Skip to main content

The emergence and evolution of linguistic tone

Periodic Reporting for period 3 - EVOTONE (The emergence and evolution of linguistic tone)

Reporting period: 2021-06-01 to 2022-01-31

EVOTONE studies the origin, acquisition, and evolution of linguistic tone: using pitch to distinguish between the meaning of words. Despite the global presence and typological ubiquity of tone, there has never been a systematic analysis of the principles that govern the evolution of tone systems. Linguists still do not have a complete phonetic, structural, and psychological model that explains how and why tones emerge (or fail to emerge) in language after language, and how they evolve once they are formed.

The EVOTONE project has three objectives: first, to provide a new, empirically based foundation for the origin and typology of tone systems; second, to study how structural and phonetic factors interact in sound change; third, to develop the first empirically grounded set of principles for tonal evolution. We address these using a combination of linguistic fieldwork, acoustic-phonetic laboratory studies, and computational analysis of language data at scale.

In its early stages, the project brings together a broad international network of scholars, including many students and early-career researchers in and from Southeast Asia. In the long term, EVOTONE will create a more nuanced understanding of how tone is produced and perceived. This is not only of abstract scientific interest. It will also help drive substantial improvements to speech technology – automated recognition, understanding, transcription, and production of human speech – for billions of speakers of tonal languages; languages that are typically marginalized because they have linguistic characteristics, including a wide variety of interactions between tone and phonation, that are not addressed by mainstream, national-language technological development.

EVOTONE recognizes the scientific centrality of these less commonly spoken languages, while helping to lay the foundation for more ambitious documentation and revitalization projects – critical in an era of widespread language extinction. In addition to resolving outstanding questions about tonogenesis and tonal evolution, the results will substantially advance our more general understanding of how language as a whole changes over time.

In sum, the EVOTONE project will have considerable societal impact; demonstrating the EU’s international commitment to documenting and preserving the languages and cultures of developing economies, and strengthening cultural and research ties between Europe and Southeast Asia.
Overall the project is proceeding as planned. To better understand how tones emerge, we have conducted a number of experimental studies and undertaken data collection from a variety of languages, including non-tonal languages like French, Italian, German, and English, tonal languages, like Thai and Cantonese, and languages at various stages of tonal development such as Khmu. Our findings thus far are consistent with our initial hypothesis: the emergence of tone is driven by abrupt shifts in the availability of certain acoustic dimensions relevant for the perception of speech sounds, rather than by a gradual enhancement of properties associated with tones. Our data also indicate a substantial role for individual differences in perceptual strategies. Some of these findings have already been disseminated in peer-reviewed publications and at international conferences and colloquia.

To extend our understanding of how tones change over time, we have focused on variation and change of lexical tones in Thai, a language in which tonal realization varies considerably with age, background, and speech style. In addition to assembling a corpus of spoken Thai encompassing speakers from diverse range of ages and backgrounds, we are leveraging archival data to study how the tonal systems of individuals change over their lifetimes.

Supporting both of the above aspects, we continue to work on our Linguistic Data Warehouse, a collection of dialect surveys, comparative dictionaries, and lexical resources for around 1,000 languages in the Asia-Pacific region. This resource, which will be made freely available upon completion, will allow us analyze dozens of potential characteristics on hundreds of languages simultaneously, in order to precisely quantify the structural characteristics of languages that correlate with tonality. In addition to the data itself, our team develops tools and software to facilitate access to these resources in a meaningful way.
Our experimental work on the relationship between speech production and speech perception has already begun to help us to better understand the conditions under which it is favorable for tones to emerge. Ongoing work will allow us to extend this to the domain of tonal change.

When it becomes possible to resume fieldwork, we will advance the state of the art in instrumental studies of rare sound patterns such as voiceless sonorants. Languages which retain these sound patterns provide us with valuable "living fossils", allowing us to gain insight in the pre-conditions necessary, or at least favorable, for linguistic tone to emerge.

Finally, the development of the Linguistic Data Warehouse will allow us to test the relative contribution of different structural aspects statistically for the first time. Such an analysis has never been undertaken, due at least in part to the challenges of assembling the necessary database. We anticipate that both the database and the generalizations we draw from it will have a substantial and lasting impact on the study of tone systems, and language typology more general, for years to come.