100 Rosalind A. M. Temple
In §2, I examine a range of CSPs to ascertain whether the range of phonetic
patterns found in (t,d) consonants is consistent with a CSP analysis and whether
these patterns are exclusive to (t,d) consonants. The analysis will touch on issues
which must be taken into account in deciding whether word-final clusters and/
or other CSPs are amenable to analysis in terms of variable rules. These are issues
which have long been the subject of discussion in the phonetics literature and they
have not gone entirely unnoticed in discussions of (t,d), having been raised by e.g.
Wolfram (1993), but there is little subsequent evidence that Wolfram’s concerns
have been heeded. In the discussion in §3, I turn to the implications of these
observations for modelling the behaviour of word-final stop consonants in the
grammar in the light of ongoing debates about the phonetics-phonology inter-
face, a prerequisite to sociophonetic/sociophonological modelling. I thus hope
to demonstrate how, paradoxically, advances in sociophonetics might sometimes
be achieved by stepping back and re-examining the phonetic detail behind a rule
which is generally held to be predominantly a categorical phonological one. It will
be seen that much can emerge from such an apparently retrospective approach
which can contribute to advances in sociophonetics and wider debates concerning
the relationship of its findings to phonetic and phonological theory, albeit there
are questions which will remain unanswered until further advances are made by
applying particularly articulatory techniques to this variable.- (t,d) and Connected Speech Processes^7
In contrast to the phonologically based accounts of (t,d), which posit a categori-
cal alternation between the presence and absence of a surface reflex of underlying
word-final /t,d/, CSPs provide, in Nolan’s words, “a way of describing a continuum
of decreasing phonetic explicitness” (1996: 15). The degree of explicitness is influ-
enced by adjacent segments or by prosodic and other factors like speech rate or
by language-specific or variety-specific conventions or, most likely, by a combina-
tion of some or all of these factors. Thus some processes are more “phonetically
natural” than others in that they arise more directly from the physical constraints
inherent in the vocal mechanism, while others must be seen as arising from cogni-
tive processes (Nolan 1996: 19). Between the two extremes “phonetic naturalness”
is a matter of degree, rather than there being a simple dichotomy between effects- The process-based characterisation of these phenomena implies an analysis in terms of
rules operating on segments in citation forms; the discussion here will adopt that descriptive
convenience, following Nolan and others, but this should not be taken as representative of a
commitment to any theoretical analysis in such terms.