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.

Pettersson and Wood started their investigation by first verifying the existence <strong>of</strong><br />

acoustically neutralizing vowel reduction using spectrographic evidence - the formant<br />

frequencies measured for unstressed /e,o,a/ were found to coincide with those<br />

measured for the vowels /i,u,/, respectively. Since there was no acoustic difference<br />

in vowel quality reflecting these underlying vowel contrasts, we can state that<br />

Bulgarian vowel reduction is acoustically neutralizing” (Crosswhite 2001: 42).<br />

But this is simply false, both as a characterization <strong>of</strong> vowel reduction in the S<strong>of</strong>ia dialect,<br />

and as a characterization <strong>of</strong> the experimental findings <strong>of</strong> Pettersson and Wood (1987).<br />

What Pettersson and Wood actually report is the opposite 16 : On pages 244-247 they show<br />

scatterplots <strong>of</strong> their results for unstressed vowels spoken in isolation, and also in frame<br />

sentences. Unsurprisingly, there is a tendency toward more reduction in the frame<br />

sentences than in forms spoken in isolation. For one speaker in the isolated words even<br />

unstressed /a/ and /â/ form clearly distinct, if overlapping, distributions. Since they<br />

provide no statistical analysis, it is hard to see whether otherwise /a/ and /â/ form one<br />

distribution or two. /o/ and /u/ overlap for both speakers, but in both contexts many <strong>of</strong> the<br />

realizations <strong>of</strong> /o/ fall well outside the distribution for /u/. Again, statistical analysis<br />

would say for certain. For /e/ and /i/, however, there is very little overlap for either<br />

speaker at either rate, to say nothing <strong>of</strong> a single distribution. Most importantly, to the<br />

16<br />

It may be that the confusion is in their use <strong>of</strong> the term neutralizing, which they employ in several places<br />

in an articulatory sense, meaning “becoming more neutral”. For example, on p. 240, <strong>of</strong> the failure <strong>of</strong> mid<br />

and low vowels to merge completely in many dialects they say the following: “... selective relaxation <strong>of</strong><br />

articulatory control and neutralization <strong>of</strong> components <strong>of</strong> the unreduced vowel would diminish articulatory<br />

precision and lead to a spectral transition between unreduced and fully reduced forms.” Describing as they<br />

are a gradient “transition” from one vowel to another, they clearly do not mean neutralization in the sense<br />

<strong>of</strong> Aufhebung. They then continue, “Alternatively (italics mine), switching between complete<br />

configurations, by substituting [i] for [e] etc. in the underlying motor programme would lead to discrete<br />

shifts between reduced and non-reduced forms”, viz. actual neutralization.<br />

63

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

Saved successfully!

Ooh no, something went wrong!