20.07.2013 Views

Positional Neutralization - Linguistics - University of California ...

Positional Neutralization - Linguistics - University of California ...

Positional Neutralization - Linguistics - University of California ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

preceding consonants, etc.). It seems unlikely, however, that a vowel would leave <strong>of</strong> all<br />

things glottal constriction on a preceding consonant unless it were itself glottalized. We<br />

might hypothesize then that the direct ancestor to the Gonja forms above was one with a<br />

short, low amplitude, creaky-voiced final vowel, whose laryngealization may either have<br />

affected the sonorant itself in implementation, or at least have obscured the already<br />

lower-salience acoustic modulation <strong>of</strong> the transition from the sonorant to the short<br />

vowel 93 , such that the sequence could be reinterpreted as merely a laryngealized<br />

sonorant 94 . This would mean Hyman is correct to see this correspondence as the product<br />

<strong>of</strong> reduction <strong>of</strong> a final segment, with the sole caveat that the pre-deletion phonetic<br />

realization <strong>of</strong> the final segment in question was quite probably the result <strong>of</strong> the<br />

phonologization <strong>of</strong> prepausal creakiness. This may seem a pedantic distinction, but it is<br />

an important one as well if the question we are posing is an etiological one 95 .<br />

Before looking more closely at the crosslinguistic typology <strong>of</strong> final glottalization,<br />

it is necessary to address the phonetic issues surrounding the development <strong>of</strong> creaky<br />

voice in phrase-final syllables. Most sources treating creaky phonation characterize its<br />

production and acoustic and aerodynamic consequences similarly: Gordon and<br />

93<br />

As compared with, e.g., that <strong>of</strong> the transition from obstruent to vowel.<br />

94<br />

The additional duration supplied by the misparsed final vowel could actually help in this regard, making<br />

a (probably shorter) onset sonorant sound more like a (probably longer) coda sonorant.<br />

95<br />

In treating these facts, <strong>of</strong> course, one might wonder about the role <strong>of</strong> the ATR specification <strong>of</strong> the lost<br />

vowel in this connection. The cognate dialect forms retaining the vowel, however, show both vowel series<br />

yielding the same result upon deletion.<br />

213

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!