Papers by Yueh-chin Chang
This paper is an acoustic and articulatory study of word-level prominence in Taiwanese Mandarin. ... more This paper is an acoustic and articulatory study of word-level prominence in Taiwanese Mandarin. Data from three male speakers were collected and analyzed, using co-registered EMA and ultrasound. Recording materials included 9 disyllables comprised of repetitive syllables (as nicknames, e.g., rá nrá n) and 9 disyllabic words with identical/similar syllable structures. Results show that there are significant kinematic differences in tongue movement between the initial and final repetitive syllables, although no difference can be found in acoustic properties. Taken together, it is evident that the trochaic foot (Strong-weak) may still be reflected in articulation, even though word-level prominence (or, word stress) is not perceptually salient in Taiwanese Mandarin.
This study explores articulatory differences of Er-suffixation (Erhua; diminutive suffixation) be... more This study explores articulatory differences of Er-suffixation (Erhua; diminutive suffixation) between Beijing Mandarin (BJM) (see [1], among many others) and Northeastern Mandarin (NEM). With the help of an NDI Wave, we compared temporal changes of the tongue configuration at different duration deciles of Er-suffixed monophthongs (represented with ten solid lines for each stem: t1-t10; the tongue postures of corresponding unsuffixed monophthongs are represented with a green dashed line throughout). Six speakers (BJM=3; NEM=3) participated in the experiments and 10 tokens (5 Er-suffixed and 5 unsuffixed) were analyzed for each stem. Only representative results from one BJM speaker (BJ-01) and one NEM speaker (NE-01) are provided here since the results are highly consistent within each dialect group. High front vowels /i, y, ɿ, ʅ/: We can see from Figs. 1-2 that although tongue root retraction is attested in both BJM and NEM speakers when a high front vowel is being Er-suffixed, one significant cross-dialectal difference emerges, namely that BJM speakers employ a "curling-up" posture (negative Tongue Tip Angle (TTA); see [2]) to produce the Er-suffix, whereas the curlingup posture is completely absent in NEM. Instead, NEM speakers opt for a downward tongue tip (positive TTA) in Er-suffixation. Taken together, it is evident that Er-suffixation involves tongue root retraction in both BJM and NEM, while BJM features an "additional" gesture of tongue tip curling. This cross-dialectal difference also leads to that the temporal lingual changes (solid lines) are of much greater magnitude in BJM Er-suffixation. Interestingly enough, regarding the "retroflex" apical vowel /ʅ/, no obvious change is found, especially in NEM Er-suffixation ([3]). Back vowels /ɑ, u/: We can see in Figs. 3-4 that BJM also shows a significant tongue shape change in Er-suffixation, whereas NEM's tongue postures are kept "stable" throughout.
This paper investigates the articulatory characteristics of Er-(diminutive) suffixation in Northe... more This paper investigates the articulatory characteristics of Er-(diminutive) suffixation in Northeastern Mandarin using an NDI Wave. We compared temporal changes of the tongue configurations at different deciles of the duration of both unsuffixed and Er-suffixed monophthongal stems. In addition, the vowel phoneme /ɚ/ (rhotic schwa) is included for comparison. Results show that, details aside, Ersuffixation predominantly involves a retraction of the tongue body (i.e., a bunched tongue configuration exhibiting a positive value of the Tongue Tip Angle (TTA) index across the board), while a raising of Tongue Tip (TT) is found in the articulation of the rhotic schwa /ɚ/. Diachronically, the Er-suffix is derived via adding the rhotic schwa to a stem. So, the well-known retroflex vs. bunched distinction of /ɹ/ in American English is replicated in Northeastern Mandarin, albeit in a different fashion.
In this paper, we examined different methods of modeling prosodic features of tones, and their ef... more In this paper, we examined different methods of modeling prosodic features of tones, and their effects on a speaker-independent Taiwanese tone recognition system. Tones can be modeled either by plain or curve-fitted features. Plain features represent the original curve faithfully using pitch values, while curve-fitted features can be thought of as an approximation to the values using mathematical functions, such as a Legendre polynomial. In addition, durational information of tones was also proven effective in previous researches. Thus, we proposed a new approach of modeling Taiwanese tones using curve-fitted features extracted from fractions of the pitch curve, along with duration as an additional prosodic feature. Our experimental results showed that using these features in an SVM classifier could substantially improve the accuracy of tone recognition in Taiwanese. Besides, we provided an empirical perspective for theoretic studies on tonal neutralization.
Interspeech 2021, 2021
This paper is a comparative study of the articulation of the "apical vowels" in three Mandarin di... more This paper is a comparative study of the articulation of the "apical vowels" in three Mandarin dialects: Beijing Mandarin (BJM), Northeastern Mandarin (NEM), and Southwestern Mandarin (SWM), using co-registered EMA and ultrasound. Data from 5 BJM speakers, 5 NEM speakers and 4 SWM speakers in their twenties were analyzed and discussed. Our recording materials include the dental and retroflex apical vowels, and their er-suffixed forms. Results suggest that distinct lingual configurations are found among the three dialects of Mandarin, even though these apical vowels are not perceptually distinguishable. Specifically, the dental apical vowel [ɿ] has a grooved tongue shape in BJM, a retracted tongue dorsum in NEM, and a relatively flat tongue shape in SWM. The retroflex apical vowel [ʅ] has a domed tongue shape as well as a bunched tongue body in NEM, while a slightly domed tongue posture is found in SWM. Moreover, the retroflex apical vowel [ʅ] is, articulatorily speaking, very similar to the er-suffix in BJM (cf. [10]). In sum, we observed yet another instance of the articulatory-acoustic mismatch.
Journal of the Acoustical Society of America, 2016
This study examines the temporal organization of prenuclear glide (onglide) gestures within a syl... more This study examines the temporal organization of prenuclear glide (onglide) gestures within a syllable. Kinematic data were collected from four Standard Chinese (SC) and four Taiwanese Mandarin (TM) speakers using an NDI Wave. Mean relative timing of C/G gestures to a (heterosyllabic) consonantal anchor /p/ were calculated using the lag between achievement of constriction of the C/G and /p/. By comparing a given subject’s mean lags for Cs/Gs with CGs, we calculated the subject’s leftward (C vs. CG) and rightward (G vs. CG) shifts associated with /pj/ and /kw/. Results show that in SC, onglides /j/ and /w/ don’t pattern alike: for /kw/, all SC speakers show both leftward and rightward shifts (i.e. the C-center effect), except for subject S3, who fails to show a leftward shift (due to a different regional accent). For /pj/, only rightward shifts are found for all SC speakers. In contrast, regarding /kw/, only leftward shifts are found for all TM speakers, except for subject S8 (due to...
Cahiers De Linguistique Asie Orientale, 1985
Chang Yueh-chin. Chang Yueh-chin : Contribution à la recherche tonale sur un des dialectes Min-na... more Chang Yueh-chin. Chang Yueh-chin : Contribution à la recherche tonale sur un des dialectes Min-nan parlé à Taiwan.. In: Cahiers de linguistique - Asie orientale, vol. 14 2, 1985. pp. 294-295
Journal of the Acoustical Society of America, 2015
Coronal consonants hold a special status for their crowded space for articulatory contrasts. Mand... more Coronal consonants hold a special status for their crowded space for articulatory contrasts. Mandarin and its variants are known to have rich inventories of coronal consonants, where a three-way coronal place contrast is usually maintained. Nonetheless, previous studies have reported a (partial) neutralization of coronal places of articulation in Taiwan Mandarin, while other acoustic studies contend that the three-way contrast remains. The aim of this study is to evaluate the purported coronal neutralization from an articulatory perspective. Five native speakers of Taiwan Mandarin participated in a set of electromagnetic articulographic (EMA), palatographic and linguagraphic experiments in order to investigate the articulatory targets of coronal consonants in Taiwan Mandarin. Our articulatory results are in general consistent with the acoustic results in previous studies, in that the three-way coronal place contrast does exist in Taiwan Mandarin, and the so-called “retroflexes” in T...
Journal of the Acoustical Society of America, 2015
It has been noted that vowel backness is largely preserved in English-to-Mandarin loanwords, but ... more It has been noted that vowel backness is largely preserved in English-to-Mandarin loanwords, but not vowel height. This asymmetry contradicts the native phonological patterns. The aim of this study is to re-examine this issue from an experimental perspective. Ten Taiwanese Mandarin (TM) speakers (aged 23–30) participated in this experiment. English Stimuli consisted in /CVmi/ sequences with C corresponding to one of the three consonants {b, d, g} and with V to one of the four vowels {ɛ, ɔ, æ, ɚ}. Mandarin stimuli also consisted in /CVmi/ sequences with C corresponding to one of the three consonants {p, t, k} and with V to one of the following vowels {i, a, u, ei, ie, au, ou, uo}. The participants were asked to rate similarity between American English and TM vowels (on scale 1–7). Our results partially support the generalizations in loanword adaptation, namely that TM speakers tend to map English [æ] onto Mandarin [ei], rather than [a], while [ɔ] is mapped to [ou/uo], [ɛ] to [ei], an...
This study shows that Southern Min chainshift tone sandhi rules have been reduced to a binary H v... more This study shows that Southern Min chainshift tone sandhi rules have been reduced to a binary H vs. L contrast among young generation speakers of Penang Hokkien. More remarkably, both complete and incomplete neutralization of sandhi tones are attested in this language. Some possible explanations will be discussed, including preservation of phonological identity and/or the (non-)completion of sound change as well as its consequences for redistribution of exemplars.
The Linguistic Review, 2012
Contextual tonal variations in Malaysian Hokkien (Southern Min) display three features of unusual... more Contextual tonal variations in Malaysian Hokkien (Southern Min) display three features of unusual typological interest. First, our acoustic study shows that carryover and anticipatory tonal influences do not substantially differ in the magnitude, suggesting that tonal coarticulation may not be universally asymmetric, or, the carryover bias is not as robust as previously assumed. Second, carryover coarticulation is not predominantly assimilatory and anticipatory coarticulation is not dissimilatory in nature. In particular, it is attested for the first time that Progressive Dissimilation occurs in ditonal contexts. Third, another important finding is that near tone mergers are subject to distinct c oarticulatory effects and exert different contextual influences on adjacent tones, suggesting that contrast preservation may be, to some extent, evoked even in phonetic realization. In conclusion, not only does this study enrich the typology of tonal coarticulation patterns, but it also argues for a new perspective on tonal coarticulation, whereby, aside from functionally motivated f actors such as articulatory constraints and anti-merger, the "anomalous" patterns in Malaysian Hokkien may well be due to Final Prominence in Southern Min tone sandhi systems, i.e. priority for fully faithful rendition of the underlying tones in sandhi-final position.
Computer Speech & Language, 1989
This paper presents a method for the recognition of lexical tones in Mandarin speech based on vec... more This paper presents a method for the recognition of lexical tones in Mandarin speech based on vector quantization and hidden Markov models. A Kay Visi-Pitch 6087DS is used to extract the fundamental frequency (F0) contour. The features for the recognition of lexical tones are derived from an F0 contour within the voiced part of a syllable. Markov models are generated for the four monosyllabic tones and for 15 disyllabic tone pairs based on these derived features. A Viterbi algorithm is applied to find the probability scores of a test tone scored by the models during the recognition phase. In speaker-independent tone recognition experiments, the average recognition rate was 97·9% for isolated monosyllabic words, 92·9% for disyllabic words, and 91·0% for trisyllabic words. The decline comes from the tone variation and the coarticulation between syllables.
Cahiers de linguistique - Asie orientale, 1998
... par ma (pour Chai et Yu), soit plus élevé que les questions déclaratives (pour Yu), et encomp... more ... par ma (pour Chai et Yu), soit plus élevé que les questions déclaratives (pour Yu), et encomparaison avec les ... Contrairement au français, dont l'intonation est de même importance que la syntaxe et la morphologie dans la formation de l ... The Prosody of Mandarin Chinese. ...
This paper discusses the issues and requirements associated with linguistic field research using ... more This paper discusses the issues and requirements associated with linguistic field research using transnasal laryngoscopy. New developments in the portability and robustness of equipment may allow laryngoscopic experiments to be carried out at or near ...
mx.nthu.edu.tw, 2011
This paper aims at systematically investigating the aerodynamics of nasalization in Taiwanese, a ... more This paper aims at systematically investigating the aerodynamics of nasalization in Taiwanese, a language that has a nasality contrast in its vowels but are subject to stricter restrictions on nasality distribution than French. Our results show that i) the onset ...
Speech errors in Taiwanese are investigated by means of a speeded repetition task. Our results sh... more Speech errors in Taiwanese are investigated by means of a speeded repetition task. Our results show that the intrusion bias is also attested for word pairs with mismatched onsets, whereas in the alternating coda condition, reduction errors are the most frequent error type. This cross-linguistic difference can be attributed to language-specific implementation of stop codas. Keywords: speech errors, Taiwanese, gestures, stop codas, checked syllables
Controversy exists whether the syllable-final voiceless stops of (S) E Asian languages are reinfo... more Controversy exists whether the syllable-final voiceless stops of (S) E Asian languages are reinforced with glottalization in all these languages, ie- p- t- k. For example, early laryngoscopic observation of finals in Cantonese, Taiwanese (Fukienese), and Thai found that after vocalization ceases, vocal folds and false vocal folds adduct, and, in some cases, there is engagement of the aryepiglottic sphincter. This paper confirms that Taiwanese and Hakka have reinforcement of finals with glottal stop and also the variable ...
This paper aims to re-examine the issues of psychological reality of the Taiwanese Tone Circle wi... more This paper aims to re-examine the issues of psychological reality of the Taiwanese Tone Circle with a novel experimental setting. Previous results from various wug tests all suggested that the tone sandhi rules should be unproductive, due to the obtained ...
Proceedings of the 17th International Conference of Phonetic Sciences (ICPhS XVII), 2011
This study describes and examines the acoustic properties of the central vowels [ɨ] and [] in Mal... more This study describes and examines the acoustic properties of the central vowels [ɨ] and [] in Malaysian Hokkien (MH). The two central vowels are typologically special in that both of them are “full-fledged”, phonemic vowels in an inventory. Therefore our main research question is to see if the two central vowels are subject to contextual influences, just like their counterparts in other languages (eg [2, 6, 8, 9]). Our principal finding is that no significant contextual variability is attested, indicating that central vowels may be resistant ...
The Journal of the Acoustical Society of America, 2014
It has been anecdotally observed that the three-way contrasts in Standard Chinese are reduced to ... more It has been anecdotally observed that the three-way contrasts in Standard Chinese are reduced to two-way contrasts in Penang Mandarin (PM). PM is a variety of Mandarin Chinese spoken in Penang of Malaysia, which is influenced by Penang Hokkien. This work shows that the alleged neutralization of contrasts is incomplete (10 consonants x 3 vowel contexts x 5 speakers). More specifically, alveopalatal [ɕ] may range from postalveolar zone (73.33%) to alveolar zone (26.67%), and so does retroflex [ʂ] (46.67% vs. 46.67%). [s] and [n] are apical (or [ + anterior]) coronals. The goal of this study is three-fold: (i) to describe the places of articulation of PM coronals and the patterns of ongoing sound changes, (ii) to show the neutralization of place contrasts is incomplete whereby constriction length remains distinct for these sibilant sounds, and (iii) to demonstrate different coarticulatory patterns of consonants in variant vowel contexts. The intricate division of coronal consonants doe...
Uploads
Papers by Yueh-chin Chang