Chapter 4. Where and what is (t,d)? 117
penultimate /k/^24 be realised as glottals; the pertinent variability would seem, then,
to be between glottalised and non-glottalised realisations of final clusters, rather
than between C 2 alternating between zero on the one hand and [t] or [ʔ] on the
other, with word-final (cluster) codas that consist only of a glottal stop somewhat
arbitrarily being deemed as having a deleted or undeleted /t/. Whether or not these
glottal-only codas simultaneously ‘cover’ a weakened (or indeed non-weakened)
alveolar articulation is unknowable from auditory/acoustic data alone, but all except
one of the 55 (t,d) tokens with ‘preceding’ glottals and following vowels or pauses
have alveolar release, which suggests that some alveolar articulation could be present
preconsonantally too. Any alternation between the presence or absence of a ‘covered’
alveolar gesture in glottal-only codas may well be a combination of idiosyncratic
(and therefore cognitive) and physiological constraints (target undershoot). And
the presence of an observable release before almost all vowels and no stop conso-
nants, and before four out of nine following continuants is towards the natural end
of Nolan’s scale. The behaviour of all glottal codas would appear, then, to be a func-
tion of a combination of both cognitive and more ‘natural’ CSPs. Once again this
observation is reinforced by the co-occurrence of glottalisation with other CSPs, as
in (56), with its fully lenited nasal, which is illustrated in Figure 6:
(56) they went and [ðɪwɛ ̰ən] knocked on Andrew’s door
0they went an’w ~ ə nTime (s) 0.4197ðFigure 6. Spectrogram showing they went a n’ (56); male speaker.
- This occasionally also applies to /k/ before plural /s/ as in I’ve only done it for three weeks
[wiː̰ʔs].