The phonetics and phonology of non-modal vowels: a cross-linguistic perspective

100

different fricatives:  fricative(s) with greater glottal openings in a given language are

more likely to trigger devoicing in that language.

Interestingly, in Korean, the asymmetry between stops and fricatives depends on

whether the stop or fricative appears on the right or left side of the potential target of

vowel devoicing (Jun and Beckman 1994, Jun et al. 1997, 1998).  Fricatives and

aspirated stops are more likely to trigger devoicing than fortis and lenis stops when

they precede a  vowel,  an  expected  pattern  given  the  greater  glottal  apertures  of

fricatives and aspirated stops.  However, following a vowel, all stops, including fortis

and lenis stops trigger devoicing more often than fricatives, even though fricatives

have greater glottal openings than fortis and lenis stops.  Jun and Beckman suggest

that the closing gesture into a stop might be faster than the oral constriction gesture

made for a fricative; this greater velocity of the oral closing gesture could lead to a

more abrupt increase in oral pressure which could inhibit voicing in the preceding

vowel.    One  might  hypothesize  that  the  asymmetry  between  Korean  and  those

languages  in  which  devoicing  is  triggered  by  a  following  fricative  and  not  a

following  stop  (Comanche,  Goajiro,  and  Southern  Paiute--singletons)  is  due  to

language specific differences in the relative magnitude and timing of glottal opening

gestures in the two classes of consonants.

In summary, although the glottal overlap story does not account for all cases of

devoicing

20

, it nevertheless offer a coherent explanation for many of the devoicing

asymmetries.  There are two robust asymmetries, however, which do not fall out

directly from a gestural overlap account without recourse to other factors.  First,

there is the tendency for low-toned vowels to preferentially devoice over high-toned

vowels pointed out in section 3.5.  The existence of this asymmetry is presumably

linked to the inherent inability of voiceless  segments  to  carry  tone  phonetically.

Thus, devoicing both high and low toned vowels would lead to neutralization of a

tonal contrast, either a lexical contrast in the case of languages with lexical tone, or a

semantic contrast, in the case of sentence (or phrase-level) intonation, e.g. questions

vs. statements.  In order to maintain the contrast, a language could  thus  devoice

either high toned or low toned vowels but not both.  Perhaps low  toned  vowels

devoice, because the articulatory gestures involved in producing low tone are more

compatible  with  the  glottal  abduction  gestures  associated  with  devoicing.    An

explanation for the tone asymmetry must await instrumental research.

The next asymmetry left unexplained by the gestural account of devoicing is the

fact that devoicing of final vowels is so prevalent, in fact, even more prevalent than

the  devoicing  of  word-medial  vowels  in  the  vicinity  of  voiceless  consonants.

Furthermore,  devoicing  in  final  position  typically  takes  place  even  when  the

preceding consonant is not voiceless.  Strikingly, as pointed out in section 3.3., final

devoicing itself respects an implicational hierarchy.  The occurrence of devoicing in

final  position  of  a  given  domain  implies  devoicing  in  final  position  of  smaller

domains.    Thus,  utterance  final  devoicing  in  a  languages  implies  phrase  final

devoicing  which  implies  word-final  devoicing.    This  implicational  hierarchy  of

devoicing can be explained in terms of the decline in subglottal pressure throughout

the course of an utterance (Dauer 1980); this drop in subglottal pressure results in a

decrease in the volume-velocity of air flow through the glottis which in turns inhibits

devoicing.  Subglottal pressure is lesser in final position of larger domains than in

final position of smaller domains; hence, the likelihood of devoicing increases the

larger the domain.  Because subglottal pressure is lowest utterance finally, vowel

devoicing is most common in this environment; cross-linguistic devoicing patterns

reflect this  fact.    Those  vowels  whose  glottal  adduction  gestures  are  inherently

hypoarticulated either in terms of magnitude or duration, e.g.  short  vowels,  high

tải về 78.41 Kb.

Chia sẻ với bạn bè của bạn: