There is a rule (V-Fronting P-50) which converts /a/ or /u/ into /i/ before palatoalveolar /j/ or /ny/ There are also some apparent cases of //ay// or //uy// becoming /i/ in similar positions. It is reasonable to consider these as cases where V-Fronting converts //ay// or //uy// to //iy// which is then reduced to /i/ by a rule we will call iy-Contraction.
In the case of /-w1aranggay/ 'spear shaft', the two rules fail to apply in the noun forms, which require derivational noun class prefix, hence /ma-garanggay/ from //maG-w1aranggay// 'spear shaft (MANA class)'. However, the rules apply in case the stem is used as a compound initial (with following verb stem), as in /-w1aranggi=ma-/ 'to pick up spear (shaft)'. See Table 14-2.
In the case of verb /=lha-/ 'to stand', the usual Past2 inflected form is /=lha-y/ as in /wuru=lha-y/ 'they stood'. This form never contracts. However, with a preceding compound initial or a derivational prefix like Multiple /-w1aragara-/ the form /=lha-y/ frequently reduces to /=lhi-Ø/ as in /wu-waragara=lhi-Ø/ 'they (ANA class) were all standing (e.g., trees)'. The full form /wu-waragara=lha-y/ is also possible but not so common. On the other hand, the paradigmatically parallel Past2 form /=ya-y/ 'slept' does not reduce to /=yi-Ø/ even in compounds, the Multiple form for this stem being /-lhagara=ya-y/ 'all slept'. See Table 11-5 (NgA2).
There are also some other possible cases of //ay// reducing to /i/ by P-50 and P-15, but they involve fairly abstract analysis of inflectional verbal paradigms. First, the NgA1 (i.e., first NgA) verb class shows stem-final /a/ in most forms: Past1 /=bura-nga-ny/ 'sat down', etc. However, the Past2 is of the type /=buri-Ø/ We could analyse this as //=bura-y// with P-50 and P-15 converting //ay// to //iy// and then /i/ Moreover, there are clear affinities between the NgA1 class and that of /=lha-/ 'to stand' (preceding paragraph), which I call the NgA2 type (essentially the monosyllabic-stem equivalent of the NgA1 type). Again, see Table 11-5.
There is also a much more general inflected form for many verb classes which we here call Nonpast3 (mainly used as future negative). For the great majority of stems ending in /a/ or /i/ the Nonpast3 is of the form /i-Ø/ Obviously, for stems ending in /i/ we can regard this as a zero suffix, but for those ending in /a/ it is possible to set up the suffix as //-y// with P-50 and P-15 then changing //ay// to /i/ Thus /yagi niwu=ma:ndhi-Ø/ 'he will not make it (ANAwu)' from /=ma:ndha-/ 'to make'. It is also possible to set up /-y/ as Nonpast3 suffix for the relevant stems ending in /i/ so that //i-y// is also reduced to /i-Ø/ by P-15· There are one or two verbs ending in /u/ which also show Nonpast3 /i-Ø/ as with /=rugu-/ 'to weep' (class U3) , Nonpast3 /=rugi-Ø/ and of course we can set this up as //=rugu-y// with P-50 and P-15 then applying (compare Past1 /=rugi-ny/ 'wept' showing P-50 before palatoalveolar). However, there are other Nonpast3 forms, notably /u:-'/ for some other stems ending in /u/ where it seems that an underlying Nonpast3 allomorph other than /-y/ should be recognised. See Table 11-3.
It is possible to formulate P-15 as an automatic rule reducing //iy// to /i/ in syllable-final position. The various morphological restrictions on the reduction of //ay// to /i/ noted above, are restrictions on P-50 (which changes the //a// to /i/ here) rather than on P-15.
| (P-15) | iy-Contraction | ||
| y → Ø // i___ | | # | |
| C | |||