Language Acquisition A Refresher on Clausal Syntax

Children become adults Language Acquisition A refresher on clausal syntax, and working with child data • We’re looking at acquisition of syntax. • What is the target knowledge? • How does a child’s knowledge differ from the target? • How does the child’s knowledge develop over time? Syntactic assumptions • Sentences have hierarchical structure. • Built of phrases (NPs,VPs, CPs, etc., ...XPs) • Syntactic theory has been developed over Clauses • A full clause is understood to be a CP (Complementizer Phrase). • The head of the CP is usually something like that, if, or simply Ø. time, but papers are only written and published at a moment in that development. Sometimes labels change, concepts revised. • The features of C determine the “force” of the IP CP—IP—VP • The “IP” area of the clause requires some comment. • This is where tense features (past, non- past), modals (could, would, should), and agreement features (governing subject-verb agreement, for example) are found. • Back in the day, this was basically lumped together as “inflection”—hence “IP”. clause—declarative, interrogative, imperative. • In English wh-questions, the first wh-word moves to SpecCP. • The basic syntactic tree, kind of abstracting away from some of the specific details of different approaches: CP, IP, VP. • Each of these has been subsequently discovered to have a finer structure. But often the finer structure is only highlighted when the fine structure itself is relevant. Sometimes, this is called “IP.” Sometimes it is also called “TP.” Other times, “TP” is used to refer to just a part of what “IP” refers to. CP C ! C that IP DP Pat • French has verb raising. • English does not. • Diagnosed in part by positions of adverbs VP t V! V meet CP and negation. DP Chris V2 CP C ! C! C que IP DP Pat C that I ! IP DP Pat I+V VP mange AdvP VP rapidement t V! t • Languages differ as to whether the verb moves up to the inflectional area of the clause or not: V-to-I (or “V-to-T”) raising. I! I will V-to-I • V2 languages, including German, are taken I! I VP pres AdvP VP quickly t V! DP cette pomme V DP eats yonder apple to involve a couple of further movements, beyond the V-to-I that French has. • One is I-to-C (after V-to-I has already happened, carrying the verb to C). • The other is “topic” to SpecCP (putting C in second position) CP DP eine Fase Splitting INFL C! C+I+V IP hab DP I! ich VP AdvP • At one point, the “IP” was hypothesized to contain three different projections (or, four if you count NegP): AgrSP—TP—AgrOP. t • This isn’t usually relevant, but it will be VP t relevant when we discuss the “A/TOM” model of root infinitives in child speech. V! t t Short people, short trees? • Kids up to about 3 years old often seem to leave a lot of the functional stuff. Agreement is missing, have or be is missing, do is missing. Tense is missing. 1) Papa have it (Eve, 1;6) • The stuff that is considered “functional” is generally also the stuff that’s higher in the tree—IP, CP, maybe DP too. • Why might kids leave the functional stuff out? 2) You nice (Sarah, 2;7) 3) Fraser not see him (Eve, 2;0) Small children’s small clauses • Radford (1990): Children’s clauses are just VPs at a certain point. (Cf. I saw Mary eat an apple.) (What things does this predict?) • Well, maybe their trees are still growing. MLU • Kids’ linguistic development is often measured in terms of Mean Length of Utterance (MLU). • Can be measured in various ways (words, morphemes) VP NP V Mumma V NP ride horsie ! 2-year olds • Around 2 years old • Around MLU 1.75 • Around 400 words in the vocabulary • 1-3 word utterances • Word order generally right • Grammatical words (the, is) generally missing Lexical and functional • Gives an idea of kids’ normal utterance length • Seems to correlate reasonably well with other qualitative changes in kid productions 2 1/2 year olds • • • • About 2 1/2 to 3 years • Over-regularization errors (He goed in the house), indicating they’ve grasped the rule of past tense formation. • Single clause sentences About MLU 2.25 About 900 words in the vocabulary Some grammatical devices (past tense -ed, verbal ing). 3 and 4 year olds • About 3 to 3 1/2 years, MLU about 2.75, about 1200 words, beginning to use syntactic transformations (Is Daddy mad? Where is he going?) • About 3 1/2 to 4 years, MLU about 3.5, about 1500 words, multi-clause sentences, still some over-regularization Root infinitives ! French: ! Pas • Another, fairly recentlynoticed aspect of child speech is that they will use infinitive verbs sometimes when adults would use finite verbs. In lots of languages. manger la poupée not eat[inf] the doll ! Michel dormir Michel sleep[inf] ! German: ! Zahne putzen teeth brush[inf] ! Thorstn das haben Thorsten that have[inf]. ! Dutch: ! Ik ook lezen I also read[inf.] No root infinitives children learning some languages • However, (e.g., Italian) seem to show very few root infinitives or none at all. these languages with very few root • Often infinitives... • Allow null subjects fairly complex agreement • Have morphology 4 and 5 year olds years, MLU around 4, about 1900 • 4-5 words, using more conjunctions and temporal terms (before, after), gain some metalinguistic awareness. 5, MLU stays about the same (no • After longer predictive), sentences get more complex, vocabulary increases (more slowly), over-regularization decreases… Root infinitives kids do this too, it turns out, but • English this wasn’t noticed for a long time. • It only write on the pad (Eve 2;0) • He bite me (Sarah 2;9) • Horse go (Adam 2;3) • It looks like what’s happening is kids are leaving off the -s. the crosslinguistic facts into • Taking account, we now think those are nonfinite forms (i.e. to write, to bite, to go). Distinguishing finiteness • In languages where root infinitives are found, however, it appears that children treat infinitives and finite verbs differently —and in general, appropriate for their type in the target language. • So—it winds up being quite sophisticated knowledge we need to attribute to them. Children’s spontaneous speech CHILDES Language Data Exchange System • Child http://childes.psy.cmu.edu • When we look at approaches to in 1984, Concord, MA. • Founded Director Brian MacWhinney understanding a child’s syntactic development, we try to see what each predicts... and then try to see whether the predictions are met. source of, among other things, computerized, • Asearchable, transcripts of child speech. When using data from CHILDES, you • Note: must always cite the original source of the • This is where computer-searchable corpora can be very useful. Enter: CHILDES. data. See the CHILDES database manual for details on what to cite for each corpus. Components of CHILDES Longitudinal data • The most generally useful corpora within CHILDES are the longitudinal studies of single children. • Chart: Snyder (2007:53) CHILDREN'S SPONTANEOUS SPEECH 53 TABLE4.1 Principal longltudlnal corpora for Fnghsh Child # Child utterances Main investigator Adam Anne Aran Becky Carl Domimc Eve Gail Joel John Liz Naomi Nicole N'I~ Peter Ruth Sarah Shem Warren Mean gap between recordings (in Days) Roger Brown Manchester Group Manchester Group Manchester Group Manchester Group Manchester Group Roger Brown Manchester Group Manchester Group LMaiche&ter Group Manchcstcr Group Ja~yuehneSachs Manchester Group Patrick Suppes LOISBloom Manchester Group Roger Brown Eve Clark Manchester Group • CHAT: CHAT is a transcription protocol common to most transcripts in the CHILDES database. • CLAN: CLAN is a program (actually a collection of programs) used to transcribe data and analyze transcripts. • CHILDES: The database itself consists of the transcripts (or other data, e.g., video, audio). as well as the age span covered and the frequency of recording (e.g. once a week, twice a month, once a month). As a result, the researcher needs 10 decide at the outset which corpora are suitable for the research question at hand. In the present case I will take the corpus in Table 4.1 that has the smallest mean gap between recordings: that of Sarah. This high frequency of recording, together with the large size of the corpus (over 37,000 child utterances), should provide the finest-grained picture available of how the child's speech changes over time. In addition to differences in the size, age span, and recording frequency of these corpora, one needs to consider differences in the approach to transcription. Unfortunately for researchers, children's CHAT • • • • advantage ofdistinguishingclearlybetween "2,zj" (two months, 25 days) and "2.25 inontl~s" @UTF8 (two-md-a-quarter months), for example. lists only those CHILDEScorpora available in July 2004 that follow non no lingual Table 4.1@Begin @Languages:! enany known developmental abnorndity, that arc based English-acquiring children without on audio- or videotape recordingsCHI of spontaneous speech (rather than simple diary notes, LOI Lois Investigator, @Participants:! Peter Target_Child, MOT Mother, or recordings of controlled experiments), that cover a time span of a1 least nine months, ! the PAT Investigator, LYN Lynn begin when childPatsy is no older than 2;o3, that have an average gapInvestigator, between recording JEN Jennifer of no more! than Child one month, that contain at least 10,000 child utterances, and that are described in the available documentation as being iu a finished state. @ID:!en|bloom70|CHI|2;1.|male|normal||Target_Child|| The CHAT format guidelines for coding your own transcripts are quite involved headers • @Participants speaker “tiers” • *CHI:, *PAT: unintelligible speech • • “xxx”, ignored. “xx”, a word. @ID:!en|bloom70|MOT|||||Mother|| @ID:!en|bloom70|LOI|||||Investigator|| @ID:!en|bloom70|PAT|||||Investigator|| @ID:!en|bloom70|LYN|||||Investigator|| @ID:!en|bloom70|JEN|||||Child|| @Tape Location:! Tape 16, side 1 @Comment:!MLU 2.39 CLAN • Useful commands: calculate • freq: frequency of words @Time Start:! 15:00 @Situation:! Peter is just waking up from nap when Lois and Patsy arrive, ! adults talk about Jennifer who is now five and a half months ! old *PAT:! ! hey Pete # that's a nice new telephone # looks like it must do everything # it must ring and talk and . %mor:! in transcript(s) search for • combo: things in the transcripts co|hey n:prop|Pete pro:dem|that~v|be&3S det|a adj|nice adj|new n|telephone ! n|look-PL v|like pro|it v:aux|must v|do pro:indef|everything pro|it v:aux|must ! v|ring conj:coo|and n|talk conj:coo|and . calculate mean • mlu: length of utterance %exp:! Peter has a new toy telephone on table next to him %com:! untranscribed adult conversation *CHI:! xxx telephone go right there . %mor:! unk|xxx n|telephone v|go adv|right adv:loc|there . %act:! reaches out to lift phone receiver, pointing to place where ! wire should connect receiver and telephone *MOT:! the wire . %mor:! det|the n|wire . in the transcripts freq > freq How this helps sample.cha freq sample.cha Sun Sep 12 19:48:56 2004 freq (10-Sep-2004) is conducting analyses on: ALL speaker tiers **************************************** From file 1 a 1 any • 1 are The freq command tallies up the number of times each word appears in the transcript. • Useful to figure out which words are most common (or which words are used at all) in a child’s transcript. 3 chalk • Harkening back, we talked about a couple of ideas about what’s wrong with kids’ trees. • Each idea makes predictions about what kids will and won’t say—and CHILDES can be used to see to what extent these predictions are met. 1 chalk+chalk 1 delicious 1 don't 1 eat [ . . . ] 1 toy+s 2 toys 3 want 1 what 2 what's 1 wonderful 2 yeah 2 you -----------------------------34 Total number of different word types used 50 Total number of words (tokens) 0.680 Type/Token ratio Searching • For example, we can search for all examples of the pronouns me, I, and my, and look at correlations between the form of the verb and the form of the pronoun. • Note on excluding imitations and repetitions—we are trying to characterize the child’s grammar. • Relatively painless computerized searching • A lot of data available, a lot of kids available (relative to pen and paper, at least)