Other projects
Voicing of glottal consonants and non-modal vowels across the world's languages (with Marc Garellek, Yuan Chai, and Maxine Van Doren)
This joint work is published online:
Garellek, M., Chai, Y. , Huang, Y. , & Van Doren, M. Voicing of glottal consonants and non-modal vowels. Journal of the International Phonetic Association. (2021). https://doi.org/10.1017/S0025100321000116
This collaborative work investigates voicing among laryngeal sounds of the world’s languages. We analyzed audio recordings from Illustrations of the International Phonetic Alphabet to explore the variation in voicing across glottal consonants and non-modal (breathy and creaky) vowels in three phrasal positions. We found that the systematic variation in voicing is due to prosody; for example, /h/ is voiced except in utterance-initial positions. We interpret these results through an articulatory lens: glottal consonants and non-modal vowels are both modulations in phonation resulting from laryngeal constriction and vocal fold spreading. We argue further that, because voicing during [ʔ] and [h] is largely predictable from respiratory and prosodic constraints, many cases of [ʔ] and [h] can be considered to be phonetically underspecified for voicing.
Vowel harmony
These projects are deposited in the San Diego Linguistics Paper, hosted by UCSD Linguistics.
Rere: click for Paper
This paper presents a phonological as well as acoustic analysis of the elicited data from one male native speaker of Rere, Taitas Kanda, who was born and raised in the town of Kwandaŋ in the Abri area in Sudan. We elicited weekly sessions with Taitas from January to May 2019, and annotated the data using ELAN (all the examples can be accessed in Kwaras Rere corpus at rere.ucsd.edu). For the acoustic analysis, I segmented words and phonemes in Praat (Boersma & Weenink, 2019), and processed the acoustic measures in VoiceSauce (Shue et al., 2011). I investigate vowel harmony and raising alternations, which appear in certain grammatical processes including instrumental suffixation and valence-changing constructions. The results show that vowel harmony is not always categorical; there are disharmonic words and many exceptions to the harmonic processes. This implies a more complex relationship between phonological systems and phonetic realizations.
Tibetan: click for Paper
This project studies vowel height harmony in Tibetan using linguistic analysis on archived data under computational phonotactic models. I discuss failures of idempotency in Tibetan and approaches for modeling this failure in OT. Vowel harmony in Lhasa Tibetan exhibits an interesting pattern that nonhigh vowels raise to assimilate the height of high vowels in disyllabic words bidirectionally (Ultan, 1973; Chang & Shefts, 1964; Dawson, 1983). Two exceptions occur: in progressive harmony, a geminate low vowel [a:] in the second syllable blocks height assimilation; in regressive harmony, a schwa derived from an unstressed /a/ word-finally does not trigger assimilation. These two blocking environments both speak to the need of failing idempotency in Optimality Theory (Magri, 2017) by treating the [a:] and the derived schwa differently from the rest of the input/output inventory for vowel harmony. This paper thus presents a comprehensive overview of these approaches and a discussion of their different shortcomings.
Predicting the Spoken-Written Lexical Frequency Asymmetry (with Sin Hang Lau, Ed Vul, and Victor Ferreira)
This joint work with Sin Hang, Ed, and Vic, is published online:
Lau, S.H.*, Huang, Y.*, Ferreira, V.S., Vul, Edward. Atten Percept Psychophys. (2019). https://doi.org/10.3758/s13414-019-01682-y (* co-first authors)
In this study, we aim to investigate what predicts the spoken-written asymmetry by examining modality specific measures (orthographic and phonological) in three families of predictors: length, neighborhood density, and bigram positional probabilities. The first two predictors are commonly controlled for in psycholinguistics studies, while the latter is more prominent in natural language processing. Even though the last measure seems to be much more computationally demanding and statistically-driven, it has been shown that even infants are sensitive to these complicated probabilistic patterns (Jusczyk & Luce, 1994). Additionally, phonotactic probability facilitates word learning (Gupta & Tisdale, 2009) and affects processing speed even when the phonological complexity is controlled for (Goldrick & Larson, 2008).
The Phonological Word in Mandarin - Evidence from Tone 3 Sandhi in Reduplication
Presented at IACL-26th @UW-Madison (2018), SCAMP 2018 @USC (click for Slides)
For a detailed version, click for the unpublished paper.
What consists of a phonological word in Mandarin? I use evidence from tone sandhi to define the prosodic word in Mandarin as one full-tone syllable + neutral-tone syllables specified by morpho-phonological rules.
The Effect of Focus on Creaky Phonation in Mandarin Tones (with Irene Vogel, and Angeliki Athanasopoulou)
Presented at PLC-41 @UPenn (2017) (click for Slides), PWPL 24.1 (click for Paper)
Huang, Yaqian; Athanasopoulou, Angeliki; and Vogel, Irene (2018) "The Effect of Focus on Creaky Phonation in Mandarin Chinese Tones," University of Pennsylvania Working Papers in Linguistics: Vol. 24: Iss.1, Article 12. Available at: https://repository.upenn.edu/pwpl/vol24/iss1/12
IACL-26th @UW-Madison (2018) (click for Slides)
We investigated the presence of creak under focus with well-controlled stimuli in a production study. We found that creaky voice increased in the high-falling Tone 4 in Mandarin, likely due to its expanded pitch range.
Neural Underpinnings of Identification Performance following Nonnative Phonetic Training (PI: F. Sayako Earle)
This is an archived project on nonnative phonetic training at U of Delaware.
Identification and discrimination are two tasks that are commonly used to assess speech perception, and have traditionally been considered two means to assess the same underlying representation. The aim of this study is to determine the neural underpinnings of this behavioral dissociation between identification and discrimination, following nonnative phonetic training and overnight consolidation.
Filler-Gap Dependencies in Mandarin Parasitic Gaps (with Shota Momma, and Qi Cheng)
The fundamental question can be traced back to, are parsers risk-takers or conservatives?
Mandarin parasitic gap is a case in point where the legitimacy of the first gap in the adjunct phrase crucially depends on the second gap in the matrix clause. We know that in filler-gap dependencies designs, parsers are always willing to posit a gap after a verb in the sentence to be able to put the filler encountered earlier in the gap. The past findings were that parsers are willing to take risks during online sentence processing.
We ask, if parsers actively predict and presuppose a parasitic gap structure (e.g., augmented with priming), will they posit gaps and expect to encounter a transitive verb even before the verb appears in the sentence or they will rather wait to see if the verb is transitive or not to be able to posit a gap for the filler? aka Hyperactive gap-filling vs. active gap-filling (Omaki et al. 2015).