Another look at velar deletion in Turkish, with special ... - Linguistics
Another look at velar deletion in Turkish, with special ... - Linguistics
Another look at velar deletion in Turkish, with special ... - Linguistics
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
salacak salaʤak salaʤakɯ ‘slab for corpse’<br />
Selanik selaːnik selaːniki ‘Salonika’<br />
s<strong>in</strong>agog s<strong>in</strong>agog s<strong>in</strong>agogu ‘synagogue’<br />
sitreptokok sitreptokok sitreptokoku ‘streptococcus’<br />
In conclusion, when we take <strong>in</strong>to account the part of speech condition, the fact th<strong>at</strong><br />
suffix-<strong>in</strong>itial <strong>velar</strong>s are immune, and the existence of lexical exceptions, it is clear th<strong>at</strong><br />
while be<strong>in</strong>g <strong>in</strong> a morphologically derived environment may be necessary for an <strong>in</strong>tervocalic<br />
<strong>velar</strong> to undergo <strong>velar</strong> <strong>deletion</strong>, it is not sufficient.<br />
5. Phonological condition<strong>in</strong>g factors<br />
We turn next to an exam<strong>in</strong><strong>at</strong>ion of phonological condition<strong>in</strong>g factors, to see whether the<br />
phonologically derived environment condition is necessary or sufficient.<br />
Velar <strong>deletion</strong> is subject to two well-known conditions on its applic<strong>at</strong>ion: vowel length<br />
and prosodic size.<br />
5.1. Vowel length<br />
As noted by Sezer (1981), <strong>velar</strong> <strong>deletion</strong> does not apply when the preced<strong>in</strong>g vowel is long:<br />
(12)<br />
nom<strong>in</strong><strong>at</strong>ive d<strong>at</strong>ive gloss<br />
/<strong>in</strong>filaːk/ [<strong>in</strong>.fi.lak] [<strong>in</strong>.fi.laː.ka]<br />
<strong>in</strong>filak <strong>in</strong>filaka<br />
/meraːk/ [merak] [me.raː.ka] ‘curiosity’<br />
merak meraka<br />
This is consistent <strong>with</strong> the f<strong>in</strong>d<strong>in</strong>g of Zimmer & Abbott 1978, whose experimental<br />
subjects were more likely to preserve the f<strong>in</strong>al /k/ <strong>in</strong> the nonsense word istisāk, <strong>in</strong> which a<br />
long vowel preceded the k, than <strong>in</strong> comparable k-f<strong>in</strong>al words <strong>with</strong> a short vowel. We can<br />
only specul<strong>at</strong>e as to the reason for the correl<strong>at</strong>ion between vowel length and <strong>velar</strong> <strong>deletion</strong>.<br />
Vowel length is an <strong>in</strong>dic<strong>at</strong>or of non-n<strong>at</strong>ive orig<strong>in</strong>, and Zimmer & Abbott found th<strong>at</strong><br />
nonsense words th<strong>at</strong> resembled Arabic loans were less likely than other nonsense words to<br />
particip<strong>at</strong>e <strong>in</strong> <strong>velar</strong> <strong>deletion</strong>. In the TELL d<strong>at</strong>abase, however, words <strong>with</strong> long vowels <strong>in</strong><br />
syllables other than the f<strong>in</strong>al (e.g. tarik /taːrik/ ‘profession’) undergo <strong>velar</strong> <strong>deletion</strong> <strong>at</strong> an<br />
80% r<strong>at</strong>e, suggest<strong>in</strong>g th<strong>at</strong> the connection is not simply etymological. <strong>Another</strong> possibility is<br />
th<strong>at</strong> <strong>velar</strong> <strong>deletion</strong> is <strong>in</strong>hibited when the result would be a sequence of long vowel followed<br />
by short vowel. <strong>Turkish</strong> does have some words conta<strong>in</strong><strong>in</strong>g th<strong>at</strong> configur<strong>at</strong>ion, e.g. nihai<br />
[nihaːi] ‘f<strong>in</strong>al’, vaiz [vaːiz] ‘Muslim preacher’, Azrail [az.raːil] ‘angel of de<strong>at</strong>h (Islam)’, but<br />
it is possible th<strong>at</strong> they are a closed, marked class.<br />
9