330 Chapter 10 Modifications to argument structure
barung ‘small shop’ marung ‘run a barung’
kole ‘day laborer’ ngole ‘work as a day laborer’
tokang ‘artisan’ nokang ‘work as an artisan’
sapedha ‘bicycle’ asapedha ‘ride a bicycle’
jala ‘net’ ajala ‘fish with a net’
motor ‘car’ amotor ‘ride in a car’
saba ‘rice paddy’ asaba ‘work in a rice paddy’
tane ‘farm’ atane ‘farm’
sakola ‘school’ asakola ‘go to school’
sapatu ‘shoe’ asapatu ‘wear shoes’
sorban ‘turbin’ asorban ‘wear a turbin’
kalambi ‘shirt’ akalambi ‘wear a shirt’
kaos ‘t-shirt’ akaos ‘wear a t-shirt’
The forms in (169) portray the actor voice in a distinctly derivational manner.
The left-hand column appears to consist of a list of nouns and the right-hand
column as verbs derived from them through affixation of ng- or a-. Additional-
ly, as the affixation of ng- or a- creates an argument-taking predicate, this
process introduces an argument slot into the lexical item, much as an applicative
morpheme can. This increases the parallelism between some uses of -e and -agi
and ng- and a-. Clearly, when used with transitive and ditransitive verb roots,
actor voice morphology is not always associated with an increase in the valence
of its host or derives a verb from an item from a different lexical category. It is
equally clear that the same is true of -e and -agi. Outside of causatives, -e and
- agi do not add to the semantic valence of the verbs to which they are affixed.
Thus, -e and -agi are no more straightforwardly derivational than is actor voice
ng- and a-.
Other analyses of the facts in (169) are, of course, conceivable. The forms
in the left-hand column need not be analyzed as nouns. It is possible that they
are zero-derived verbs (such as hammer and saw in English), in which case the
actor voice forms in the right-hand column are simply inflected verbs. That said,
there is no particular evidence that determines such an analysis as necessarily
correct. Discussion in Chapter 3 section 3.4 suggests the possibility of analyzing
these as precategorial roots. But such an analysis entails that the actor voice
morphology in these cases derives a verb from a precategorial root. The preca-
tegorial root analysis for Philippine languages entails that voice morphology
there is associated with derivation of just this sort (Foley 1998, Kroeger 1998,
Himmelmann 2007). So, in this view, Philippine voice is derivational and not
inflectional. If Austronesian voice systems (including that of Madurese) should
be considered purely inflectional or derivational begs an easy answer.
The distribution of -e and -agi points to an additional similarity to Taga-
log voice. While -e and -agi can occur with verbs taking actor voice, speakers