Control and null subjects are governed by morphosyntax in Finnish Pauli Brattico Talk presented at the Uralic Syntax Workshop Budapest May 2016 (Version 1.0)
Null subjects and control (1) Pekka väitti että saa ylennyksen. Pekka claimed that get.3sg promotion 'Pekka claimed that he will get a promotion.' (2) Pekka väitti saavansa ylennyksen. Pekka claimed get.va.3sg promotion 'Pekka claimed that he will get a promotion.' (3) Pekka halusi lähteä. Pekka wanted leave.a 'Pekka wanted to leave.' (4) Pekka käski *(Merjan) lähteä. Pekka asked Merja.GEN leave.a 'Pekka asked Merja to leave.'
Goals (1) Pekka väitti että saa ylennyksen. Pekka claimed that get.3sg promotion 'Pekka claimed that he will get a promotion.' (2) Pekka väitti saavansa ylennyksen. Pekka claimed get.va.3sg promotion 'Pekka claimed that he will get a promotion.' Question A. When a subject can or must be null and/or controlled? (Licensing) Question B. How are they controlled? (Theory of control) Question C. What types of null subjects are there? (Typology of NSs)
Answers for Finnish (1) Pekka väitti että saa ylennyksen. Pekka claimed that got.3sg promotion 'Pekka claimed that he will get a promotion.' (2) Pekka väitti saavansa ylennyksen. Pekka claimed get.va.3sg promotion 'Pekka claimed that he will get a promotion.' Question A. Appearance of null subjects is regulated by morphosyntax Question B. There are two mechanisms for finding antecedents: (A) grammatical mechanism that is based on c-command and (B) discourse strategy based on topic. They operate in paralle and interact with each other. Question C. Two types in Finnish: pro and PRO, that have different properties with respect to A-B.
Type I null subject pro Generated optionally by heads that exhibit agreement, and only when they exhibit agreement, and have a Spec/EPP position for overt pronouns. Own rules for control (explained later) (1) He väittivät että (he) saa-vat nukkua pitkään They claimed that can.3sg sleep long 'They claimed that they can sleep long.' (2) He väittivät että täytyy nukkua tarpeeksi. They claimed that must.0 sleep enough 'They claimed that one must sleed enough.'
Finite clause... Noun phrase (NP/DP) (Minun) auto-ni 'I.gen car-1sg' TUA-adverbial Adposition phrase Lapsi nukahti {pro lue-ttu-aan kirjaa} child fell.asleep read-tua-3sg book 'The child fell asleep after s/he read the book.' ESSA-adverbial (Minun) lähellä-ni 'I.gen near-1sg' Lapsi nukahti {pro luki-essa-an kirjaa} child fell.asleep read.essa-3sg book 'The chid fell asleep when s/he was reading the book.' MA-participle Pekka palautti {{pro löytä-mä-nsä} kirjat} Pekka returned found-ma-3sg book 'Pekka returned the books he found.' VA-infinitival Pekka uskoi {pro lähte-vän-nsä} Pekka believed leave-va-3sg 'Pekka believed that he will leave' The rules are: 1. Null subject only if there is agreement 2. Overt subject can replace the null subject 3. Overt subject does not require agreement 4. Control works the same way (more about later) 5. Null subject optional
Type II null subject PRO Generated obligatorily by heads that never exhibit agreement and have no Spec/EPP position for overt pronouns. Own rules for control (explained later) (1) Pekka halusi PRO lähteä. Pekka wanted leave.a 'Pekka wanted to leave.' (2) *Pekka halusi Merjan lähteä (nothing can be at Spec). Pekka wanted Merja.GEN leave.a (3) *Pekka halusi PRO lähteä-nsä (this verb does not agree) Pekka wanted leave.a-3sg
want + A-infinitival Pekka halusi PRO lähteä Pekka wanted leave.a 'Pekka wanted to leave.' MA-infinitival Pekka näki Merjan {PRO lähtemässä} Pekka saw Merja leave-ma 'Pekka saw Merja leavintg.' E-adverbial Pekka nukahti yleensä {PRO lukien kirjoja} Pekka fell.asleep often read.e books 'Pekka fell asleep often by reading books.' VA-participle adjective phrase The rules are: 1. There is never agreement on the verb/predicate, Pekka näki {luuta PRO syövän} koiran Pekka saw bone.par eat.va.0 dog 'Pekka saw a dog that was eating/ate the bone.' 2. An overt argument cannot replace the null subject, 3. The null subject is obligatory.
Control for PRO Type II PRO null subject C-command Most local antecedent Requires antecedent in all number and person (if it has such) Notes When determining locality/c-command, complement vs adjunct (adverb) position is relevant. Antecedent cannot be located extrasyntactically from the discourse
Control for pro (1) * on aina paikalla ajoissa. is always there in.time??<----- (Antecedent is missing; 3.person NS requires it) (2) Pekka väittää että on aina paikalla ajoissa. Pekka claims that is always there in.time <------------------------------- (C-commanding antecedent found = OK) =Strategy A (3) Pekkaa ei tarvitse muistuttaa tapaaisesta. Pekka does not require remaning.... on aina paikalla ajoissa is always there in.time <---------- (Antecedent is the discourse topic, no c-command = OK) =Strategy B
Interaction between A and B If there is a constituent that attracts both A and B, then that must be the antecedent. Here the matrix subject that is the topic attracts both A and B Pekka väittää että on ajoissa paikalla Pekka claims that is in.time there <topic> <------------A-------------- <------------B-------------- If A and B can converge to different constituents, then the sentence is ambiguous. Mitä tulee Pekkaan(1), joku(2) väitti että (1,2) varasti häneltä kellon. what comes to Pekka somebody claim that stole a watch from him <------------A------------ <-----------------------B----------- (possible but?)
Let's look how it works for TUAadverbial and its pro Mitä tulee Pekkaan, ollaan oltu yhdessä paljon kalassa {pro juoduttuaan työttömäksi} <...topic...> <...impersonal passive...> <...TUA-adverb> what comes to Pekka, have been together must fishing {pro after he got unemployed} *<----------------------A------------------ No antecedent here! <-------------------------------B------------------- Topic found!
I have argued that Finnish has two kinds of null subjects, pro and PRO Their distribution is governed by morphosyntax, either by abundance of morphosyntax (pro) or by the lack thereof (PRO) Their control properties differ.
The big picture I believe that morphosyntax is a reflex of some operation in narrow syntax that affects PF- and LF-visibility of arguments: PRO pro Invisible at PF = (gap) Invisible at LF = reference fixed by control lack of any type of morphosyntax (Case assignment) makes it totally invisible at PF,LF, and no later operation can repair the issue; the verb steals something morphosyntactic from the element that shows up as agreement, making it invisible at PF,LF This is I think close to what was proposed in the GB-theory? I think it can be argued that in Finnish morphosyntax does govern the distribution of null subjects.