As best I can tell from my limited sources, Armenian always allows pro-drop of subject with a definite reading. However, it does not allow all objects to be dropped, since the agreement morphology on the verb is only with the subject. For a long time, I wasn't sure if it allowed the dropping of objects with an indefinite reading (as in the English "He ate"), but I planned to implement it for the verb "eat" on the assumption that it was like English. Luckily, in reading about the dative case, I came across a sentence which glossed as "I give to [name]", where the direct object was omitted. I've assumed that this isn't an ungrammatical example, and that it also receives an indefinite reading on the omitted direct object. I created instances of the basic-head-opt-subj-phrase and basic-head-opt-comp-comp-phrase as described in the lab, although I later had to make an intermediate types in armenian.tdl (without "basic-") that restricted those rules to fully-inflected finite verbs, otherwise they allowed lone infinitives to serve as the main verbs of sentences. I also rearranged my verb types as described in the lab. All my verbs now derive from verb-lex, with the exception of negative-aux-verb-lex, which still anomalously derives from basic-scopal-adverb-lex as described in the last lab. I added ditransitive verbs: both the verb "tal" (give), which can have its direct object omitted, and the verb "tsaxel" (sell), which doesn't (I'm not sure this contrast actually exists, but I wanted a non-optional ditransitive for testing purposes). I added OPT and DEF-OPT as appropriate to my verb types: nom-acc-trans-verb-lex has [OPT -] its 2nd arg nom-acc-optobj-trans-verb-lex has [DEF-OPT -] on its 2nd arg nom-acc-dat-ditrans-verb-lex has [OPT -] on its 2nd and 3rd args nom-acc-dat-optobj-ditrans-verb-lex has [DEF-OPT -] on its 2nd arg and [ OPT - ] on its 3rd. potential-verb-aux-lex has [OPT -] on its 2nd arg negative-verb-aux-lex has [OPT -] on its 2nd arg I had to add these last two because auxiliary verbs were going into an infinite loop (exceeding the maximum number of edges), since the optional comp rule also applied to them. My test suite contains: Auxiliary verbs with missing subjects, transitive and intransitive The ditransitive verb "give" with all arguments, subject missing, direct object missing, and subject and direct object missing, all of which are correct. The ditransitive verb "sell" with all arguments, subject missing, direct object missing, and subject and object missing, the last two of which are incorrect. The transitive verb "eat" with and without a direct object, both of which are OK. The transitive verb "buy" with and without a direct object, the second of which is incorrect. I've included these sentences in test.all as well, and included test.out for a batch parse of all the sentences. The only sentence from previous labs for which the results have changed is: mard e krnay utel man a can eat This now receives two parses: one where the man is the subject and the direct object is omitted (i.e. "the man can eat"), and one where the subject is omitted and the man is the direct object (i.e. "he can eat the man"). Neat! Accordingly, I've removed the star on this sentence. I tried parsing some sentences and checking at the semantics, and they look right. For example, the sentence "utem" ("(I) eat") has a definite subject and an indefinite object. The sentence "mardi ke-tam" ("(I) give to the man") has a definite subject, an indefinite (omitted) direct object, and an indirect object that is unspecified for definiteness (since the article is omitted).