendors in full confrontation with each other: the Indo-Trinidadian vendor is represented as having a high pitched voice whereas the Afro-Trinidadian has a markedly deep voice. What lay people describe as pitch is in fact a perceptual phenomenon. Al­ though frequency and pitch are related, acoustically speaking the two are not the same. It is important to bear the distinction between production versus per­ ception in mind. In speech, humans produce frequencies which are measured in Hertz (Hz). Frequency in correlation to pitch is known as the fundamental frequency or F0. In contrast, when humans hear a voice, they perceive pitch or the “auditory realization [or] impression…of that property” (Ball & Müller 2005, 8). That is to say, in terms of acoustic and psycho-perceptual realities, humans pro­ duce frequencies, but hear pitch. For the sake of convenience, we use the word pitch to refer to F0 although we are fully aware of the technical difference between the two terms. As regards perceptual research on speaker ethnicity, an extensive amount of work has been done on European-American and African-American voices over the last 60 years (see Thomas & Reaser 2004 for a comprehensive overview; also Rahman 2008), ranging from studies on vowel quality to prosodic features, such as pitch, intonation, and phonation. Recent works outside of an African-Ameri­ can context include Mendoza-Denton’s (2011) research on Chicano gang perso­ nas, Szakay’s (2012) work on voice quality as an ethnicity marker in New Zealand, and Newmann and Wu’s (2011) work on Chinese- and Korean-Americans. Pho­ nation – which may manifest as modal, creaky, falsetto, or breathy, depending on the state of the glottis (Laver 1994; Gordon & Ladefoged 2001) – has also received 5. These descriptors were collected from our survey participants. (See Appendix, Section 2 of the survey.)

© 2014. John Benjamins Publishing Company All rights reserved



Glenda Alicia Leung and Dagmar Deuber

notable attention in relation to gendered ethnicity (Stuart-Smith 1999; Alim 2004; Yuasa 2010; Podesva 2013, and others). The current study endeavours to contribute to this growing body of research on speech perception by investigating the suprasegmental cues that Trinidadians use to assess ethnicity. We start out from the popular stereotype described above on the basis of which we have formulated the following three hypotheses to be tested in the study: H1: Indo-Trinidadians use a wider F0 range than Afro-Trinidadians. H2: Listeners will identify Indo-Trinidadians as Afro-Trinidadians if the overall F0 is decreased. H3: Listeners will identify Afro-Trinidadians as Indo-Trinidadians if the overall F0 is increased. After a detailed description of the method in Section 5, the results relating to these hypotheses will be reported in Sections 6.1–3, followed by an analysis of interspeaker variation in Sections 6.4–5 and further discussion of the results, also including findings on phonation in Section 7.

5. Method In order to examine the aforementioned hypotheses, an experiment was designed and presented in a survey format to Trinidadians. They were asked to listen to recordings and to identify the ethnicity of the speakers. Participants were given two options: they could either identify the speaker as Afro- or Indo-Trinidadi­ an. A follow up question “How sure are you?” probed the certainty of the par­ ticipant’s response, to which the following five response options were provided: extremely unsure, unsure, fairly sure, sure, and extremely sure. (See Appendix). It should be noted that we report only the results for ethnicity identification. A sample of 16 clips of Indo- and Afro-Trinidadian speakers was used, which was balanced for ethnicity and gender (4 clips per cell). The clips used were taken from sociolinguistic interviews conducted in Trinidad by the first author in 2009. Speaker names used in this paper are pseudonyms. Clips included neither Indic loanwords nor narratives of any theme which could bias the listener’s perception of the speaker’s ethnicity. F0 was modified in all of the original 16 clips, thus yielding 32 clips in total. Unmodified clips were not altered in any way, meaning that the F0 of the clips was not manipulated. In contrast, for the modified clips, the F0 was either increased or decreased by a constant semitone value, depending on the speaker’s gender: +/– 1.25 for females and +/– 1 for males. This was done © 2014. John Benjamins Publishing Company All rights reserved

Chapter 2.  Indo-Trinidadian speech

using Audacity, an audio editing software. F0 was manipulated by semitones since they are closer to human perception than Hz. How pitch was manipulated in the clips depended not only on the gender of the speaker but also on the speaker’s ethnicity: the F0 of the Indo-Trinidadian speakers was decreased whereas the F0 of the Afro-Trinidadian speakers was increased. Below is a summary of the mod­ ifications made: 1. 2. 3. 4.

The Indo-Trinidadian female voices were lowered by –1.25 semitones. The Indo-Trinidadian male voices were lowered by –1 semitone. The Afro-Trinidadian female voices were increased by +1.25 semitones. The Afro-Trinidadian male voices were increased by +1 semitone.

Although there was a total of 32 clips, each participant was only asked to judge 16 clips (8 unmodified and 8 modified). In order for all 32 clips to be evaluated, two versions of the survey with different clips were used for data collection. It should be noted that each listener never heard both the unmodified and modified voice of the same speaker. A total of 88 responses (male = 27, female = 61) was collected in March 2011. Each version of the survey was completed by 44 participants. Chi-square tests for independence were conducted to establish if there was a relationship between the dependent variable “ethnicity identification” (Afro- and Indo-Trinidadian) and the independent variable “clip status” (unmodified and modified). Chi-square tests were run both on pooled data (grouped by both ethnicity and gender) and individual speaker data in an R environment. Acoustic analysis was also conducted on the clips so as to ascertain if In­ do-Trinidadians use a wider pitch range than Afro-Trinidadians and whether Indo-Trinidadians have a higher mean F0. All clips were segmented into into­ national phrases, using the software Praat (Boersma & Weenink 2013). In Praat, the following F0 range settings were used according to the speaker’s gender: 75–300 Hz for males and 100–500 Hz for females. Exclusions were made in the measurement process wherever creaky voicing occurred and wherever there was overlap or background noise that affected the pitch curve. For each speaker’s clip, three F0 measurements (Hz) were made: the pitch floor, the pitch ceiling, and the mean F0 per intonational phrase. The pitch floor (F0 minimum) and pitch ceil­ ing (F0 maximum) were used to calculate pitch ranges across gender and ethnic groups. Pitch range was calculated as the F0 difference between the pitch ceil­ ing and pitch floor. Additionally, the measurement of mean F0 per intonational phrase was used to calculate the overall mean F0. Wilcoxon tests were conduct­ ed to look for differences in F0 values between the various ethnic and gender groupings. © 2014. John Benjamins Publishing Company All rights reserved



Glenda Alicia Leung and Dagmar Deuber

This study was originally designed to test F0 as an ethnicity assessment cue. However, during the course of examining interspeaker variation, phonation – namely, creaky, and modal voicing – emerged as a potential factor for ethnicity identification. Although our research was not designed to deal with phonation, we decided to report it because of its apparent relevance to the results. A simple binary coding scheme, referring to “creaky and modal” voice within the intona­ tional phrase, was used to classify phonation in unmodified clips.6 6. Results 6.1

Hypothesis 1: Difference in F0 range between ethnicities

As a point of comparison, we referred to Gut’s (2009) values for mean F0 (av­ erage pitch) and F0 range (pitch range) for men and women. As can be seen in Table 2.1, our results were consistent with mean F0 and F0 pitch ranges for both males and females. Pitch ranges for the Indo- and Afro-Trinidadian men illustrat­ ed little difference; both the pitch floor and ceiling were nearly identical. However, among the women, differences in pitch range were noticeable. The Indo-Trinidad­ ian women had both a higher pitch floor and ceiling than the Afro-Trinidadian women. Moreover, the Indo-Trinidadian women had a considerably wider pitch range (351 Hz) than their Afro-Trinidadian counterparts (258 Hz).7 Further­ more, Wilcoxon tests revealed differences in mean pitch between the Afro- and Indo-Trinidadian women to be statistically significant, W (159) = 1267, p 
View more...


Copyright © 2017 DOCIT Inc.