Department of Linguistics

Despite abundant research on the distribution of word-initial glottal stops, it is still unclear which factors matter most in predicting where glottal stops occur and why. In this study, logistic mixed-effects regression modeling is used to predict the occurrence of word-initial full glottal stops in an English corpus. The results indicate that prominence and phrasing are overwhelmingly the most important factors in predicting full glottal stop occurrence. Moreover, prominent word-initial vowels that are not preceded by a glottal stop show acoustic correlates of glottal constriction, whereas non-prominent phrase-initial vowels do not. Rather, phrase-initial voicing (even for sonorants) is less regular, but in a manner inconsistent with glottal constriction. These findings are subsequently con_rmed using articulatory measures from electroglottography, and extended to Spanish. Based on the results, a prominence-driven theory of word-initial glottalization is proposed and motivated, with higher phrasal domains responsible for the strength of the glottal stop gesture.

Since Cutler and Norris (1986), it has been held that the role of the syllable in on-line word segmentation is fundamentally language specific; syllabification-based strategies are said to be available in syllable-timed languages, but unavailable in languages that are stressed-timed. The present study used word-spotting and found listeners in English (a prototypically stress-timed language) to be highly sensitive to sonority patterns. In particular, it was found that listeners more readily parsed sonorant consonants as codas, making vowel-initial target words like “absent” easier to spot in nonsense strings like “jeemabsent” compared to “jeebabsent”. This pattern mirrors both English-speaking listeners’ off-line syllabification preferences, and also the on-line behavior of listeners of syllable-timed languages (e.g., French), suggesting the syllable-based segmentation routine is not language specific.

In many dialects of Japanese, high vowels between voiceless consonants are often devoiced. This devoicing phenomenon is generally considered a phonological assimilation process. It is almost obligatory in theTokyodialect, except for some marked environments in which complete devoicing is often blocked. One such case is so called consecutive devoicing, where two or more consecutive vowels are in devoiceable environments. Although several accounts of consecutive devoicing have been proposed (e.g., Kondo, 2005; Yoshida, 2004; Tsuchida, 1997), the nature of its markedness, namely whether it is phonologically or phonetically driven, is still being debated. In order to provide a more comprehensive account of consecutive devoicing in theTokyodialect, the current study investigated the relative contribution of various factors on its occurrence as well as its across-speaker variability. The results revealed that the manner of consonants surrounding the second devoiceable vowel (C2-C3) had the strongest effect on the likelihood of consecutive devoicing among the factors tested, followed by the quality of vowel in the following mora (V3). Further, large and gradient speaker variability in the occurrence of consecutive devoicing was observed. These results indicate that consecutive devoicing is a phonetically driven process, rather than a phonological process as traditionally considered.

Recent research has sought to identify the systematic features that make Singa-porean English (SgE) distinct from other varieties of English. Although the intonation of SgE has been described previously (Deterding 1994; Lim 2004; Ng 2011), no phono-logical model has yet been proposed. This paper proposes a model of SgE intonational phonology within the Autosegmental-Metrical phonology framework (eg. Pierrehum-bert 1980). Three native speakers were recorded reading declarative and question sentences of varying length and stress pattern. Preliminary results suggest that SgE has three prosodic units above the word: the Accentual Phrase (AP), Intermediate Phrase (ip) and Intonational Phrase (IP). An AP is slightly larger than a word and is characterized by a general LH (rising) contour. The L can be attributable to either an L* tone on a lexically-stressed syllable or an L initial boundary tone if the stressed syllable occurs late in the AP. The AP-_nal syllable always has a phonologically high boundary tone (Ha). Intermediate phrases are marked by L- and H- tones and IPs are marked by L% and H% tones. Finally, preliminary data suggests that SgE speakers use an expanded pitch range and occasionally lengthening to mark contrastive focus.

This paper presents a model of the intonational system of Tongan, an Austronesian language, taking the autosegmentalmetrical theory as its framework. Tongan has lexical stress which appears on the penultimate syllable of prosodic words and is marked postlexically with one of two bitonal pitch accents—a rise, LH*, or a low tone, L*. Measurements show that the first tone of both pitch accents aligns with the stressed syllable onset, while the second tone aligns with the stressed syllable offset. There is evidence for two tonally marked levels of prosodic phrasing in Tongan, the intonational phrase (IP) and the accentual phrase (AP). The IP is about the size of a full utterance or major phrase and is marked by a final boundary tone and are realized on the IPfinal syllable. Four boundary tones have been observed. The smaller unit, the AP, usually contains one lexical word plus preceding functional elements. Two APfinal tones have been observed, realized on the final syllable of the phrase. Lastly, focus is only realized intonationally through increased pitch range on the focused element. Tongan is typologically interesting because it provides another case in a growing list of languages that intonationally marks both head and edge prominence.

The present study explored listeners’ expectations about how prosodic prominence can be used to disambiguate information structure in English. In particular, the contribution of prenuclear accents to the prosodic disambiguation of the size of the focus constituent (broad VP vs. narrow object focus) in SVO constructions was tested using the cross-modal priming paradigm. In two experiments, listeners were presented with visual targets (e.g., “brunette”) following contrastively related primes (e.g., “blonde”), which were heard as objects in SVO sentences (e.g., “He kissed a blonde.”). In Experiment 1, listeners heard the sentences produced with a single pitch accent on the object, and the focus structure varied from broad VP focus to narrow object focus. No significant differences in priming patterns across conditions were found, supporting theories of Focus Projection (e.g., Selkirk 1995, Gussenhoven 1984), which predict prenuclear accents to be optional. In Experiment 2, the information structure of the sentences was held constant as narrow object focus, and their prosody varied with respect to the presence of a prenuclear pitch accent on the verb. For these narrow focus sentences, it was found that priming occurred only when the sentence lacked a prenuclear accent, suggesting that prenuclear pitch accents contribute meaningfully to the information structural contrast. Sensitivity to the prosodic manipulation, however, was found to be modulated by individual differences in listeners’ “autistic” traits. The implications for on-line lexical processing and theories of themapping between prosody and information structureare discussed.

This study explores the properties of one type of phonation contrast - the tense vs. lax phonation contrasts of Yi (Loloish) languages – in terms of their glottal articulations, acoustic correlates, and perceptual salience. To determine the glottal articulations involved in the phonation contrasts, we adopted Functional Data Analysis to analyze entire EGG glottal pulse shapes. This method is found to capture differences in the abruptness of contact, a key property that is not captured by traditional EGG parameter measures. The primary contact patterns for phonation contrasts are very consistent across languages, speakers, and genders, and there is only one underlying articulatory pattern for these tense vs. lax contrasts. Overall, we found consistent correlations among the different kinds of measures, which means that speakers and listeners are able to establish a stable link between articulation, acoustic signals and perception.

In this study, we determine the acoustic correlates of primary and secondary stress in Tongan. Vowels with primary stress show differences in F0, intensity, duration, F1, and voice quality, but F0 is the best predictor of primary stress. Vowels with secondary stress are mainly cued by a difference in F0. With regards to the effects of stress on the vowel space, we find that all five Tongan vowels are higher in the vowel space (have lower F1) when unstressed, with no differences in F2. Moreover, there is no reduction in the overall size of the vowel space. We interpret this pattern as evidence that unstressed vowels in Tongan are not undergoing centralization, nor are they otherwise reduced. Rather, Tongan speakers have separate targets for stressed and unstressed vowels.

This study revisits the issue of tonal registers by exploring the cues used in producing andperceiving the five level tones of Black Miao. Both production and perception experiments show that non-modal phonations are very important cues for tonal contrasts. Two different kinds of non-modal phonations that either enhance pitch contrasts or provide an additional contrastive cue divide tonal levels into several registers. Benefiting from more than one cue, 11, 33 and 55 are well distinguished in the tonal space; by contrast, 22 and 44, only contrasting in pitch, are the most confusable tones. The tonal registers model can explain the different uses of non-modal phonations across languages.

This study investigates the importance of source spectrum slopes in the perception of phonation by White Hmong listeners. In White Hmong, non-modal phonation (breathy or creaky voice) accompanies certain lexical tones, but its importance in tonal contrasts is unclear. In this study, native listeners participated in two perception tasks, in which they were asked to identify the word they heard. In the _rst task, participants heard natural stimuli with manipulated F0 and duration (phonation unchanged). Results indicate that phonation is important in identifying the breathy tone, but not the creaky tone. Thus, breathiness can be viewed as contrastive in White Hmong. Next, to understand which parts of the source spectrum listeners use to perceive contrastive breathy phonation, source spectrum slopes were manipulated in the second task to create stimuli ranging from modal to breathy sounding, with F0 held constant. Results indicate that changes in H1-H2 (di_erence in amplitude between the _rst and second harmonics) and H2-H4 (di_erence in amplitude between the second and fourth harmonics) are independently important for distinguishing breathy from modal phonation, consistent with the view that the percept of breathiness is inuenced by a steep drop in harmonic energy in the lower frequencies.