Ryan Mattson 5-15-05 LING 567 Lab 6 Write Up Interrogatives and Imperatives in Finnish Matrix interrogative clauses in Finnish are typically created by inverting the verb and the subject and inflecting the verb with a -ko/-kO (where the capital o represents the fronted o, usually marked with an umlaut; this vowel fronting occurs based on vowel harmony) suffix, which is probably a clitic. A typical matrix declarative looks like 1) Me pelkAAmme tomaattit we-nom fear-1p-pl tomato-acc we fear tomatoes The interrogative corollary to this would be the following 2) PelkAAmmeko me tomaattit? fear-1p-pl-ques we-nom tomato-acc Do we fear tomatoes? / We fear tomatoes? An intransitive verb works in the same way, like the following 3a) MinA kAvelen I-nom walk-1p-sg I am walking b) kAvelenko minA? walk-1p-sg-ques I-nom Am I walking? c) *minA kAvelenko And the ditransitive verbs follow in suit: 4a) hAn antaa tomaattit miehelle he/she-nom give-3p-sg tomato-acc man-allat He/she gives tomatoes to the man b) antaako hAn tomaattit miehelle? give-3p-sg-ques he/she-nom tomato-acc man-allat Is he/she giving tomatoes to the man? Matrix imperatives are simply the uninflected root of the verb, that is to say, the form without the minA or sinA ending. A typical phrase heard in Finnish is in the imperative (though it uses the copula, and I've not yet implemented this particular verb into my grammar): 5) ole hyvAA (cannot be parsed) be-imp good-part be good! An example of an intransitive imperative: 6) kAvele walk-imp walk! Transitive verbs keep the same complements: 7) pelkAA tomaattit fear-imp tomato-acc fear tomatoes! 8) sano ettA minA nAytAn apinalta say-imp comp i-nom look-1p-sg monkey-ablat say that i look like a monkey! And ditransitives work as expected: 9) anna tuot tomaattit miehelle give-imp that-pl tomato-pl man-allat give those tomatoes to the man Current coverage of imperatives The intransitive 'kAvele' (walk!) currently works as expected, however transitive and ditransitive verbs are parsing, but not with correct structures. I implemented imperatives by reinflecting the 2p-Sg form of the verb (for instance, 'kAvelet' walk-2p-Sg) to remove the 't' ending using the imp_verb-lex-rule, then the verb must go through the imperative-head-opt-subj-phrase grammar rule so that it doesn't need a subj, and finally it goes through the finnish-imperative- clause grammar rule to get the semantics. Parsing 'kAvele' provides 3 parses, but two would be removed if the root condition lex-root was removed, thus giving the correct parse with the root node being finnish-imperative. The only concern is that the root node is being labeled 'V' (above an 'S') which leads me to believe some information may be being in the finnish-imperative rule. The transitive and ditransitive seem to be connecting their complements via head-adj-int, which projects to finnish- declarative, as opposed to finnish-imperative. I think I need to constrain head-adj-int to not allow head-dtr's with [IMPERATIVE +], as well as modify my head-comp rule to correctly grab the complements appropriate to transitive and ditransitive verbs. Current coverage of interrogatives Matrix interrogatives are working well, though not perfectly. Certain constructions are overgenerating (though this again seems to be due to an underconstrained head-adj-int rule, as well as redundant inflectional rules that differ in their case marking). A working transitive: 10) pelkAAnko minA tomaattit fear-1p-sg-ques i-nom tomato-acc i fear tomatoes? A working intransitive: 11) kAveleeko hAn walk-3p-sg-ques he/she-nom he/she is walking? A working (though overgenerating, through adj-head-int and head-adj-int) ditransitive: 12) annanko minA tomaattit miehelle give-1p-sg-ques i-nom tomatoes man-allat i am giving tomatoes to the man? The semantics of the matrix interrogatives closely match the example, though they don't exactly match. The obvious differences are in the order of the predications, given that Finnish inverts the subject and the verb in questions. The only other obvious difference in the semantic representation is that where there is an indefinite predication in the English representation, my grammar has a predsort. A representation for 'kAveleeko hAn' (is he/she walking?) follows (lacking the predsort, but with a pronoun predication instead): For embedded interrogative clauses, I treated the wh-like words, such as the current word I have implemented, 'mitA' (what-like, and partitive), as complementizers, as they show up in the same places, for instance in the following sentence: 13) minA tiedAn mitA sinA syOt i-nom know-1p-sg wh-comp you-nom eat-2p-sg I know what you eat (meta-discussion: it just occurred to me that mitA may not be the type of embedded interrogative marker that is desired for this lab, however the syntactic construction would be treated the same for the appropriate marker which would create the sentence: 'I know whether you eat' mitA seems like it would be better handled by long-distance dependency machinary. For the sake of sanity, I'm going to pretend until tomorrow, 5-17, when I can ask my Finnish professor about a corellate to 'whether' in Finnish, that mitA is the same.) The semantics I was trying to implement with a non-branching rule that almost directly mirrors the finnish-interrogative-clause. I called this one a cp-interrogative-clause, but it doesn't seem to be working. The current difference in the rule is that the HEAD-DTR's HEAD must be of type comp. This can be fixed by changing cp-inter- rogative-clause's HEAD-DTR's MSG to be of type message instead of no-msg, but this creates a circular lexical rule, so I've left it as type no-msg in order to parse sentences effectively.