3.8 w-Insertion P-7.

The alternations discussed in this section, though superficially resembling those seen in the preceding section, are sufficiently distinct in detail to justify separate treatment.

The NA and NgARA nonhuman noun classes are marked, in what I call the Continuous series, by prefixes /na-/ and /ngara-/ respectively. In the Punctual series the two are merged as /yi:-/ (ยง4.7).

Several nouns in these classes begin with short /a/ After /na-/ and /ngara-/ the usual VV-Contraction rule P-49 applies and we get such forms as //na-aragu///na:-'ragu/ 'the young male dugong' and //ngara-anarag///ngara:-'narag/ 'the young goanna'. However, with prefix /yi;-/ VV-Contraction does not occur (since the prefix vowel is long). For some speakers the outputs are /yi:-yaragu/ and /yi:-yanarag/ with surface /y/ However, other speakers (especially younger and middle-aged ones) prefer /yi:-waragu/ and /yi:-wanarag/ with surface /w/ in all such forms. For the /w/ dialect, a rule inserting the /w/ is needed, and even speakers of the /y/ dialect are compatible with this analysis, since /w1/ usually becomes surface /y/ in the environment /i__a/ (see P-9 and P-10).

For some (apparently most) speakers of the /w/ dialect, in the one relevant case where the noun root begins with long /a:/ we get /y/ instead of /w/ the form being /yi:-ya:muny/ 'common fly' from /a:muny/ This suggests that /w/ is not inserted in this form, except for certain speakers who do say /yi:-wa:muny/

(P-7)w-Insertion (for some speakers, mostly young)
Ø → w // yi:-___a
where /yi:-/ is the noun class prefix, and (for most speakers) the /a/ is short

Reasons for dissociating this rule from P-6 are that the latter applies only in a long-vowel form (with /a:/) while P-7 applies for most speakers only in short-vowel forms, and that P-6 applies only to a root preceded by a derivational prefix or compound initial while P-7 applies after an inflectional prefix.

P-7 is designed to apply to forms involving nouns which clearly begin with /a/ or for which there is no specific evidence that there is any underlying segment (such as a semivowel) preceding the /a/ It should be noted that there are some other stems which appear to begin with underlying /w1a/ which loses its /w1/ in some surface environments (including word-initial position). For these roots (which normally cannot undergo VV-Contraction P-49, for example) an alternation of the type word-initial /a.../ vs. prefixed /yi:-wa.../ need not be due to P-7. The latter rule applies to all nouns beginning in /a/ which occur after /yi:-/ including those which clearly have underlying initial /a/ instead of /w1a/ such as /ala:di/ 'bad'.