Children go Through a Number of Stages As They Acquire Language

Maturation and optional infinitives Wexler, Kenneth (1996). The development of inflection in a biologically based theory of language acquisition. In Mabel Rice (ed.), Toward a genetics of language. Mahwah, NJ: Lawrence Erlbaum Assoc. Reviewing the stages… Children go through a number of ‘stages’ as they acquire language: • Preliminary stage (0-1;0) (prelinguistic development, babbling) • Single word stage (1;0–1;6) • Two word stage (1;6–2;0) (simple sentences, vocabulary explosion) • Simple sentences • Complete sentences (embedding) The interaction between inflection and syntax. Inflection: tense marking, agreement; syntax: word order French: (1) Mary embrasse Jean. Mary kisses Jean (2) a. b. c. d. (3) French: Two-word stage (1;6–2;0) or so This is where kids start to produce multi-word utterances. First evidence of syntax. (doesn’t mean it was there before, but one word can’t give us evidence) Tremendous vocabulary explosion. Pinker (1994) [citing Brown 1973] notes that at this point, it seems like the two-word stage is a “production bottleneck”—all of the “semantic roles” are represented, and in the right order. Vfinite pas … pas Vnonfinite * Vnonfinite pas * pas Vfinite a. b. pattern of adult French Finite verbs precede pas. Nonfinite verbs follow pas. Evidence that morphology and syntax are deeply inter-related in the adult language. A little bit of syntax: AGENT ACTION RECIPIENT OBJECT LOCATION Mother Mommy Mommy Baby Give gave fix John lunch in the kitchen. I Tractor Adam (4) IP qp NP I′ subject qp I VP tense qp agr’t V NP verb object pumpkin table doggie Put Put ride go Give Put put light floor horsie doggie paper truck it floor I (INFL) (inflection) is a functional head. window box (5) Two-word stage? • West Greenlandic (polysynthetic) looks like it goes to something more like a “two morpheme” stage apparently. a. b. Mary [past,3sg] buy the car → Mary [bought] the car. Mary [pres,3sg] buy the car → Mary buys the car. (A little bit like, e.g., French: à + le → au, de + le → du, à + les → aux…) (6) IP qp NP I′ subject qp I NegP tense qp agr’t Neg VP not qp pas V NP verb object Checking: (7) The finite verb has endings on it marking tense and agreement The tense and agreement features are in INFL. In order for the structure to be well-formed, these features must be “checked” — the verb and INFL have to get together to check that they have the same features. IP qp I′ NP subject qp I+V NegP " qp 1 Neg VP 1 not qp 1 pas tV NP z----------m object Conclusion: Children know about tense, and they know what effect tense has on word order. What’s different between kids and adults seems to be that they don’t always produce finite verbs, but when they produce a nonfinite verb, it is treated like a nonfinite verb would be in the adult language. A similar experiment was carried out for German: [+finite] 197 11 V2 V-final [–finite] 6 37 So, kids know a complex fact about the adult language, the correlation between finiteness and verb position (explained for the adults in terms of the checking story). But adults only allow nonfinite verbs in embedded sentences— kids seem to allow main clauses to have nonfinite verbs. (10) (11) a. pas tomber bebe not fall baby b. pas attraper une fleur not catch a flower a. ich der Fos hab’n I the frog have ‘I have the frog.’ b. Thorstn das haben T. that have ‘Thorstrn have that.’ So, in adult French finite verbs we have: (8) subject V+Infl neg object For nonfinite verbs, we can suppose that there are no tense features, hence the verb does not need to check against Infl, resulting in: (9) subject (Infl) neg V object (12) Back to children and knowledge of syntax: French kids produce negative sentences both in the verb pas and pas verb orders, but… pas verb verb pas [+finite] 11 185 In fact, this kind of thing is found in many languages—a stage, around 2;0, where kids do the following: [–finite] 77 2 The optional infinitive stage: a. There are main clauses with finite verbs. b. There are main clauses with nonfinite verbs. c. Children move the verb as required in the adult language for each type of verb. Other languages for which there is evidence of an OI stage: All Germanic languages studied to date (Danish, Dutch, English, Faroese, Icelandic, Norwegian, Swedish, Faroese). Also, French, Irish, Russian, Brazilian Portuguese, Czech, Hebrew. Very Early Parameter Setting. From the earliest observable stage, that is, from the time that children produce multi-word utterances, they have correctly set the basic inflectional/clause structure parameters. These include the V-to-TNS (verb raising) parameter, the V2 parameter, and parameters of basic word order , including the relative order of the verb and the object. Almost no examples where a kid can be said to have mis-set a parameter (Swiss German, embedded V2 in kids but adult V-final?) English: We expect to see things like: (14) She like me. She go to the store. a. b. vs. She likes me. vs. She goes to the store. (In English, the nonfinite verb is just the stem; (to) like, (to) go.) (cf. French parl+er = parler ‘to speak’, German sprech+en = sprechen ‘to speak’) So, lots of studies of English L1 acquisition failed to notice that kids were using infinitives; it just seemed like kids were slowly acquiring the -s ending in 3sg present. Note: This means kids aren’t learning this via some kind of positive reinforcement when they say a sentence right—they haven’t said anything correctable by the time they show evidence that they have it right. Agreement goes with finiteness—if the kid shows agreement, the verb is finite. So we expect that kids won’t say: A way to describe the OI stage: (15) a. b. * I goes to the store. * You likes me. Tense optionality: Tense is optional. (Everything else [UG, language-particular grammar] is known to the child) (16) a. b. c. d. c. * * * * * This seems to be true of kids around 2;0, but it goes away soon after, and it is not true of adults. Maturation. The hypothesis put forward by Wexler is that this is an example of maturation, analogous to losing one’s baby teeth and growing new ones. The part of grammar that requires tense in an adult utterance is not in place yet in the child’s grammar, though pretty much the rest of it is. (13) Mary wanted [ e to leave]. wanting happens before leaving; the temporal interpretation of to leave depends on the context (main clause tense). Perhaps for adults the reason you can’t have nonfinite main clauses is that there is no higher tense to be contextually based on. Perhaps kids have more permissive conditions on anchoring tense, allowing it to be filled in from the discourse context as well. I is going to the store. She am here. He are here. I does not like ice cream. You does not belong here. English do is generally believed to be added to sentences to “support stranded tense.” (17) a. b. c. She TNS go → She goes. She TNS not go → She does not go. * She not goes So if do depends on tense, we expect not to see do if tense is missing, if the verb is nonfinite. So we expect not to find cases like (18) a. b. (19) a. b. * She do not go. * She not goes. She does not go. She not go. So, it’s this pragmatic principle which “matures” for kids around 2;5. This is all essentially as predicted. [Modal drop explanation? English can’t use dummy did except as last resort, an odd thing to drop; Faroese seems not to even have a dummy modal. Plus, the sentences don’t appear to have a modal meaning.] It could also be argued that be is inserted to support tense, like do is: (20) a. b. She is eating. She is quiet. (be is not contributing anything to the meaning here—plus not all languages even need be for such sentences, being perfectly content with something akin to She quiets). If be depends on tense like do does, we have predictions: (21) a. b. She eating. She quiet. This is very common with kids. In many languages it is possible to leave the subject out if it is understood. Often this has to do with the fact that the verb is marked with the person of the subject. These languages are “null subject” languages (includes Italian). Kids seem to produce a lot of null subjects even in non-null subject languages— But in general, null subjects are allowed even in non-null subject languages in embedded nonfinite clauses: (27) a. b. I want [e to leave]. I want [Bill to leave]. We do not expect to see: (22) a. * She be going. So we might suppose that the null subjects kids are saying happen when they leave out Tense, meaning that they’re grammatically allowed to leave out the subject. [There is a main verb be too; Wexler’s example: You be a cowboy which means you (should) act like a cowboy, not a statement that you are a cowboy.] Turns out: In English: be and have act more like French verbs—they do move to I. (23) a. b. c. She is not going. * She not is going. * She did not be going. But this is dependent on there being tense in INFL, so we don’t expect to see something like: (24) a. a. b. Do you have a quarter? Is she here? But the verb shouldn’t even be in INFL if there was no tense; correspondingly, we don’t expect to see questions like: (26) a. b. c. This doesn’t seem to be a mis-set parameter—the fact that there is a difference in proportion between finite & non-finite suggests that they know that they are speaking a non-null subject language. Specific Language Impairment (SLI). SLI comes in a variety of forms, but there is at least a common type that Rice & Wexler have been investigating for a while. * She be not going. In yes-no questions, INFL is moved to the front of the sentence: (25) In non-null subject languages, kid null subjects are much more frequent with nonfinite verbs than with finite verbs. But, there are still a number of null subjects with finite verbs too. * Do shehave a cake? * Be she here? * Be she going? Null subjects— It appears that (this type of) SLI can be characterized as an “extended optional infinitive” stage, showing many of the same characteristics as the OI stage in 2-year-olds (but with SLI kids, the sentences are longer). This might tend to further bolster the maturation argument (with abnormal development and individual differences in timings of other maturational processes in mind). Universality: It appears that, although many languages have shown OI’s, not all languages seem to. Italian, for example, does not show OI’s. However, it also seems to correlate with whether a language is a null-subject language or not. The Null Subject/Optional Infinitive generalization. Some such languages: Italian, Spanish, Catalan, Tamil, Polish. Hebrew: null subject in 1st and 2nd person, non-present only. % of optional infinitives, Hebrew: 1st/2nd past/fut null subject 0.6% (1/171) overt subject 1.4% (1/72) all else 25% (85/337) 0.6% (3/530) Basically no OI’s in contexts where the null subject is possible. Basically all the OI’s are where a null subject is not allowed, but most have an ungrammatical null subject (OI licenses null subjects). At this point, the technical proposal of how exactly this correlation comes about might be beyond us, but if this correlation is real, the syntacticians and acquisitionists must try to find a theory that predicts it. Korean. Adult Korean has two ways to negate a sentence—we’ll concentrate on the first. (28) a. Chelswu-ka chayk-ul an-ilk-ess-ta. Chelswu-NOM book-ACC NEG-read-PAST-DECL ‘Chelswu did not read the book.’ b. Chelswu-ka chayk-ul ilk-ci anh-ass-ta. Chelswu-NOM book-ACC read-CI NEG.do-PAST-DECL ‘Chelswu did not read the book.’ Kids don’t tend to show the second kind until later (3;5ish). Kids often make a mistake with the first kind, though: (29) a. ‡ na an pap mek-e I NEG rice eat-DECL (‘I do not eat rice’) (C&H:34 [2;2-6]) They put the object in between an and the verb. This happens right around age 2;0. There is no other real evidence of OI’s in Korean, but this might be a reflex of the same thing. (To explain how it is a reflex of the same thing would be difficult at this stage, since we have tied the OI stage to tense—to explain it, we want to find the reason behind why tense is left out and show that this reason applies to this case too). (30) Na-nun I-topic pap-ul an t rice-acc neg 1 " 1 z--------------m mekessta ate Where adults move the object over negation (something like how French-speaking adults move the verb over negation). Kids, however, in the same internal situations that would cause an OI in French, will sometimes not move the object, resulting in the na an pap meke type word order.