Contrasting phonation types in languages can differ along several acoustic dimensions, depending on whether the contrast involves differences in open quotient, in glottal closing velocity, and in noise excitation/periodicity. Listeners thus potentially have multiple perceptual cues to such contrasts. Two perception experiments, classification and similarity rating, show that listeners from different language backgrounds attend to different acoustic correlates of modal and breathy voice vowels. Contrastive tones in tone languages may also vary in phonation quality.

To determine if the speaking fundamental frequency (F0) profiles of English and Mandarin differ, a variety of voice samples from male and female speakers were compared. Two methods of pitchtracking were also compared. Differences in the F0 data from the two methods were small, except that the lowest creaky F0 values were better detected manually. The F0 profiles of English vs. Mandarin speakers were sometimes found to differ, but these differences depended on the particular speech samples being compared. Most notably, the physiological F0 ranges of the speakers, determined from tone sweeps, did not differ between the two languages, indicating that the English and Mandarin speakers’ voices are comparable. Their use of voice pitch in single-word utterances was, however, quite different, with the Mandarin speakers having higher maximums and means, and larger ranges, even when only the Mandarin high falling tone was compared with English. In contrast, for a prose passage, the two languages were more similar, differing only in the mean F0, Mandarin again being higher. The study thus contributes to the growing literature showing that languages can differ in their F0 profile, but highlights the fact that the choice of speech materials to compare can be critical.

How are listeners able to identify whether the pitch of a brief isolated sample of an unknown voice is high or low in the overall pitch range of that speaker? Does the speaker’s voice quality convey crucial information about pitch level? Results and statistical models of two experiments that provide answers to these questions are presented. First, listeners rated the pitch levels of samples taken over the full pitch ranges of male and female speakers. The absolute f0 of the samples was by far the most important determinant of listeners’ ratings, but with some effect of the sex of the speaker. Acoustic measures of voice quality had only a very small effect on these ratings. This result suggests that listeners have expectations about f0s for average speakers of each sex, and judge voice samples against such expectations. Second, listeners judged speaker sex for the same speech samples. Again, absolute f0 was the most determinant of listeners’ judgments, but now voice quality measures also played a role. Thus it seems that pitch level judgments depend on voice quality only indirectly, through its information about sex. Absolute f0 is the most important information for deciding both pitch level and speaker sex.

Marginal phonemes and contrasts occupy a complex position in linguistic theory, as traditional theories of phonemehood do not account for marginality. However, contemporary linguistics has found that phonemic contrast strength is not fixed in childhood but rather continues to change, thus implicating the lexicon, the set of words a speaker knows, as a factor in the behavior of phonemic contrasts.This dissertation takes this link between category strength and the lexicon and treats it as an empirical question. I identify token frequency and type informativity, measures of frequency and predictability within a lexicon, as potential predictors of individual behavior. I then justify and present an experimental procedure for an eye tracking, two-alternative forced choice, categorization study on three phonetic continua — [a͡ɪ]-[ʌ͡i], a marginal contrast; [a͡ɪ]-[ɔ͡ɪ], a classic phonemic contrast; and [ʌ͡i]-[ɔ͡ɪ], a mixed case — in Canadian English, using the visual world paradigm. I discuss decisions that were made in the design of the experiment, including how individual lexicons were probed and why multiple continua were examined. Analyzing the resultant eye tracking data both graphically and by GAMM model comparison, I find that behavior was not interpretably predicted by my selected predictors, though their contribution to bias, a normalized preference measure, was statistically significant. I report on the behavioral patterns that were found in the categorization data and show that participants with differential behavior did not have statistically significant differences in either frequency or informativity in nearly all cases. My findings come as a surprise, as predicting that variation in the lexicon (operationalized as frequency and informativity) should influence linguistic behavior is both obvious and supported by the literature. I thus present my thoughts on why these predictors were not significant ones as well as my suspicion that the process of calculating these lexical statistics was poisoned by the likely incorrect assumption that a marginal phoneme can be treated as if it were a strong phoneme for the purposes of calculation. I close with suggestions for future work that could advance understanding of this issue, including potential test cases and the need for alternative operationalizations of frequency and predictability for marginal phonemes.

This dissertation investigates how glottal stops are produced and perceived, and why they occur so frequently before word-initial vowels in languages of the world. Specifically, the goal of the production chapter was to determine whether glottal stops are truly glottal sounds. High-speed imaging of glottal stops uttered by five phonetically-trained English speakers was obtained using trans-oral videoendoscopy. When produced as plosives, glottal stops always had some form of vocal fold incursion, though full glottal closure is not always observed. Moreover, glottal stops were usually - but not always - accompanied by ventricular incursion. Glottal stops are thus necessarily glottal but not necessarily ventricular sounds. The timing of ventricular incursion suggests that it may be used to sustain, rather than produce, glottal closure.

The specific goal of the perception study was to determine whether glottal stops are perceived distinctly from phrase-final creak, which shares similar acoustic features to glottalization. Sixteen English listeners were asked to identify words with glottal stops as allophones of /t/, e.g. 'button,' 'atlas,' and 'dent.' They were also presented with near-minimal pairs with no glottal stop, e.g. 'bun,' 'Alice,' and 'den.' The target words either had creak or no creak. The results indicate that words with glottal stop allophones are more accurately and more confidently identified when no creak is present, when the glottal stop is acoustically longer, and/or when the glottal stop is word-medial. Moreover, glottal stops are more accurately identified in creak when they occur in 'button'-type words, compared with worse identification in 'atlas'- and 'dent'-type words. For words with no glottal stops (e.g., 'bun,' 'Alice,' and 'den'), the presence of creak does not render them confusable with words with glottal allophones, except for 'den'-type words where word-final creak is sometimes mistaken for a glottal stop. Thus, glottal stops are generally harder to detect in creak, but are mutually confusable with creak only word-finally after nasals.

To determine why and when glottal stops occur word-initially, the occurrence of word-initial full glottal stops in an English corpus was analyzed using logistic mixed-effects regression modeling. Prominence and phrasing are overwhelmingly the most important factors in predicting full glottal stop occurrence in English. Moreover, prominent word-initial vowels that are not preceded by full glottal stops 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 were subsequently confirmed 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 only for the relative strength of the glottal stop gesture. Glottalization before word-initial vowels is thus a marker of prominence and is used amplify cues to prominence when they would otherwise be weakened.

San Felipe Jalapa de Díaz (Jalapa) Mazatec is unusual in possessing a three-way phonation contrast and three-way level tone contrast independent of phonation. This study investigates the acoustics of how phonation and tone interact in this language, and how such interactions are maintained across variables like speaker sex, vowel timecourse, and presence of aspiration in the onset. Using a large number of words from the recordings of Mazatec made by Paul Kirk and Peter Ladefoged in the 1980s and 1990s, the results of our acoustic and statistical analysis support the claim that spectral measures like H1-H2 and mid-range spectral measures like H1-A2 best distinguish each phonation type, though other measures like Cespstral Peak Prominence are important as well. This is true regardless of tone and speaker sex. The phonation contrasts are strongest in the first third of the vowel and then weaken towards the end. Although tones remain distinct from one another in terms of F0 throughout the vowel, for laryngealized phonation the tone contrast in F0 is partially lost in the initial third. This study shows that the complex orthogonal three-way phonation and tone contrasts do remain acoustically distinct according to the measures studied, despite partial neutralizations in any given measure.