Tonogenesis

Vietnamese (Vietic, Mon-Khmer, Austroasiatic) is monosyllabic and tonal.  Most Mon-Khmer (MK) languages are multisyllabic and atonal.  Evidence suggests that Vietnamese (VN) has had its tones less than one millennium, and that other languages (both MK and non-MK) are also acquiring tones, a process called tonogenesis.
    Comparing VN’s phonology to that of closely related languages, Haudricourt has identified several stages of tonogenesis.  We know that Vietnamese tonogenesis included monosyllabization, loss of onset voicing distinctions, and loss of laryngeal codas, but how and why, and in what order?  And why do other languages follow similar paths?
    Usage-based phonology claims that language change is driven by articulatory reduction, which motivates similar paths of change across related and unrelated languages.  If speakers use a word often, they conserve effort by saying it the ‘easy’ way.  And even if a word is reduced, its frequency helps listeners to perceive it.  Frequent words reduce first, then infrequent words.
    I argue that tonogenesis can be given such an articulatory account, and I detail what such an account should include.  Comparing VN’s sound changes to those in other MK languages, and to reductive processes noted by vocal physiology and articulatory phonetics, I demonstrate that each change along the tonogenetic path is an articulatory reduction.  There is much about past tonogenetic processes that we cannot test, but the most recent of those processes are still synchronically evident in comparisons of frequent to infrequent words.
    To show that a usage-based account of tonogenesis is supported by synchronic facts, I analyze the spontaneous conversations and directed readings of 16 VN speakers in Hanoi and near Saigon.  Acoustic measurements reveal that (relative to infrequent words) VN’s frequent words are reduced in three ways: they exhibit less negative spectral tilt, less duration, and less pitch excursion.  The significance of each is explained.
    Empirical results support the usage-based, articulatory account that I propose.  Future studies of sound change will gain useful diachronic insights by including considerations of usage frequency and articulatory reduction.

Asian languages reveal a continuum from quasi-disyllables to highly eroded monosyllables. These variegated states are now understood to be different stages along a common evolutionary path. An overview is proposed, beginning with the earliest stages of monosyllabicization, before the generalization of a strictly monosyllabic phonological structure. The well-documented processes of consonantal depletion which lead to the development of phonation-type registers and tones are presented within a general model. Lastly, phonological evolutions observed at advanced stages of segmental depletion are discussed. The concluding note concerns the aftermath of segmental depletion: the re-creation of polysyllables.

This paper provides a systematic account of the emergence of contour tone, based on a ‘syllable-tone-register’ model and a large body of new firsthand acoustic tonal data. The emergence of tone is a process of pitch upgrading from an accompanying feature of phonation types to a distinctive feature. It involves three linguistic prerequisites, which bring about three processes: (L1) various phonation types, clear voice, breathy voice, falsetto, etc., which intrinsically induce pitch differences. (L2) Mono-syllabic (with great sonorous power) morpheme structure that enhance the pitch differences. (L3) Phonologicalization, which conceptualizes the enhanced pitch differences to a phonological category, tone. Furthermore, this paper proposes two ecological preconditions behind the linguistic prerequisites: (N1) Humid climates, which tend to induce a breathy voice. (N2) Small body-size, which tends to induce falsetto and creaky voice. A flowchart is designed to show how these prerequisites lead to different types of the world’s languages and finally output contour tone.

Tonogenesis is the development of distinctive tone from earlier non-tonal contrasts. A well-understood case is that of Vietnamese (similar in its essentials to that of Chinese and many languages of the Tai-Kadai and Hmong-Mien language families), where the loss of final laryngeal consonants led to the creation of three tones, and the tones later multiplied as voicing oppositions on initial consonants waned. This is by no means the only attested diachronic scenario, however. There is tonogenetic potential in various series of phonemes: glottalized vs. plain consonants, unvoiced vs. voiced consonants, geminates vs. simple (and, more generally, tense vs. lax), and even among vowels, whose intrinsic pitch can transphonologize to tone. But the way in which these common phonetic precursors to tone play out in a given language depends on phonological factors, as well as on other dimensions of a language’s structure and on patterns of language contact, resulting in a great diversity of evolutionary paths in tone systems. In some language families (such as Niger-Congo and Khoe), recent tonal developments are increasingly well-understood, but working out the origin of the earliest tonal contrasts (which are likely to date back thousands of years earlier than tonogenesis among Sino-Tibetan languages, for instance) remains a mid- to long-term research goal for comparative-historical research.

Unlike many languages of Southeast Asia, Khmer (Cambodian) is not a tone language. However, in the colloquial speech of the capital Phnom Penh, /r/ is lost in onsets, reportedly supplanted by a range of other acoustic cues such as aspiration, a falling- or low-rising f0 contour, breathy voice quality, and in some cases diphthongization, e.g. /krɑː/ ‘poor’ > [kɔ̀ɑ], [kʰɔ̌ɑ], [kɔ̤ɑ̤], /kru:/ ‘teacher’ > [khùː] [kʰǔː], [kṳː]. This paper presents the results of production and perception studies designed to shed light on this unusual sound change. Acoustic evidence shows that colloquial /CrV/ forms differ from reading pronunciation forms in terms of VOT, f0, and spectral balance measures, while a pair of perceptual studies demonstrate that f0 is a sufficient cue for listeners to distinguish underlying /CrV/-initial from /CV/-initial forms, but that F1 is not. I suggest that this sound change may have arisen via the perceptual reanalysis of changes in spectral balance, coupled with the coarticulatory influence of the dorsal gesture for /r/.

Suprasegmental contrasts of tone and register are commonplace phonological phenomena among the languages of Mainland Southeast Asia and its periphery (MSEA) (Matisoff 1990, 2001). Insofar as we have come to understand the origins and evolution of such contrasts, two theories predominate: tonogenesis (Haudricourt 1954) and registrogenesis (Huffman 1976). In their classical forms, tonogenesis and registrogenesis are well suited for modeling the development of tone and register in the best known, most studied languages of MSEA, but there is much additional complexity that they fail to capture. This is especially true for languages of Austroasiatic stock, which in many cases have developed tone and register in ways that must be considered ‘unorthodox’ with respect to the received models (Ferlus 1979, 2004, 2011; Diffloth 1982a, 1982b; Svantesson 1989; Gehrmann 2015; Sidwell 2015, 2019).

The goal of this thesis is to present a possible way forward towards a unified conceptual framework for tone and register evolution in the languages of MSEA: desegmentalization. Expanding on Dockum’s (2019) concept of desegmental phonology, desegmentalization is the process by which one or more segmental properties (onset phonation, vowel height, vowel length or coda phonation) condition changes in the distribution of a language’s suprasegmental contrasts. A general survey of the Austroasiatic language family is presented, in which documented examples of desegmentalization are presented and discussed. Austroasiatic constitutes a useful laboratory for such a survey, because the identification of the segmental origins of suprasegmental contrasts in Austroasiatic languages is relatively straightforward in comparison to the other language families of MSEA. Based on this survey of desegmentalization processes in Austroasiatic, ten discrete desegmentalization models are proposed. The output typologies for the suprasegmental contrasts produced by each model are compared and implications for a general model of tonogenesis and registrogenesis are explored.

This thesis offers (1) a digestible introduction for the non-specialist to the historical development of suprasegmental contrast in MSEA, (2) a resynthesis of current tonogenetic theory which integrates classical tonogenesis, classical registrogenesis and various other, lesser-known evolutionary pathways under the larger umbrella of desegmentalization and (3) a comprehensive overview of tone and register origins in the Austroasiatic family.

This paper presents some novel and hard-to-access data from Kalkoti, an Indo-Aryan language spoken in northern Pakistan. The particular focus is on showing how this Shina variety in a relatively short time span has drifted apart from its closest known genealogical relatives and undergone significant linguistic convergence with a Kohistani variety in whose vicinity Kalkoti is presently spoken. Among other features, we explore what seems like an ongoing process of tonogenesis as well as structural “copying” in the realm of tense and aspect.

This paper discusses the diachronic processes of tonal development of Tai languages. Tonogenesis is treated as the very first step of the arising of tones in proto-Tai. After tone arose, two groups of proto-initials (voiceless-voiced) conditioned primary tonal splits whereas splits conditioned by other pharyngeal features, such as aspiration and preglottalization represent stages of secondary development. Some secondary tonal split patterns result in three-way register splits. However, the bearing capacity of the tone numbers in a language often cause the third register to merge into one of the other two to make two-way register splits always be the mainstream.

This paper aims to investigate how tonogenesis in Seoul Korean affects non-native language acquisition in a multilingual context with L1 Cantonese, L2 English and L3 Korean. We are also interested to examine the source and direction of cross-linguistic influences in the acquisition of VOT and F0 patterns which are used for the stop contrasts. Sixteen L3 Korean learners differing in proficiencies (beginning vs. intermediate) were recorded reading a list of words embedded in carrier sentences in the three languages. Results from acoustic analyses found that the learners did not merge VOTs for the Korean lax and aspirated stop contrast, but they showed the tendency for tonogenesis, such as their heavy use of f0 in producing different Korean stops. Furthermore, comparison between the beginners and the intermediate learners revealed various cross-linguistic influences: regressive transfer of f0 on L1 and L2, bidirectional interaction of f0 values between L2 and L3, and L1 influence on the production of VOT in L2 and L3. These findings suggest that L3 transfer happens in a cue-by-cue manner, i.e. VOT and f0 need to be accounted for separately.

In the colloquial Phnom Penh dialect of Khmer (Cambodian), lexical use of F0 is emerging together with an intermediate VOT category and breathy phonation following the loss of /r/ in onsets (e.g. /kruː/ ‘teacher’ > [khṳ̀ː]). I show how this incipient tonogenesis might arise in a series of computational simulations tracing the evolution of multivariate phonetic category distributions in a population of ideal observers. Acoustic production data from a fieldwork study conducted in Phnom Penh was used as the starting point for the simulations. After establishing that the basic framework predicted relative stability over time, two possible responses to a phonetic production bias were considered: one in which agents correctly identified the source of (and thereby compensated for) the effects of the bias, and one in which agents misattributed the acoustic effects of the bias as a property of the onset. Good qualitative fits to the empirical production data were found for the latter group of learners, while the outcome for compensating learners resembled production data from a related dialect. These results are consistent with the sudden and discontinuous nature of many sound changes, and suggest that what appear to be enhancement effects may also emerge under different assumptions about the number of cue dimensions accessible to or deemed relevant by the learner.

The development of phonological tone has been linked in many languages to consonant voice quality contrasts that impart pitch differences to preceding or following vowels. In particular, modal voicing on a syllable-initial consonant has been shown to correlate with low pitch on a following vowel in both tone and non-tone languages (Hombert et al. 1979). In the Austronesian language Malagasy, a strong relationship between consonant voicing and vowel pitch has been observed in the literature; previous studies state, however, that Malagasy does not have phonological tone. In contrast with these previous descriptions, the results presented in this paper from a production study of homorganic fricative pairs in several different dialects of Malagasy suggest that in certain dialects in the center of the country, in and around the capital city, pitch has replaced modal voicing as the primary phonetic cue to fricative voicing category. Based on word list data from 11 Malagasy speakers, this study finds that the homorganic fricatives of speakers of the Central dialects are best classified based on the pitch of the following vowel, while those of speakers of non-Central dialects are best classified based on the duration of modal voicing on the fricative. Secondary phonetic cues typically associated with a voicing distinction (i.e., duration, frication intensity) appear to be undergoing neutralization among the Central speakers.

This paper focuses on the phonetic nature and path of evolution of Cantonese short-long stopped tones. Firstly, the phonetic nature of Cantonese stopped tones is explored in light of recent tonal acoustic studies. Secondly, drawing on examples from other Southeast Asian languages, an asymmetry of pitch differences is shown to exist in languages with both short and long stopped tones: the ‘short-high VS long-low’  pattern is much more common than the ‘short-low VS long-high’ one. An evolutionary perspective on this synchronic asymmetry is proposed, in light of (i) Ohala's (1981) listener-based perspective on sound change and hypercorrections, and (ii) facts pointing to an evolutionary path from ‘short VS long’ to ‘high VS low’. Finally, the dating of pitch splitting of Cantonese Yin Ru tone is discussed, as well as the issue to what extent the similarity of the stopped tone between Cantonese and Tai languages is due to substratum, recent influence, or parallel evolution.
[Keywords] long and short vowel, stopped tones, tonal evolution, hyper-correction

Prior work has suggested that proto-Rma was a non-tonal language and that tonal varieties underwent tonogenesis (Liú 1998, Evans 2001a-b). This paper re-examines the different arguments for the tonogenesis hypothesis and puts forward subgroup-internal and subgroupexternal evidence for an alternative scenario in which tone, or its phonetic precursors, was present at the stage of proto-Rma. The subgroup-internal evidence comes from regular correspondences between tonal varieties. These data allow us to put forward a working hypothesis that proto-Rma had a two-way tonal contrast. Furthermore, existing accounts of how tonogenesis occurred in the tonal varieties are shown to be problematic. The subgroup-external evidence comes from regular tonal correspondences to two closely related tonal TransHimalayan subgroups: Prinmi, a modern language, and Tangut, a mediaeval language attested by written records from the 11th to 16th centuries. Regular correspondences among the tonal categories of ...

The Third Mesa dialect of Hopi as recently described has a simple tone (to be precise, pitch accent) system.  In this article we show that in almost cases (two morphemes remained unexplained) the low tone appears in those places where other and/or earlier forms of Hopi have and/or had a vowel followed by a glottal stop, h, preaspiration, or a voiceless sonorant.  A side note is that this immediately shows that B. L. Whorf's descriptions of Hopi were highly accurate at least as to the phonetics--contrary to what some people often say.  In fact, the one thing that we would now correct is that in a fn. in this paper, we still expressed some doubt about the three vowel lengths that Whorf recorded.  Soon after this was published it became clear to us that these must also have been real because they correlate with features of the proto-language that Whorf, unless he had owned a time machine, could not have known about, since they were discovered (by us) a half-century after his death.  In particular, the shortest of the three vowel lengths occurs consistently in the first syllable of *CVCVC stems, which would have had accent on th second syllable in the proto-language.

The development of phonological tone has been linked in many languages to consonant voice quality contrasts that impart pitch differences to preceding or following vowels. In particular, modal voicing on a syllable-initial consonant has been shown to correlate with low pitch on a following vowel in both tone and non-tone languages (Hombert et al. 1979). In the Austronesian language Malagasy, a strong relationship between consonant voicing and vowel pitch has been observed in the literature; previous studies state, however, that Malagasy does not have phonological tone. In contrast with these previous descriptions, the results presented in this paper from a production study of homorganic fricative pairs in several different dialects of Malagasy suggest that in certain dialects in the center of the country, in and around the capital city, pitch has replaced modal voicing as the primary phonetic cue to fricative voicing category. Based on word list data from 11 Malagasy speakers, this ...

In the 1970’s, linguist Franklin Huffman undertook extensive field work on the Katuic languages, an Austroasiatic sub-family. Thanks to the efforts of Paul Sidwell and Doug Cooper, WAV files of Huffman’s audio recordings, high quality scans of his notebooks and accompanying digitized databases of the data contained therein have been made available on sealang.net.  The Huffman Archives document register languages from most Katuic sub-groups, including Ta’oi (Ir & Bru Talan), Kriang (Ngeq), Bru (Bru Ubon, So Mangkong, So Sakol-Nakorn, So Khammouane & Katang) and Kuay (Souei Saravan, Kuy Surin & Kuy Srisaket). These languages have diverse register systems both in terms the phonetic features, largely qualitatively defined at this point, which uphold the phonological register contrast and also in terms of the registrogenetic processes which produced them. The diverse sample of Katuic register languages contained in the Huffman archives presents a unique opportunity to qualitatively asses the acoustic correlates of the register phenomenon across Katuic and to set a baseline which will inform further work on the production and perception of Katuic register. The purpose of this paper is to determine (a) if Huffman’s recordings are of sufficient quality for acoustic analysis, especially spectral slope analysis, and (b) if so, which acoustic cues most readily distinguish register contrast in various Katuic languages.

Here, we have begun with an analysis of Huffman’s Souei recordings (hereafter, Suay). Suay [kdt] is a particularly interesting test case as Huffman (1976) describes it as a Transitional register language, in which register is not yet phonologized but stereotypical lax register features (breathy voice, F1 lowering, etc…) are present. These register features have not yet become a feature of the language’s vocalism, but rather remain associated with the historical voiced stop onsets, which are in the middle of a devoicing process. Huffman (1979) further notes that the register contrast is ambiguous and possibly neutralized in the close vowels of Suay.

The Suay data contain 1,408 items spoken 2 times each in isolation by a single male speaker. Of these, we analyse 303 instances (151 unique lexical items) with long vowels or diphthongs which can be organized into two register categories. After hand-segmenting the TextGrids to demarcate the syllabic onset and rime, acoustic measures were taken using Praat. We measured Voice Onset Time (VOT) as well as spectral characteristics (f0, F1/F2/F3, and harmonic amplitudes H1-H4 and A1-A3) at 5 msec intervals. Harmonic amplitudes were then corrected for the effects of formant frequencies and bandwidths (Iseli & Alwan 2004). We focused primarily on the analysis of vowels following voiceless unaspirated stops, as this is the context where register features are present in Suay.

Our preliminary analysis of the Suay data shows that the two expected register categories are readily identified in terms of VOT, vowel quality (especially F1) and spectral tilt (H1*-H2*) over the first half of the syllable rime. As anticipated, these categorical differences are more apparent in non-close vowels. Similar results are obtained for other spectral tilt measures but importantly, pitch (f0) does not appear to play a role for this speaker.

Despite considerable background noise, our initial investigations demonstrate that spectral analysis of the Huffman archives is indeed feasible and that the results are useful both for aiding in interpreting Huffman’s transcriptions, as well as for reasoning about historical sound change. Segmentation and analysis of additional Kuay language data is currently underway.

References
Iseli, M., and A. Alwan. 2004. An improved correction formula for the estimation of harmonic magnitudes and its application to open quotient estimation. Proceedings of the 2004 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Montreal, 669-672.

Huffman, Franklin. 1976. The register problem in fifteen Mon-Khmer languages. In Philip N. Jenner, Laurence C. Thompson & Stanley Starosta (eds.), Austroasiatic Studies, 575-590. Honolulu: The University Press of Hawaii.

Huffman, Franklin. 1979. Analysis of Souei. In Paul Sidwell (ed.),

Huffman Papers. Online: www.sealang.net/archives/huffman (Accessed 27 January, 2019).

Punjabi, a language primarily spoken throughout Pakistan and in the northern Indian state of Punjab, is one of a few closely related Indo-Aryan languages, including Lahnda and Western Pahari, or Dogri-Kangri, which are counted among the world’s tone languages, despite having no genetic link to other recognized tone languages.  Few grammars have been published for Punjabi, and of those available, the grammars either fail to discuss the existence of lexical tone or note tone only in passing, and these disagree among themselves on even the number of tones.  Unfortunately, those grammars which do make note of the presence of lexical tone often fail to discuss the tone patterns or tonemics of Punjabi in a linguistically meaningful way or provide substantial evidentiary support for their own claims regarding tone pattern.  This may be due to the fact that, unlike Chinese, which has a contrastive pitch on each syllable, Punjabi “does not lean heavily on pitch phonemes” (Malik, 1995).  Still, they are widely evident in the spoken language and are in need of descriptive research supported by significant empirical data. It is the conclusion of this research that the high and low tones found in the Panjabi language can be directly correlated to the classic Gurmukhi orthography.  The script features historically aspirated and unaspirated variations of most consonants, which, in certain phonemic environments, are explicit indicators of the tonal qualities found in the spoken language.

Eastern Cham is a language of Vietnam. This language is fairly phonologically complex, especially with respect to pitch. Because of extensive contact with other languages, the structure and phonology have undergone some dramatic changes, including the ongoing process of tonogenesis. This paper focuses on the occurrence of pitch in this language, its distribution, and it argues that pitch is not yet phonemic in this language.

La présente contribution récapitule certaines avancées majeures réalisées dans l'étude des systèmes tonals d'Asie orientale, et discute leurs implications pour la typologie tonale et la modélisation phonologique. Les travaux phonologiques et typologiques accordent une place relativement limitée aux langues dans lesquelles les tons ne sont pas définis simplement en termes de hauteur, et ne se prêtent pas à une décomposition en niveaux tonals (par exemple le tamang et le vietnamien). Ces langues manquent assurément d'attrait au plan morpho-phonologique, étant dénuées d'alternances tonales. Elles n'en présentent pas moins une complexité bien réelle, qui doit être prise en considération dans la modélisation phonologique et les généralisations typologiques. Le panorama présenté ici ramène à une distinction proposée il y a plus d'un demi-siècle (Pike 1948) entre deux types d'organisation tonale : d'une part les systèmes de tons ponctuels*, d'autre part les systèmes de tons complexes*. Pour approfondir cette distinction synchronique, une attention particulière est accordée à l'évolution des systèmes tonals : les faits d'évolution sont en effet d'excellents révélateurs de la nature des systèmes phonologiques.
This contribution recapitulates major findings about East Asian languages to discuss general issues of tonal typology, diachronic evolution and phonological modeling. Relatively little attention has been paid to Far Eastern languages in which linguistic tone cannot simply be equated with pitch, or be decomposed into tonal levels, such as Tamang or Vietnamese. These languages are admittedly somewhat unattractive for phonologists, since they lack tonal alternations. However, they clearly need to be taken into account in phonological modeling and typological generalizations – and they present highly interesting complexities of their own. The conclusion which emerges in light of a synchronic and diachronic survey is very close to a point of view expressed more than half a century ago (Pike 1948), namely that there exists a major typological divide between two types of tonal organization: (i) level-tone systems and (ii) complex-tone systems. With a view to drawing all the implications of this divide, special attention is devoted to the evolution of tonal systems: diachronic facts provide precious insights into the nature of phonological systems.

Chrau, a Mon-Khmer language of southern Vietnam, has been described as preserving a voicing contrast in onset stops. Instead of voicing, closely related languages have a register contrast in which voiced and voiceless stops have respectively evolved into breathy and modal registers. Unexpectedly, the Chrau dialect investigated in this paper has a register system based on f0, voice quality and vowel quality, just like its sister languages. However, onset stop voicing is not fully neutralized: stops have a short positive VOT in the modal register, but in the breathy register, they either have closure voicing or a VOT slightly longer than modal register stops. There is no evidence for an acoustic trade-off between voicing and register cues. The preservation of VOT differences in onset stops suggests that Chrau has a conservative register system. We discuss how the cooccurrence of VOT and register cues sheds light on models of registrogenesis.

The paper deals with the particular subset of Proto-Indo-European phonemes called laryngeals. It is generally hypothesized that PIE had three or four such laryngeals, the nature of which has been studied mostly in respect with their segmental features. In this draft, Balto-Slavic data and tonogenesis, especially Štokavian, are analyzed in order to determine which tonogenetic features can be retrieved for the different laryngeals hypothesized in PIE.