GRS LX 700 Language Acquisition n Recall the basic structure and of adult sentences. n IP (a.k.a. TP, INFLP, …) is the position of modals Linguistic Theory and auxiliaries, also assumed to be home of Week 2. The emergence of syntax tense and agreement. n CP is where wh-words move and where I moves in -aux-inversion

Splitting the INFL Splitting the INFL n Syntax since 1986 has n Distinct syntactic been more or less functions assigned to driven by the distinct functional heads. principle “every n T: tense/modality n AgrO: agreement, separable functional accusative case element belongs in its n AgrS: subject agreement, own .” nominative case n Neg: negation n Various syntactic tests support these moves n Origins: Pollock (1989) as well (cf. CAS LX (split INFL into Agr and T), 523). Chomsky (1993) (split INFL into AgrS, T, AgrO).

Functional heads Functional heads n The DP, CP, and VP n VP was split into two parts, vP where the agent all suffered a similar starts, and VP where the fate. patient starts. V and v combine by head movement. n DP was split into DP n Origins: Larson (1988) and NumP proposed a similar structure for double-object , Hale & Keyser (1993) proposed something like n Origin: Ritter 1991 and this structure, which was related work adopted by Chomsky (1993).

1 Functional heads Functional structure n CP was split into n Often, the “fine structure” n The heart of “syntax” several “discourse- of the functional heads is really in the related” functional does not matter, so functional heads, on people will still refer to heads as well (topic, “IP” (with the this view. Verbs and focus, force, and understanding that under give us the “finiteness”). a microscope it is lexical content, but probably AgrSP, TP, functional heads (TP, AgrOP, or even more n Origins: Rizzi (1997) AgrSP, etc.) give us complex), “CP”, “DP”, the syntactic structure. etc.

How do kids get there? Testing for functional structure n Given the n Kids learn it (patterns of input). n Trying to answer this n It’s not very easy. It’s structure of adult n Chickens and eggs, and creoles, and so hard to ask forth. question involves sentences, the trying to determine judgments of kids, question we’re what evidence we and they often do concerned about n Option 1: Kids start out assuming unhelpful things like the entire adult structure, learning have for these here will be in repeat (or garble) large part: how do just the details (Does the move? functional structures How is tense pronounced?) things they just heard kids (consistently) in child syntax. (probably telling us arrive at this n Option 2: Kids start out assuming nothing about what structure (when their grammar they become some subpart of the adult structure, complexity increasing with actually is). adults)? (predetermined?) development.

Testing for functional structure Helpful clues kids give us n We do know n This isn’t foolproof. If a child n Null subjects n Root fails to pronounce the past what various n Kids seem to drop the n Kids seem to use functional tense suffix on a verb that was subject off of their nonfinite forms of clearly intended to be in the sentences a lot. More main (root) projections are past, does this mean there’s than adults would. supposed to be verbs where adults no TP? Does it mean they There’s a certain wouldn’t. Again, crosslinguistic responsible for, simply made a speech error there’s a certain and so we can (as adults sometimes do)? systematicity to it as well, from which we crosslinguistic look for evidence Does it mean they haven’t systematicity to it that figured out how to pronounce might take hints about of their effects in kids’ functional can provide clues as to the past tense affix yet? child language. structure. what’s going on.

2 Radford (1990, 1995) adult syntax ≠ child syntax n A proposal about Early Child English. n Adults: CP—IP—VP n Kids’ syntax differs from adults’ syntax: n Kids: VP n kids use only lexical (not functional) elements n structural sisters in kids’ trees always have a q- relation between them. n Evidence for absence of IP: VP n No modals (repeating, kids drop them) “Small Clause NP q V’ Hypothesis” n No auxiliaries (Mommy doing dinner) man n No productive use of tense & agreement (Baby V q NP ride truck, Mommy go, Daddy sleep) chase car

Absence of CP Absence of DP n No CP system: n No DP system: n no (that, for, if) n no non-q elements n no expletives (raining, outside cold) n no preposed auxiliary (car go?) n no of before complements of nouns (cup tea) n no wh-movement (imitating where does it go? n Few (Hayley draw boat, want duck, yields go?; spontaneous: mouse doing?) reading book) n kids bad at comprehending wh-object n No ’s, which may be a D. questions (out of canonical order). (—What are n No , which are probably D. you doing? —No.)

Small children’s small To T or not to T n The Small Clause Hypothesis is not prima facie n Focusing specifically on tense (and subject crazy. Child English does seem to look agreement), the fact that kids sometimes use something like what it would predict. tense and sometimes do not does not indicate n On the other hand, when looking across that they know or represent T in their syntactic languages, we find that the SCH doesn’t fare structure. very well. n In languages where tense/agreement is more n The question is: When tense is there, does it act visible, we find kids using infinitives, but only like tense would for an adult? Do kids sometimes, other times using finite verbs. The differentiate between tensed and verbs, case that kids do not represent tense weakens or are these just memorized Vs at this point? (but is not yet out of the running!).

3 Full Competence Hypothesis Adult German

n Poeppel & Wexler (1993). Data: Andreas (2;1, n Phrase structure consists of CP, IP, VP. from CHILDES). n German is SOV, V2 n The morphosyntactic properties associated with finiteness and attributable to the availability of n The (or auxiliary or modal) is the functional categories (notably head movement) second constituent in main clauses, following are in place. some constituent (subject, object, or n The best model of the child data is the standard adverbial). analysis of adult German (functional projections n In embedded clauses, the finite verb is final. and all). The one exception: n V2 comes about by moving the finite verb to n Grammatical Infinitive Hypothesis: (head-initial) C. n Matrix sentences with (clause-final) infinitives are a legitimate structure in child German grammar.

German clause structure German clause structure CP CP

DP C¢ n This “second position” is DP C¢ n Things other than generally thought to be C, subjects can appear in C+I IP where something else C+I IP “first position”. hat Hans kaufte (like the subject, or any den n When the tense appears I¢ DP I¢ — other XP) needs to appear Ball on an auxiliary, the verb in SpecCP. VP — Hans VP stays in place. n This only happens with V¢ finite verbs. Nonfinite V¢ verbs remain at the end DP — of the sentence (after the — V object). gekaufte den Ball

In brief… Results

n Kids can choose a finite or a nonfinite verb. n There is a strong contingency. n A finite (matrix) verb shows up in 2nd position n Conclude: the finiteness distinction is made n A nonfinite verb appears clause-finally correctly at the earliest observable stage.

ich mach das nich I do that not +finite -finite V2, not final 197 6 du das haben V final, not V2 11 37 you that have

4 Do kids learn “this is a second Verb positioning = position verb” for certain verbs? functional categories n In adult German, V2 comes n (Are some verbs used as auxiliaries?) from V Æ I Æ C. n If we can see non-subjects n Andreas used 33 finite verbs and 37 nonfinite to the left of finite verbs, we verbs, 8 of which were in both categories— know we have at least one FP n —and those 8 were finite in V2 position and functional projection (above Object F¢ nonfinite in final position. the subject, in whose Spec the first position non- F+V VP Subject n Remaining verbs show no clear semantic core subject goes). V¢ that one might attribute the distribution to. — —

When V is 2nd, what’s first? CP

n Usually subject, not a big surprise. n The Full Competence Hypothesis says not only that functional categories exist, but that the child n But 19 objects before finite V2 has access to the same functional categories that (of 197 cases, 180 with overt subjects) the adult does. n And 31 before finite V2 n In particular, CP should be there too. n Predicts what we’ve seen: n finite verbs are in second position only n Conclude: Kids basically seem to be acting (modulo topic drop leaving them in first position) like adults; their V2 is the same V2 that n nonfinite verbs are in final position only adults use. n subjects, objects, adverbs may all precede a finite verb in second position.

P&W’s predictions met—how P&W’s predictions met—how did the other guys fare? did the other guys fare? n Radford and related approaches (No functional n “No C hypothesis” (kids don’t use overt categories for the young)? complementizers) n Well, we see V2 with finite verbs n finite verb is second n Of course, kids don’t really use embedded n non-subjects can be first clauses either (a chicken-egg problem?) n and you can’t do this except to move V out of VP n Purported cases of embedded clauses without a and something else to its left… aren’t numerous or convincing. n You need at least one functional category. n Andreas uses agreement correctly when he uses it—adults use IP for that. n Absence of evidence ≠ evidence of absence.

5 P&W’s predictions met—how P&W’s predictions met—how did the other guys fare? did the other guys fare? n Can we get away with only one functional n Empirical argument: category? n negation and adverbs are standardly supposed to mark n The word order seems to be generable this way the left edge of VP. so long as F is to the left of VP. n A subject in SpecVP (i.e. when a non-subject is topicalized) should occur to the right of such n subject can stay in SpecVP elements. n V moves to F n 19 Object-initial sentences 31 -initial n non-subject could move to SpecFP. sentences, 8 have an(other) adverb or negation, n …though people tend to believe that IP in and all eight have the subject to the left of the German is head-final (that is, German is head- adverb/negation. final except for CP). How do kids learn to put I n [ Object C+I+V [ Subject [ neg/adv t t ] t ]] on the right once they develop CP? CP IP VP Subj V I

The Full Competence Harris & Wexler (1996) Hypothesis n The idea: Kids have full knowledge of the n Child English bare stems as “OIs”? principles and processes and constraints n In the present, only morphology is 3sg -s. n Bare stem isn’t unambiguously an infinitive form. of grammar. Their representations are n No word order correlate to finiteness. basically adult-like. n OIs are clearer in better inflected languages. Does English do this too? Or is it different? n Hypotheses: n What’s different is that kids optionally n Kids don’t “get” yet; go and goes are allow infinitives as matrix verbs (which basically homonyms. kids grow out of). n These are OIs, the -s is correlated with something systematic about the child syntax.

Harris & Wexler (1996) Harris & Wexler (1996) n Exploring a consequence of having T in n Empirically, we expect: the structure: do support. n She go n She goes n Rationale: n She not go (no T no do) n Main verbs do not move in English. n She doesn’t go (adult, T and do) n Without a modal or auxiliary, T is stranded: The verb -ed not move. n but never n She not goes (evidence of T, yet no do). n Do is inserted to save T. n Note: All basically options if kids don’t n Predicts: No T, no do insertion. “get” inflection.

6 Harris & Wexler (1996) Harris & Wexler (1996)

n Looked at 10 kids from 1;6 to 4;1 n Affirmative: n 43% inflected aff neg n Adam, Eve, Sara (Brown), Nina (Suppes), Abe (Kuczaj), Naomi (Sachs), Shem (Clark), April n Negative: (Higginson), Nathaniel (Snow). n < 10% inflected -inflec 782 47 n Counted sentences… n It not works Mom n with no or not before the verb n no N. has a microphone n without a modal/auxiliary n no goes in there +inflec 594 5 n with unambiguous 3sg subjects n but the horse not stand ups n with either -s or -ed as inflected. n no goes here!

Harris & Wexler (1996) Harris & Wexler (1996)

n Small numbers, but in the right direction. n Also, made an attempt to ascertain how the form n Generalization: Considering cases with no correlated with the intended meaning in terms auxiliary, kids inflect about half the time of tense. (Note: a nontrivial margin of error…) normally, but almost never (up to performance n Inflected verbs overwhelmingly in the right context. errors) inflect in the negative. n If do is an indicator of T in the negative, we might expect to see that do appears in negatives present past future about as often as inflection appears in bare stem 771 128 39 affirmatives. -s 418 14 5 n Also, basically true: 37% vs. 34% in the pre-2;6 group, 73% vs. 61% in the post-2;6 group. -ed 10 168 0

Harris & Wexler (1996) Harris & Wexler (1996) n Last, an elicitation experiment contrasting affirmative, never (no T dependence for adults), and not. n Affirmatives inflected often, not inflected rarely, never sort of inbetween. n Does the cow always go in the barn, or does she never go? n n Does the cow go in the barn or does she not go in the barn? Looking at the results in terms of whether the n Do you think he always goes or do you think he never goes? question was inflected: n Do you think that he goes, or don’t you think that he goes? n Kids overall tended to use inflection when there was inflection in the question. n n Processing load? Extra load of not alleviated by When the stimulus contained an -s: leaving off the -s? If that’s the case, we’d expect never n affirmative: 15 vs. 7 (68% had an -s) and not to behave the same way—in fact, never might n never: 14 vs. 16 (48%) be harder, just because it’s longer (and trigger more -s n not: 4 vs. 12 (25%) —quite a bit lower. drops).

7 Some alternatives… Modal drop n Root infinitives due to “modal drop”? n n Idea: I want to eat pizza. Adult modals are in position 2, regardless of what is in position 1. n RI? I want to eat pizza. n n First question: why modals? If kids are dropping modals, we should expect a certain proportion of the n Second, they don’t (always) seem to mean dropped modals to appear with a non- what they should if there is a null modal. subject in position 1. 20/37 seem to be clearly non-modal. n But none occur—nonfinite verbs also n Thorsten Ball haben (T already has the ball) seem to come with initial subjects.

Modal drop Modal drop n On the other hand, if nonfinite final V n Just to be sure (since the numbers are small), indicates failure to raise to I and C, we P&W check to make sure they would have expected non-subjects in position 1 with nonfinite don’t expect CP to be available for verbs if the modal drop hypothesis were true. “topicalization” (the assumption is that V2 n 17% of the verbs are infinitives involves both movement of V to C and n 20% of the (finite) time we had non-subject movement of something else to SpecCP; topicalization n So 3% of the time (20% of 17%) we would expect non- but no need to move something to SpecCP subject topicalization in nonfinite contexts. unless V is in C). n Of 251 sentences, we would have expected 8. n We saw none.

Subject case errors Finiteness vs. case errors n Various people have observed that kids Schütze & Wexler Loeb & Leonard learning English sometimes will use (1996) (1991) accusative subjects. Nina 7 representative kids 1;11-2;6 2;11-3;4 subject Finite Nonfinite Finite Nonfinite n It turns out that there’s a sort of a correlation with the finiteness of the verb he+she 255 139 436 75 as well. Finite verbs go with nominative him+her 14 120 4 28 case, while nonfinite verbs seem to go % non-Nom 5% 46% 0.9% 27% with either nominative or accusative case.

8 What to make of the case EPP and missing INFL errors? n If there were just an IP, responsible for both n Case is assumed to be the n Why are non-AgrSP NOM and tense, then they should go together (cf. jurisdiction of AgrSP and subjects accusatives? “IP grammar” vs. “VP grammar”) AgrOP. n Probably a default case in English: n n So, nominative case can Yet, there are many cases of root infinitives with n Who’s driving? Me. Me too. NOM subjects serve as an unambiguous It’s me. signal that there is an n n Other languages seem And, even ACC subjects seem to raise out of the AgrSP. VP over negation (me not go). not to show this n Accusative case, “accusative subject” error n We can understand this once we consider IP to conversely, may signal a but also seem to have a be split into TP and AgrP; tense and case are missing AgrSP. nominative default separated, but even one will still pull the subject (making an error up out of VP. (ATOM:+Agr –Tns) undetectable).

Agr/T Omission Model “ATOM” (ATOM) n Schütze & Wexler n For a finite verb, both TP and propose a model of AgrSP must be there. English n Adult clause structure: this in which the inflection (3sg present –s) relies case errors are a on both. If one or the other is AgrP result of being able missing, we’ll see an infinitive to either omit (i.e. bare stem). AgrSP or Tense. NOMi Agr¢ n Thus, predicted: finite n For a subject to be (AgrSP+TP) verbs show Nom Agr TP in nominative case, (AgrSP), but only half of the AgrSP must be nonfinite verbs (not both AgrSP ti T ¢ there (TP’s presence and TP) show Nom (AgrSP). is irrelevant). We should not see finite+Acc. T VP

ATOM ATOM

n Kiddie clause, missing TP (—TNS): n Kiddie clause, missing AgrP (—AGR):

AgrP

NOMi Agr¢ Agr TP

ACC fi defaulti T ¢

VP T VP

9 Pronunciation of English One prediction of ATOM n T+AgrS(+V) is n Layers of “default”, most n +AGR+TNS: NOM with inflected verb (-s) pronounced like: specific first, followed by next most specific n +AGR–TNS: NOM with bare verb (“Distributed n n /s/ if we have Morphology”, Halle & –AGR+TNS: default (ACC) with bare verb features Marantz 1993). n –AGR–TNS: GEN with bare verb [3, sg, present] n Notice: 3sg present –s (the GEN case was not discussed by Wexler 1998, requires both TP and but see Schütze & Wexler 1996) n /ed/ if we have the AgrSP, but past –ed feature [past] requires only TP (AgrSP might be missing, so we n Nothing predicts Acc with inflected verb. might expect some n Ø otherwise accusative subjects of past tense verbs).

Finite pretty much always goes ATOM and morphology with a nominative subject.

n [+3sg +pres] = -s n Schütze & Wexler Loeb & Leonard [+masc, +3sg, +nom] = he (1996) (1991) n [+past] = -ed n [+masc, +3sg, +gen] = his Nina 7 representative kids n — = Ø n [+masc, +3sg] = him 1;11-2;6 2;11-3;4 n [+fem, +3sg, +nom] = she n [+masc +3sg +nom] subject Finite Nonfinite Finite Nonfinite n [+fem, +3sg] = her play+[3sg+pres] he+she 255 139 436 75 n he plays. n [+1sg, +nom] = I n [+2sg +nom] n [+1sg, +gen] = my him+her 14 20 4 28 play+[2sg +past] n [+1sg] = me n you play. % non-Nom 5% 46% 0.9% 27% n [+2, +gen] = your n [+2] = you n But is this knowledge built-in? Hint: no.

ATOM and morphology Schütze (2001, inter alia) n What if the child n ATOM predicts that n No. n If the kid thinks her is the produces a lot of agreement and n Her goes to school is not nominative feminine 3sg utterances like nominative case necessarily a , her goes to school counterexample to n her sleeping should correlate. is perfectly consistent ATOM (although it is a with ATOM. n her play n Her goes to school is candidate). n and even predicted never to n Morphology must be n Hence, we should really n her sleeps occur. learned and is crosslinguistically only count her+agr n her goes to school n So does this child’s variable. correlations from kids n but never uses the use of her goes to school n She is known to emerge who have demonstrated word she? mean ATOM is rather late compared to that they know she. wrong? other pronouns.

10 ATOM and morphology Rispoli (2002, inter alia) n Morphology (under n [+masc, +3sg, +nom] = he n Rispoli has his own n The kid’s success in “Distributed Morphology”) is theory of her-errors. finding the form is a system of defaults. n [+masc, +3sg, +gen] = his n Pronoun morphology is affected by “gravity”. n The most specified form n [+masc, +3sg] = him possible is used. organized into “tables” “Heavier” forms are n more likely to be picked n Adult English specifies her as a [+fem, +3sg, +nom] = she (paradigms) basically, when accessing the table, feminine 3sg pronoun, and she n [+fem, +3sg] = her where each form has a as a nominative feminine 3sg certain weight. even if it’s not quite the pronoun. n [+1sg, +nom] = I right form. If it’s close n When a kid is trying to n If the kid doesn’t know she, the n [+1sg, +gen] = my pronounce a pronoun, and it’s heavy, it’ll win result will be that all feminine out a lot of the time. 3sg pronouns will come out as n [+1sg] = me s/he attempts to find the n Her by virtue of being her. That’s just how you n [+2, +gen] = your entry in the table and pronounce nominative 3sg pronounce it. both acc and gen is extra- feminine, if you’re the kid. n [+2] = you heavy, and pulls the kid n Just like adult you. in fairly often.

Her plays Monotonicity n ATOM and Rispoli make n Rispoli’s complaints n Schütze assumes that use n Rispoli (2002) set out different predictions with about Schütze’s studies: of she is a matter of to show that there is a respect to her plays. n Excluding kids who knowledge of she. Once the n ATOM says it should happen not to produce kid knows it, and given certain amount of never happen (up to she in the transcript under that the adult version of “yo-yo’ing” in the simple performance evaluation is not good the kid will know it, it’s production of she. error) enough. The assumption there, for good. n Rispoli says case errors is that this learning is n Rispoli claims that the are independent of monotonic, so if the kid “weight” of she can n We’ll focus on Nina, ever used she fluctuate, so that it could agreement, her plays is for whom we can get perfectly possible, even (productively) in the past, be “known” but mis- expected. the her errors should not retrieved later if her the data. be excluded. becomes too heavy.

Nina she vs. her Checking Rispoli’s counts n Rispoli’s counts show she her n 2;2 n These are the Nina using she from n *CHI: she have hug a lady . times when Nina 2;2 2 43 basically the outset of n *CHI: she have jamas@f on . used she (twice at 13-15 4% 96% her use of pronouns, n 2;3 2;2, once at 2;3, and also shows a 2;3 1 12 n *MOT: does she like it ? once at 2;4). decrease of use of she 16-19 8% 92% n *CHI: she drink apple juice . n *CHI: her like apple juice . at 2;5. 2;4 1 6 n 2;4 n Rispoli found 7 at 20-23 14% 86% n *MOT: he's up there ? 2;5, we’ll deal 2;5 7 73 n *CHI: no # she's not up there . with them later. n *CHI: he's up there . 24-31 9% 91%

11 Checking Checking n 2;2 n These three and one other time n *MOT: what happened when I n 2;5: n *CHI: helping her have a Nina said her’s ok are the only shampooed Miriam yesterday ? yellow blanket . candidate counterexamples at n *CHI: her was cried . n I found about 76 n *MOT: she has a yellow 2;2. her+bare/past blanket ? n At 2;2, 45 her+bare verb. n *MOT: oh # there's the dolly's bottle . verbs. n *CHI: yeah [= yes] . n (R got 43, possibly including n *CHI: her's not going to drink it . n *CHI: her's ok . her’s ok) n I found 3 n *CHI: her ok . n At 2;3, no candidate n *MOT: I'll start washing it . potential n *MOT: she's ok ? counterexamples, 14 her+bare n *MOT: see how clean it comes ? counterexamples. n *CHI: ok . verbs. n *MOT: you want to use the pot ? n *CHI: her's ok . n (R got 12) n *CHI: a little bit . n *CHI: her ok . n At 2;4 none, 7 her+bare. n *CHI: her don't . n *CHI: her's ok . n (R got 6) n *CHI: her's not dirty . n *MOT: she's ok . n *CHI: not dirty .

Two hypotheses about Bottom line? learning n It doesn’t seem like n The point about variation in anything was usage of she is valid, worth n VEPS (very early parameter setting) particularly affected, being aware of the Basic parameters are set correctly at the even if Nina’s early assumptions and being sure earliest observable stages, that is, at least files were fully we’re testing the right things. from the time that the child enters the two- included. n Rispoli was trying to make word stage around 18 months of age. the point that if we’d n The number of accidentally missed a she in n VEKI (very early knowledge of inflection) possible the early files, we might have counterexamples excluded counterexamples At the earliest observable stage (two-word seems well within the there. Yet, even including stage), the child knows the grammatical and “performance error” everything, the asymmetry is phonological properties of many important range. strong. inflectional elements of their language.

Two-word stage? Very Early Parameter Setting n The reason both VEPS and VEKI mention n As soon as you can see it, kids have: the two-word stage is just because this is n VO vs. OV order set (Swedish vs. German) the first stage where we have evidence of n VÆ>I [yes/no] (French vs. English) utterance composition. n V2 [yes/no] (German vs. French/English) n Null subject [yes/no] (Italian vs. Fr./E.)

n So, at least by the 2-word stage, they have the parameters set (maybe earlier)

12 VEKI? Ok, but… n Generally, when kids use inflection, they use n So: Kids have the full functional it correctly. Mismatches are vanishingly structure available to them, and they set rare. the parameters right away and know n English (Harris & Wexler 1995) the inflection. n German (Poeppel & Wexler 1993) n What then do we make of the fact that kids make non-adult utterances in the n Again, this is kind of contrary to what the face of evidence that they aren’t learning field had been assuming (which was: kids the parameters? are slow at, bad at, learning inflection). n KW: Certain (very specific, it turns out) properties of the grammar mature.

Root infinitives vs. time NS/OI n The timing on root n But some languages appear not to infinitives is pretty undergo the “optional infinitive” stage. robust, ending How can this be consistent with a around 3 years old. maturational view? n OI languages: studied to date (Danish, Dutch, English, Faroese, Icelandic, Norwegian, Swedish), Irish, Russian, Brazilian Portuguese, Czech n Non-OI languages: Italian, Spanish, Catalan, Tamil, Polish

NS/OI and Hebrew NS/OI (Rhee & Wexler 1995) n What differentiates the OI and non-OI n Hebrew is a NS language but only in 1st languages? and 2nd person, non-present tense. n Agreement? Italian (non-OI) has rich agreement, Everywhere else (3rd past, future, present) but so does Icelandic (OI). subjects are obligatory. n Null subjects! n Hebrew-learning 2-year-olds showed n Null Subject/OI Generalization: optional infinitives except in 1/2-past, and Children in a language go through an OI stage allowed null subjects elsewhere, with iff the language is not an INFL-licensed null subject language. infinitives.

13 NS/OI and Hebrew Implementing ATOM (Rhee & Wexler 1995) % of RIs n The basic idea: In adult clauses, the subject needs to move both to SpecTP and (then) to kids up to 1;11 1/2 past/fut (NS) else (non-NS) SpecAgrP. null subjects 0 (of 21) 32% (36/112) n overt subjects 0 (of 6) 0 (of 28) This needs to happen because T “needs” something in its specifier (≈EPP) and so does Agr. all OI kids 1/2 past/fut (NS) else (non-NS) n The subject DP can “solve the problem” null subjects 0.6% (1/171) 25% (85/337) for both T and for Agr—for an adult. overt subjects 1.4% (1/72) 0.6% (3/530)

Implementing ATOM Implementing ATOM n Implementation: For adults: n Implementation: For kids: n T needs a D feature. n Everything is the same except that the subject n Agr needs a D feature. can only solve one problem before quitting. It “loses” its D feature after helping out either T n The subject, happily, has a D feature. or Agr. n The subject moves to SpecTP, takes care of T’s need for a D feature (the subject “checks” the n Kids are constrained by the Unique Checking D feature on T). The T feature loses its need Constraint that says subjects (or their D for a D feature, but the subject still has its D features) can only “check” another feature feature (the subject is still a DP). once. n So the kids are in a bind. n The subject moves on, to take care of Agr.

Implementing ATOM Minimalist terminology n Kids in a pickle: The only options open to the n Features come in two relevant kinds: kids are: interpretable and uninterpretable. n Leave out TP (keep AgrP, the subject can solve Agr’s n Either kind of feature can be involved in a problem alone). Result: nonfinite verb, nom case. “checking”—only interpretable features survive. n Leave out AgrP (keep TP, the subject can solve T’s n problem alone). Result: nonfinite verb, default case. The game is to have no uninterpretable features left at the end. n Violate the UCC (let the subject do both things anyway). Result: finite verb, nom case. n “T needs a D” means “T has an uninterpretable n No matter which way you slice it, the kids have [D] feature” and the subject (with its normally to do something “wrong”. At that point, they interpretable [D] feature) comes along and the choose randomly (but cf. Legendre et al.) two features “check”, the interpretable one survives. UCC=D uninterpretable on subjects?

14 Wait—how can you say kids are Wait—how can you say kids are UG-constrained yet drop T/Agr? UG-constrained yet drop T/Agr? n So, aren’t TP and AgrSP required by UG? n Perhaps what requires TP and AgrP are Doesn’t this mean kids don’t have UG- principles of (pragmatic) interpretation… compliant trees? n You need TP so that your sentence is “anchored” in the discourse. n You need AgrSP … why? Well, perhaps something parallel…? Wexler doesn’t really say… n Actually, perhaps no. UG requires that all n Regardless, kids can check all the features be checked, but it isn’t clear that uninterpretable features even without TP or there is a UG principle that requires a TP AgrSP; hence, they can still be considered to be and an AgrP in every clause. UG-constrained.

Is there any way to see the effects NS/OI via UCC of UCC even in NS languages?

n An old idea about NS languages is that they n Italian: Mary has laughed. arise in languages where Infl is “rich” enough n Suppose that auxiliaries (like have) also have a to identify the subject. D-feature to be checked as the subject (in the adult language) passes through. n Maybe in NS languages, AgrS does not need a D (it may in some sense be nouny enough to n Not crazy: (All) the students (all) have (all) left. say that it is, or already has, D). n UCC-constrained kids will have to drop something (the auxiliary or T), even in Italian. n Lyons (1997) reports that a “substantial n If AgrS does not need a D, the subject is free proportion of auxiliaries are omitted in OI- to check off T’s D-feature and be done. age Italian.” n Ok, maybe. Consistent, anyway.

One open question… Theories of missing structure

n No functional projections. (Radford) Kids n The UCC says you can only use a D-feature don’t have any functional projections (TP, on a DP to check against a functional category CP, and so forth). This comes later. No TP, once. no tense distinction. n This explains why sometimes TP is omitted (keeping AgrSP) and sometimes AgrSP is n Structure building. (Vainikka, Guilfoyle & omitted (keeping TP). Noonan) Kids start with no functional n but if GEN infin. comes from omitting both TP projections and gradually increase their and AgrSP, what could ever cause that functional structure. (particularly given Minimize Violations)?

15 Theories of missing structure Rizzi and truncated trees n “ATOM” (Full competence). (Wexler, …) Kids n The result (of not having CP=root) is that have access to all of the functional structure and kids are allowed to have truncated have a very specific problem with tense and structures—trees that look like adult trees agreement that sometimes causes them to leave with the tops chopped off. one out. n n Truncation. (Rizzi) Like structure building but Importantly: The kids don’t just leave stuff without the time course—kids have access to all out—they just stop the tree “early.” So, if of the functional structure but they don’t realize the kid leaves out a functional projection, that sentences need to be CP’s, so they s/he leaves out all higher XPs as well. sometimes stop early.

Truncation Truncation n If kid selects anything lower than TP as n Pierce (1989) looking at French observed the root, the result is a root that there are almost no root infinitives infinitive—which can be as big as any kind with subject clitics—this is predicted if of XP below TP in the structure. these clitics are instances of subject n Note in particular, though, it can’t be a CP. agreement in AgrS; if there is no TP, there can be no AgrSP. n So: we expect that evidence of CP will correlate with finite verbs.

Truncation Truncation and NegP n There is some dispute in the syntax n But we do find negative Root literature as to whether the position of Infinitives—(Pierce 1989): in the NegP (the projection responsible for the acquisition of French, negation follows negative morpheme) is higher or lower finite verbs and preceds nonfinite verbs than TP in the tree. (that is—French kids know the movement n If NegP is higher than TP, we would properties of finiteness, and thus they expect not to find negative root infinitives. have the concept of finiteness).

16 Truncation and NegP S O Vfin?

n So, is TP higher than NegP? n Usually (Poeppel & Wexler 1993) German kids put finite verbs in second position, and n Hard to say conclusively from the existing leave nonfinite verbs at the end. French data because there are not many negative root infinitives—but further study could lead to a theoretical result of n Occasionally one finds a finite verb at the end. this sort about the adult languages. n Rizzi suggests we could look at this as an instance of a kid choosing AgrSP as root, where CP is necessary to trigger V2.

Legendre et al. (2000) Optimality Theory n Wexler: During OI stage, kids sometimes omit T, and sometimes omit Agr. Based on a choice of n Legendre et al. (2000) is set in the which to violate, the requirement to have T, to have Optimality Theory framework (often seen in Agr, to have only one. phonology, less often seen applied to n (cf. “Kids in a pickle” slide) syntax). n Legendre et al.: Looking at development (of French), it appears that the choice of what to omit is n “Grammar is a system of ranked and systematic; we propose a system to account for violable constraints” (predict) the proportion of the time kids omit T, Agr, both, neither, in progressive stages of development.

Optimality Theory Optimality Theory

n In our analysis, one constraint is Parse-T, n Parse-T and *F are in conflict—it is which says that tense must be realized in a impossible to satisfy both at the same clause. A structure without tense (where time. TP has been omitted, say) will violate this n When constraints conflict, the choice made constraint. (on a language-particular basis) of which constraint is considered to be “more n Another constraint is *F (“Don’t have a important” (more highly ranked) determines functional category”). A structure with TP which constraint is satisfied and which will violate this constraint. must be violated.

17 Optimality Theory Optimality Theory n So if *F >> Parse-T, TP will be omitted. n Grammar involves constraints on the n and if Parse-T >> *F, TP will be included. representations (e.g., SS, LF, PF, or perhaps a combined representation). n The constraints exist in all languages. n Where languages differ is in how important each constraint is with respect to each other constraint.

Optimality Theory: big picture Legendre et al. (2000)

n Proposes a system to predict the n Universal Grammar is the constraints proportions of the time kids choose the that languages must obey. different options among: n Languages differ only in how those n Omit TP constraints are ranked relative to one n Omit AgrSP another. (So, “parameter” = “ranking”) n Omit both TP and AgrSP n The kid’s job is to re-rank constraints n Include both TP and AgrSP (violating UCC) until they match the order which generated the input that s/he hears.

French v. English The idea n English: T+Agr is pronounced like n Kids are subject to conflicting constraints: n /s/ if we have features [3, sg, present] n Parse-T Include a projection for tense n /ed/ if we have the feature [past] n Parse-Agr Include a project for agreement n /Ø/ otherwise n *F Don’t complicate your tree with n French: T+Agr is pronounced like: functional projections n danser NRF n *F2 Don’t complicate your tree so n a dansé (3sg) past much as to have two functional n je danse 1sg (present) projections. n j’ai dansé 1sg past

18 The idea Floating constraints n Sometimes Parse-T beats out *F, and then n The innovation in Legendre et al. (2000) there’s a TP. Or Parse-Agr beats out *F, that gets us off the ground is the idea that and then there’s an AgrP. Or both Parse-T as kids re-rank constraints, the position of and Parse-Agr beat out *F2, and so there’s the constraint in the hierarchy can get both a TP and an AgrP. somewhat fuzzy, such that two positions can overlap. n But what does sometimes mean? *F

Parse-T

Floating constraints Floating constraints

*F *F

Parse-T Parse-T

n When the kid evaluates a form in the n (Under certain assumptions) this constraint system, the position of Parse- predicts that we would see TP in the T is fixed somewhere in the range—and structure 50% of the time, and see winds up sometimes outranking, and structures without TP the other 50% of sometimes outranked by, *F. the time.

French kid data French kid data

n Kids start out using 3sg agreement and n Looked at 3 French kids from CHILDES present tense for practically everything n Broke development into stages based on a (correct or not). modified MLU-type measure based on how long most of their utterances were (2 words, n We took this to be a “default” more than 2 words) and how many of the n (No agreement? Pronounce it as 3sg. No tense? utterances contain verbs. pronounce it as present. Neither? Pronounce n Looked at tense and agreement in each of the it as an infinitive.). three stages represented in the data.

19 French kid data French kids data n This means if a kid uses 3sg or present n We found that tense and agreement tense, we can’t tell if they are really using develop differently—specifically, in the 3sg (they might be) or if they are not using first stage we looked at, kids were using agreement at all and just pronouncing the tense fine, but then in the next stage, they default. got worse as the agreement improved. n So, we looked at non-present tense forms n Middle stage: looks like and non-3sg forms only to avoid the competition between T question of the defaults. and Agr for a single node.

Proportion of non-present and A detail about counting non-3sg verbs n We counted non-3sg and non-present verbs. 40% n In order to see how close kids’ utterances were to adult’s utterances, we need to know how often adults use non- 35% 3sg and non-present, and then see how close the kids are 30% to matching that level. 25% non-present non-3sg 20% adult non-pres n So, adults use non-present tense around 31% of the 15% adult non-3sg time—so when a kid uses 31% non-present tense, we take that to be “100% success” 10% 5% n In the last stage we looked at, kids were basically right at 0% the “100% success” level for both tense and agreement. 3b 4b 4c

Proportion of non-finite root A model to predict the forms percentages

35% n Stage 3b (first stage) 30% n no agreement 25% n about 1/3 NRFs, 2/3 tensed forms 20% NRFs 15% *F2 *F 10% ParseT 5%

0% 3b 4b 4c ParseA

20 A model to predict the A model to predict the percentages percentages n Stage 4b (second stage) n Stage 4c (third stage) n non-3sg agreement and non-present tense n everything appears to have tense and each about 15% (=about 40% agreeing, 50% agreement (adult-like levels) tensed) n about 20% NRFs *F2 *F ParseT *F2 *F ParseT ParseA ParseA

Predicted vs. observed—tense Predicted vs. observed—agr’t

40% 40% 35% 35% 30% 30% 25% 25% non-present non-3sg 20% 20% predicted non-pres predicted non-3sg 15% 15% 10% 10% 5% 5% 0% 0% 3b 4b 4c 3b 4b 4c

Predicted vs. observed—NRFs T

35% T T 30% T 25% T T 20% NRFs 15% predicted NRFs T T 10%

5% T 0% T 3b 4b 4c

21