Voicing of [sg] Buccalization and voicing of the feature [sg]
In tableau 30, the input has a root with an initial glottal stop, which has the plural morpheme [+sg]. Candidate a fatally violates high ranking OCPLar, and candidate b does the same to
P,TLar in an attempt to change its coronal value. Only candidate c preserves both markedness constraints, but at the added featural change for lateral.
The input in example 31 is compared with the faithful candidate a, but this candidate violates the OCPLar. Candidate b is an attempt to buccalize with a plain coronal stop, but is unsuccessful for
language specific markedness reasons 28. Candidate c is the optimal candidate in that it preserves OCPLar, while also allowing for a surface form of the plural morpheme [+sg].
8.3.2 Voicing of [sg]
When a voiceless bilabial stop is affixed with aspiration, the surface form is always a voiced bilabial stop. That is, aspirated voiceless bilabial stops are non-existent in word initial position, but always
neutralize with bilabial stops. We have already established the language-specific markedness constraint 28 to account for this idiosyncrasy. However, the question remains as to how to explain this
alternation with voice in a principled way. First, the fact that voiceless bilabial stops would be singled out to alternate with voice has
articulatory support Lisker and Abramson 1964, Pickett 1999. A voiced stop requires two gestures:1 the occlusion of air flow by some oral articulator and 2 near-simultaneous vocal fold vibration. The
release of a stop is the effect of air pressure built up from egressive airflow pushing against the oral occlusion. This pressure takes longer to build up for labial consonants compared to velars, due to a
longer vocal chamber. Because of this longer duration pressure build up, voicing is most optimal with labials and least so with segments articulated further back in the vocal tract such as velars or uvulars.
The preference for voiced bilabials over voiced non-bilabials can be explained by ranking a faithfulness constraint for bilabial voice below that of non-labial stops. The constraint is given in 32 followed by a
ranking in 33. 32
IDENT:P[vce] ‘Output p segments have input correspondents for [vce]’.
33 Ranking of voicing markedness
IDENT:K[vce]IDENT:T[vce]IDENT:P[vce] The ranking in 33 says that the value of voice for a voiceless bilabial stop is more violable than the
value for voice among non-bilabials e.g. coronals and velars. Taking this approach toward bilabial stop voicing, we can integrate 32 into the ranking given for
lateral buccalization in 29 to produce the final ranking in 34. 34
Ranking of bilabial voicing with other feature faithfulness constraints. OCPLar, P,TLarIDENT[lat, cor]IDENT[cor]IDENT:P[vce]
The low ranking of IDENT:P[vce] predicts that voicing for a bilabial will be preferred over a form that is the output of the buccalization process.
35 Voicing of bilabial stops when [sg] is present word initially.
Input: ʰ-pǽ OCP P,TLar IDENT[lat,cor] IDENT[cor] IDENT:P[vce] a.
pʰǽ b.
lʰǽ c.
tʰǽ d.☞
bǽ The input has a root with an initial bilabial stop affixed with [+sg]. The faithful form in candidate
a would be preferred if it were not for the presence of high ranking P,TLar. Candidates b and c both fail because the IDENTITY constraints involved in those processes are more ‘expensive’ than the
violation of IDENT:P[vce] for candidate d. Thus, d is the winner.