Argument Structure Summary for February 8, 2005

We began a discussion of how to account for the various properties of the inner and outer causatives.

First, we briefly discussed the "lexicalist" notion that the properties of inner causatives might have to do with a distinction between the lexicon and syntax. In brief, this involves inner causatives being listed (accounting for morphological and semantic irregularity) and wordlike (accounting for relative opacity to syntax).

Next, we discussed the extent to which the faire-par and faire-infinitive properties fit into the overall typology. According to the checklist we went over on February 1st, faire-par causatives are typical outer causatives. But faire-infinitives do not seem typical of inner causatives: They are not very restrictive in their selection of roots (e.g. they combine freely with transitive as well as intransitive verbs) they show no signs of morphological irregularity, and they do not show signs of syntactic opacity.

However (as pointed out by Michal), there are certain parallels: the highest argument of the base verb becomes an obligatory argument, both in faire-infinitive and in the inner causative, while the highest argument of the embedded verb becomes adjunct-like in the faire-par and the outer causative. Furthermore, according to a thesis by Guasti with which Michal has at least some passing familiarity, the faire-infinitive (or faire-à) has a direct causation reading which can be likened to the Hindi -aa causative, while the faire-par has an indirect causation reading akin to Hindi -vaa.

Then we moved on to a summary by Patrycja of her analysis (originally presented at the Argument Structure workshop in November) of the Amharic causatives, and verb classes in Amharic more generally. The analysis decomposes "as-" into a causative "a-" and a nominalizing "s-." The optionality of the causee is derived by allowing causative "s-" to optionally select either a vP with an external argument or one without, a difference ascribed to properties of two different v heads, à la Salish. In effect, this gives a faire-infinitive-like causative (if the obligatory external argument is chosen) or a faire-par-like causative (if the other v is chosen) -- note that this account implicitly seems to reject the analogy of faire-infinitive and inner causative. The nominalizer, on Patrycja's account, allows existential closure over the external argument, rather like a passive v. She also postulated a "voice" head higher up than v, which did not interact with passive but was crucial in an account of case assignment in Amharic causatives.

An important question is to what extent can such an account generalize to the other languages? As we heard from Travis at the Workshop, Malagasy has a nominalizing "f-" prefix (if I recall correctly), suggesting that a nominalization account might extend naturally to Malagasy. But Amharic does not seem to have a nominalizing "s-" prefix, nor Hindi a "-v" suffix, nor Kitharaka an "-ith" suffix, and so on. Furthermore, it is not entirely clear how many distinctive properties of nominals per se are involved in the analysis; give the level of abstraction, involving a "voice" that is not related to passive, an "X" which assigns accusative, a "v" which projects an external argument and another "v" which binds one off, plus a "cause" head (this being necessary for the spelling out of the "a-" prefix), it is unclear to me that some other "verbal" category could not do the work of the putative nominalization, e.g. in preventing passive from applying.

Next time, we will hear from Gillian, who will also recap her Argument Structure Workshop account, this one being designed for those persnickety Hindi/Urdu causatives.

As a special added bonus feature, next Tuesday Michal will address the morphological generalization that was the centerpiece of my talk, that the inner causative is phonologically a "subset" of the outer causative, and that furthermore the shared part is further from the stem than the non-shared part (not true for Saami):

Nivkh: inner V-u, outer V-g-u
Hindi: inner V-aa, outer V-v-aa
Kitharaka: inner V-i, outer V-ith-i
Amharic: inner a-V, outer a-s-V
Malagasy: inner an-V, outer an-f-V

(Complication: Nivkh allows double causative V-u-g-u, and Malagasy allows an-f-an-V; this opens the possibility that the Kitharaka outer causative is correctly parsed V-i-th-i; but on either parse Nivkh and Malagasy seem to have more options than the other languages.)

Peter