Elsevier

Journal of Phonetics

Volume 94, September 2022, 101163
Journal of Phonetics

Research Article
American English pitch accents in variation: Pushing the boundaries of mainstream American English-ToBI conventions

https://doi.org/10.1016/j.wocn.2022.101163Get rights and content

Highlights

  • African American English, Jewish English, and Appalachian English were studied.

  • The three varieties show phonetic differences in pitch accents labeled as H* and L+H*.

  • The three varieties differ in the rate of use of these pitch accents, and their position in the phrase.

  • MAE-ToBI has strengths and weaknesses for varieties which vary from standardized versions of English.

Abstract

Linguists interested in intonation have long struggled to establish a maximally broad set of annotation conventions that function equally well across varieties of American English. The current study tests the advantages and limitations of the widely-used MAE-ToBI conventions, focusing on the H* and L+H* distinction, for three varieties of American English: African American English, Appalachian English, and Jewish English. Results of quantitative analysis of production data from 30 speakers of the three varieties finds major differences in rate of use of the H* and L+H* pitch accent as well as the phonetic realizations of these pitch accents, which may not be captured solely using the MAE-ToBI conventions. These differences appear not only between MAE-ToBI and the other three varieties, but also between the varieties themselves in unique ways that may shed light on the nature of sociolinguistic variation at the level of intonation, as well as the debated status of the distinction of H* vs. L+H* as a phonological or phonetic distinction. These findings provide further motivation for the development and use of annotation systems that explicitly consider sociolinguistic variation as well as phonetic parameters. Such systems will become even more essential as both sociolinguists and phoneticians expand intonational analysis beyond so-called “standard varieties” in order to arrive at a richer and more accurate picture of the intonational system of American English.

Introduction

Intonational differences between varieties are often salient to both laypeople and linguists. However, describing and formalizing these differences has posed many challenges. Part of this difficulty comes from the fact that while there is an agreed upon annotation standard for segmental features, the International Phonetic Alphabet, an equivalent set of conventions is lacking for intonation. While the IPA is not truly purely phonetic, as the set of symbols is driven by the goal of being able to notate all of the distinct phonemes in all the world’s spoken languages, it does also allow for researchers to do a range of transcriptions from more broadly phonemic to more closely phonetic. Different types of transcriptions are useful for different things. For example, a very broad phonological transcription allows for researchers to do acoustic analyses on segments which differ significantly in their phonetic realization while accounting for phonological structure. A more narrow transcription allows for initial observations of differences in the field, or for analyses of allophonic variation.

Starting in the 1980s and 1990s and continuing to this day, linguists developed a number of variety-specific Tone and Breaks Indices (ToBI) annotation systems for intonation (see Beckman et al., 2005, Ladd, 2008 on the history of the development of the systems, and Jun, 2005, Jun, 2014 for an overview of different ToBI systems), based on the principles of autosegmental/metrical phonology. ToBI annotation systems are meant to be more phonological than phonetic, with the annotations indicating phonological tones.

The fact that ToBI systems are phonological, and are phonetically underspecified, has both benefits and drawbacks. One drawback is that ToBI systems have always been designed to be variety-specific, in contrast to the purported universality of the International Phonetic Alphabet. For example, English ToBI was intended to cover “general American, standard Australian, and southern British English” (Beckman & Elam, 1997, pg. 8); to further emphasize this, the variety for American English is specifically called Mainstream American English (MAE) ToBI. As such, researchers using this system to transcribe any variety of spoken American English which differs significantly from this imagined standard must make decisions about whether, how, and to what extent, MAE-ToBI can be used to transcribe the data without modification. Unfortunately, particularly in the United States, there remains a dearth of ToBI systems for non-Mainstream regional and ethnic varieties, so researchers frequently employ MAE-ToBI as the closest available annotation system (Thomas, 2015).

The explicitly phonological nature of ToBI has also led to extensive debates over the line between phonetics and phonology; specifically, whether some pairs of MAE-ToBI labels represent phonetic, rather than phonological, distinctions. In particular, there have been debates over whether the H* and L+H* labels, described in more detail below, represent two distinct phonological categories, or different phonetic realizations of the same category.

However, the phonological aspect of the system can also be a benefit, allowing us to do similar types of analyses to those that are done on segmental featuresː for example, in looking at phenomenon like /u/ fronting, researchers make an implicit claim that Californian English and Eastern New England both have a phonological category in the word boot distinct from bead, bode, and book; we can then look at variation in the phonetic realization of that category. This underspecification can also be useful in studying the existence of and/or location of category boundaries between phonemes in varieties with very different phonetic realizations of those categories (e.g., the distinction between /ɔ/ and /ɑ/ by a speaker with the Northern Cities Vowel Shift, where /ɔ/ is lowered to [ɑ], and, /ɑ/, fronted to [a], compared to a speaker from New York City, where /ɔ/ can be raised almost to [ʊe]).

In this paper, we explore to what extent MAE-ToBI is appropriate for use on three varieties of American English. Although some researchers have explored this problem for other varieties of English, including varieties spoken in the United Kingdom, including Glaswegian English (Mayo, Aylett, & Ladd, 1997), New Zealand English (Warren, 2005) and Australian English (Fletcher & Stirling, 2014), this remains an understudied area for varieties of English spoken in the United States. The varieties of interest —African American English (AAE), (American) Jewish English1 (JE), and Appalachian English (ApE)— have been found to show variation in their intonation systems from more standardized varieties, particularly in their use and phonetic realization of H* and L+H* pitch accents.

In this paper, we focus in on (1) whether these varieties show evidence for phonetically distinct H* and L+H*s, and (2) whether these varieties use these pitch accents in different ways. This study has immediate practical implications, including in how MAE-ToBI training materials, such as the MIT Open Courseware course on transcribing prosodic structures using ToBI (Veilleux, Shattuck-Hufnagel, & Brugos, 2006) should describe the H* / L+H* contrast. It also has more theoretical ones. Previous work has suggested that the presence or lack of a distinction between H* and L+H* may be dialect dependent in the United States (Arvaniti & Garding, 2007). This work adds to that debate by showing that how the distinction is made between these pitch accents appears to be variety dependent.

Section snippets

Variation in pitch accents

The current study works within the autosegmental/metrical approach to intonational phonology (see Ladd, 2008 for an overview). In this framework, intonational melodies (tunes) are decomposable into a series of underlying tones. This is in contrast to other theories (e.g., the “British school”, see e.g., Crystal, 1969) which take tunes (e.g., a low rise, a rise-fall) as primitives. Building off of autosegmental theories of lexical tone (Goldsmith, 1976), tones are either low (L) or high (H);

Methods

Each author interviewed 10 female speakers of each variety. The authors were all speakers of the variety in question for which they did the interviews. For the purposes of this paper, we take a more traditional sociolinguistic approach to linguistic variation (see Eckert, 2012 for a more detailed history of the field) with a speaker’s variety (and indeed, the existence of a particular variety) being defined based on set social characteristics: for Appalachian English, residing in Appalachia

Relative use of H* and L+H*

As noted above, the data were coded using MAE-ToBI, and thus the researchers approached the data with an a priori assumption that there was a distinction between H* and L+H*; acoustic evidence for the existence of this distinction in all three varieties will be laid out in more detail below in Section 4.2.

Table 1 summarizes information about the passages presented above in Section 3, as well as the average length of the passage, average number of IPs, and average total pitch accents (all pitch

Discussion

The common finding across all of these models were differences between H* and L+H* within each variety; however, the phonetic details and use of these pitch accents appeared to differ between the varieties. First, there are statistically significant differences between the pitch accents coded as H* and those coded as L+H*, with L+H* having higher peaks, steeper slopes, larger rise spans, and later peak alignments than H*.

However, we see differences in the exact phonetic implementation of these

CRediT authorship contribution statement

Rachel Steindel Burdin: Conceptualization, Methodology, Data curation, Formal analysis, Writing – original draft. Nicole R. Holliday: Conceptualization, Methodology, Data curation, Writing – review & editing. Paul E. Reed: Conceptualization, Methodology, Data curation, Writing – review & editing.

Acknowledgements

The authors would like to thank Emelia Benson Meyer for help with data processing, as well as audience members at ETAP-4 and Speech Prosody 2018. An initial analysis of the L+H* data appeared in the conference proceedings of the latter.

References (68)

  • D. Bates et al.

    Fitting linear mixed-effects models using lme4

    Journal of Statistical Software

    (2015)
  • K. Becker

    (r) we there yet? The change to rhoticity in New York City English

    Language Variation and Change

    (2014)
  • Beckman, M. E., & Elam, G. A. (1997). Guidelines for ToBI labelling, version...
  • Beckman, M. E., Hirschberg, J., & Shattuck-Hufnagel, S. (2005). The original ToBI system and the evolution of the ToBI...
  • Boersma, P., & Weenink, D. (2016). Praat: doing phonetics by computer. Version 6.0.12. Retrieved from...
  • M. Bucholtz

    Sociolinguistic nostalgia and the authentication of identity

    Journal of Sociolinguistics

    (2003)
  • R.S. Burdin

    Variation in form and function in Jewish English intonation

    (2016)
  • R.S. Burdin

    New notes on the rise-fall contour

    Journal of Jewish Languages

    (2017)
  • R.S. Burdin

    The perception of macro-rhythm in Jewish English

    American Speech

    (2020)
  • S. Calhoun

    The centrality of metrical structure in signaling information structure: A probablistic perspective

    Language

    (2010)
  • E. Chodroff et al.

    Information structure, affect, and prenuclear prominence in American English

  • Cole, J., Britt, E. R., Thomas, E. R., & Coggshall, E. L. (2005). Intonational Distinctiveness of African American...
  • N. Corver et al.

    Semi-lexical categories: The function of content words and the content of function words

    (2001)
  • D. Crystal

    Prosodic systems and intonation in English

    (1969)
  • Dainora, A. (2002). Does intonational meaning come from tones or tunes? Evidence against a compositional approach. In:...
  • A. Dainora

    Modelling intonation in English: A probabilistic approach to phonological competence

  • P. Eckert

    Three waves of variation study: The emergence of meaning in the study of sociolinguistic variation

    Annual Review of Anthropology

    (2012)
  • A. Farinella et al.

    Biases from MAE-ToBI intonational transcription conventions in the intonational analysis of African American English

    Presented at NWAV 49 (University of Texas at Austin/Virtual)

    (2021)
  • J. Fletcher et al.

    Intonational variation in four dialects of English: The High Rising Tune

  • J. Fletcher et al.

    Prosody and discourse in the Australian English Map Task

  • D.L. Gold

    Jewish English

  • J. Goldsmith

    An overview of autosegmental phonology

    Linguistic Analysis

    (1976)
  • S. Gooden

    Authentically Black, bona fide Pittsburgher: A first look in African American women’s language in Pittsburgh

  • S. Gooden et al.

    Tone inventories and tune-text alignments

    Studies in Language

    (2009)
  • Cited by (0)

    View full text