The metaphysics of coarticulation

Hammarberg, Robert

doi:10.1016/s0095-4470(19)31261-6

Cited by 52 publications

(18 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One reason why Gibson's (1979) general theory of perception has seemed unlikely to serve as a useful basis on which to develop a theory of speech perception (see Fowler, 1986 and the following commentary) is that the phonological or phonetic primitives that a listener is presumed to recover-phonetic features, for example-are generally not considered to cause, in an unmediated way, the structuring of the acoustic speech signal. That is, features are believed to be in the mind of the talker, and not to be transparently reflected in the vocal-tract actions that do causally structure the signal (e.g., Hammarberg, 1976;Pierrehumbert, 1990;Repp, 1981). Both in the field of linguistics, however (e.g., Browman & Goldstein, 1990), and in psychology (Fowler, Rubin, Remez, & Turvey, 1980;Saltzman & Kelso, 1987), theories are under development in which phonological primitives are presumed to be gestures of the vocal tract.…”

Section: Inv Arlants Specifiers Cues 599mentioning

confidence: 99%

Invariants, specifiers, cues: An investigation of locus equations as information for place of articulation

Fowler

1994

Perception & Psychophysics

109

View full text Add to dashboard Cite

This experiment explored the information for place of articulation provided by locus equations--equations for a line relating the second formant (F2) of a vowel at midpoint of F2 of the formant at consonant-vowel (CV) syllable onset. Locus equations cue place indirectly by quantifying directly the degree of coarticulatory overlap (coarticulation resistance) between consonant and vowel. Coarticulation resistance is correlated with place. The experiment tested predictions that when coarticulation resistance varies due to properties of the consonant other than place of articulation (in particular, due to manner of articulation), locus equations would not accurately reflect consonantal place of articulation. These predictions were confirmed. In addition, discriminant analyses, using locus equation variables as classifiers, were generally unsuccessful in classifying a set of consonants representing six different places of articulation. I conclude that locus equations are unlikely to provide useful place information to listeners.

show abstract

Section: Inv Arlants Specifiers Cues 599mentioning

confidence: 99%

Invariants, specifiers, cues: An investigation of locus equations as information for place of articulation

Fowler

1994

Perception & Psychophysics

109

View full text Add to dashboard Cite

show abstract

“…He suggested that the undershoot observed with lower stress levels and faster articulatory rates is a consequence of the earlier arrival of the final consonant control signal, which deflects the articulator away from its target trajectory. (See Daniloff &Hammarberg, 1973, andHammarberg, 1976, for a some-what more general assimilatory account of coarticulation. )…”

Section: Perception Of Coarticulated Segmentsmentioning

confidence: 99%

Vowels as islands of reliability

Diehl¹,

Kluender²,

Foss³

et al. 1987

Journal of Memory and Language

View full text Add to dashboard Cite

Vowel nuclei of syllables appear to provide a relatively stable (although not stationary) frame of reference for judging consonant events. We offer evidence that reliable consonant identification demands prior or simultaneous evaluation of this "vocalic frame." Listeners were presented a list of /bVs/, /dVs/, and /gVs/ syllables and were instructed to press a response key immediately upon recognizing a particular initial consonant target. Three groups of subjects monitored for /b/, /d/, and /g/, respectively. The test syllables contained 10 English vowels varying substantially in intrinsic duration. Response times to the initial consonants correlated positively with the duration of the following vowels, even when the effect of consonant-vowel formant transition duration was partialed out. The results suggest that consonant recognition is vowel dependent and, specifically, that a certain amount or proportion of the vowel formant trajectory must be evaluated before consonants can be reliably identified.

show abstract

“…In connected speech, the production of a speech sound is affected by the articulatory features of its neighboring sounds [1][2][3][4][5]; this typically results in a different surface realization of the sound from its underlying form. The changes can either be a) assimilatory, when a neighboring sound extends its features to the coarticulated sound, e.g., a labial may change to a velar when immediately preceding another velar in Korean [6]; or b) dissimilatory, when the altered sound becomes less similar to a neighboring sound, e.g., a liquid /r/ is converted to a nonliquid /d/ before/after an adjacent liquid /l/ in Southern Bavarian German [7].…”

Section: Introductionmentioning

confidence: 99%

A Segmentation Effect in Dutch listeners’ Surface-to-Underlying Mapping of Tones

Li¹,

Kager²

2018

Speech Prosody 2018

View full text Add to dashboard Cite

The current study investigates a) whether assimilatory tone sandhi processes, which are motivated by articulatory ease, and dissimilatory tone sandhi processes, which are not, differ in their recoverability of surface-to-underlying tone mapping for naïve Dutch listeners. It also investigates b) whether gradient and categorical underlying-to-surface sound changes in these processes lead to any difference in this mapping. Results from a series of Word Detection tasks reveal no evidence that the listeners perform surface-to-underlying tone mapping more successfully in the assimilation than in the dissimilation condition when the underlying-to-surface tone change is discrete. Also, no robust evidence is found that they perform the mapping more easily in the assimilatory condition when the underlying-to-surface alternation becomes more gradual. Consistent evidence suggests a tone segmentation effect at the surface level playing a pivotal role in the surfaceto-underlying tone mapping task by the listeners. Dissimilation seems to be intrinsically related with easier segmentation whereas assimilation seems to intrinsically lead to more difficult segmentation for the Dutch listeners.

show abstract

The metaphysics of coarticulation

Cited by 52 publications

References 3 publications

Invariants, specifiers, cues: An investigation of locus equations as information for place of articulation

Invariants, specifiers, cues: An investigation of locus equations as information for place of articulation

Vowels as islands of reliability

A Segmentation Effect in Dutch listeners’ Surface-to-Underlying Mapping of Tones

Contact Info

Product

Resources

About