Re: Re[2]: [tied] More nonsense: Is English /d/ truely voiced?

From: Piotr Gasiorowski
Message: 19083
Date: 2003-02-23

----- Original Message -----
From: "Glen Gordon" <glengordon01@...>
To: <cybalist@yahoogroups.com>
Sent: Sunday, February 23, 2003 1:09 PM
Subject: Re: Re[2]: [tied] More nonsense: Is English /d/ truely voiced?



> >By the way, the rest of the quotation -- the part that you snipped -- makes it clear that devoicing during the stop is common in English.
>
> Common where? Amongst whom? Describe where in English devoicing of /d/ occurs -- what environments, dialects, etc. I'm truely not aware of this devoicing. However, I would say, that at least with myself, I side towards excessive voicing. Perhaps my personal speech pattern, or dialect, plays a part in my apparent ignorance of this devoicing.

Devoicing is certainly common in all British accents and has been described by phoneticians since the nineteenth century. Here are a couple of typical descriptions:

Otto Jespersen (1909) [about English /b/]: "... the vocal chords are generally in vibration so as to make the sound 'voiced'; this is always the case between voiced sounds, as in _robber_, _to begin_; in the beginning of a sentence, as in _Begin_ after a pause, the vibrations do not, however, begin till immediately before the opening of the lips; at the end of a sentence, as in _rob_ before a pause, the vocal chords vibrate in the first part part only of the [b] and then are gradually removed from one another. ... The latter kind of gliding is probably found before voiceless sounds as well, for instance in _lobster_ [lObst&], webster [webst&]."

J.D. O'Connor (1973): "In RP the allophones of /p, t, k/ are invariably voiceless, those of /b, d, g/ may be and often are voiced but they do not generally have full voicing (i.e. from closure right to release) unless they are surrounded by other voiced sounds. In _oboe_, _table_, _under_, _tidy_, _anger_ and _ago_ [b, d, g] are fully voiced, but in _abscess_, _bedsore_, _bagpipe_, where they are in contact with a voiceless sound, or when they occur immediately before of after a pause, as in the isolated word _dog_, voicing does not go right through. Indeed, it may not be present at all.

... Taking the RP case, aspiration is of great importance in contrasts such as _pie_/_by_ and voice much less important since /b/ in this position may have little or none. In _sacking_/_sagging_ the voicing contrast is much more important for distinguishing /k/ from /g/ since /k/ has little or not aspiration but it is voiceless whereas /g/ is fully voiced. Before a pause, as in _site_/_side_, voicing is again relatively unimportant since the /t/ has none and /d/ may well have none either; the aspiration difference is minimal since /t/ has little in this position, and the wight is on the fortis/lenis feature, /t/ being more strongly and /d/ more weakly articulated. But in this position there is often help to be had in making the distinction from quite another quarter: the vowel is of noticeably different length before /t/ and /d/. In _site_ it is considerably shorter, in _side_ considerably longer, and this is regularly so before fortis/lenis consonants in final position ..."

Laboratory data collected from American English speakers, while certainly incomplete, show that at least some North American accents are not significantly different from RP as regards the pattern of devoicing. (Most of the research I'm aware of has been done at UCLA by Peter Ladefoged, Ian Maddieson, Patricia Keating and others, all of whom have also done field or laboratory research on lots of languages other than English and so are intimately familiar with all sorts of phonation mechanisms. They certainly know what they measure and what they are talking about.) I've seen it reported, however, that fully voiced and even implosive stops may be found initially in some American dialects (Wells 1992 impressionistically qualifies the occasional use of implosives as a Southernism). The role of aspiration as the source of acoustic cues for fortis stops may be generally smaller in North American Englishes (also in some British ones, e.g. in Lancashire) than in RP. The devoicing of liquids and semivowels after /p, t, k/ (in words like <pride, trap, crime, crab, cluster, pew, cute, twin, queen> seems to me to be stronger in RP than anywhere in America.

Piotr