Introducing second language acquisition hummel pdf

Linguistic Term Paper

1

Childhood Bilingualism

1. Overview and definition(s) § Bilingualism: ability to use two languages since early age (see 1.1) § Definition of ‘bilingual’? Not only one! (cf. 27 types of bilingual! Hummel 2014:226-7) § Theoretical questions:

i. Do children keep the two languages separate? ii. Do the two languages develop independently, in parallel? iii. Are the developmental stages always the same in the two languages, e.g.Null Subject stage, Root Infinitive stage, etc.? iv. Do the languages influence/interfere with each other? v. Does bilingualism affect the time-course of development?

2. Characterization of Simultaneous Bilingual Development

• regular and continued exposure to more than one language before age 3 (or 2) • Bilingual first language acquisition (BFLA) - exposure to 2 or more languages from

birth • Bilingual second language acquisition (BSLA) - exposure to a 2nd language after 1

month and no later than 2 years after birth • Successive childhood L2 acquisition: child acquired the second language after age ~3

2.1 Early Diary Studies § traditional sources of data: 20 or so case studies by linguists of their bilingual children § cross-sectional studies with older children, but are less informative

2.2 Conditions of exposure and "mixing" § Early studies focused on the effects of “conditions of exposure” in bilingual

development (relation between manner/presentation and language mixing) § Mixing was related to: poor acquisition, fused systems, confusion, etc.

§ Conditions of exposure include things like: a. Who is speaking each language? (eg. mother, father, nanny, etc.) - parents speak different languages, parents vs. babysitter, home vs. school/friends/etc. - fluency depending on on their previous experience (e.g. specific vocabulary,

Ling C140. Fall 2019 G. Silvestri

2

expressions, colloquialisms, etc.) b. How much input in each language? (eg. hours per day) - there can be different amounts of input in the two languages - Most studies want 40-60% or at worst a 30-70% split in exposure to two languages. c. Under what social conditions are the two languages presented? (eg. home, foreign country, school, etc. - language prestige, minority vs. majority language, etc. may affect bilingual development d. Nature of input: is there mixing in the input? § Romaine (1989) notes 6 different kinds of bilingual situations:

(i) One person-one language: 2 parents with two languages one of which is the dominant language of the community (ii) Non-dominant home language/one language-one environment (iii) Non-dominant home language w/o community support (iv) Double non-dominant home language w/o community support (v) Non-native parents: same native language/one parent talks to child in non-native language (iv) Mixed languages: parent are bilingual and mix languages

§ Mixing/interference includes the following:

(From Swain and Wensche (1975) French/English bilingual children: Only 4% mixed utterances)

• Lexical (English/French) (1) a. Je vais te donner un sweater. 'I am going to give you a sweater' b. Ca you tear. 'That you tear' c. His nose is perdu. 'His nose is lost' d. Is that side oú that side? 'Is that side or that side?' e. Est-ce que you give it to her? 'Did you give it to her?' • Morphological: Hildegard's examples above W. Leopold (see below): German-English: pouren, praticen, monthe • Syntax: syntactic frame of one language, words of the other. (2) a. I wanta glue it, this. b. I think Marie want not to listen.

Ling C140. Fall 2019 G. Silvestri

3

c. That's to me. d. That's what? e. A house pink • Semantic: meaning of word in one language taken on by a related word in the other language (3) a. I went to the library. (Italian libreria = book store) b. I went there on this opportunity. (Italian opportunità = occasion) c. bureau (Fr.) = office, not dresser § Two influential early studies and their conclusions:

(A) Jules Ronjat (1913) studied one French-German speaking child who did not mix the 2 Ls - proposed une personne/une langue ("one person-one language") principle: apparent lack of confusion between 2 languages (cf. also Clyne 1987; Bain and Yu 1980; Kielhofer and Jonekeit 1983)) - Later termed ‘one situation-one language.’

§ child has no linguistically-based mechanism to keep Ls separate at the beginning § child has to rely on extra-linguistic cues (who is doing the talking)

(B) Werner Leopold (1939, 1947, 1949a, 1949b) studied one German-English bilingual (his daughter, Hildegard) - First major American study - father was German and mother American. - 4 volumes: vocabulary (to age 2); phonology (to age 2); syntax (to 2) and language age 2- 12.

§ Leopold's major findings:

• very few bilingual synonyms (translation equivalents) • little phonetic interference • interference/mixing at the lexical and morphological levels after age 2. eg. pouren, praticen, monthe (see below)

§ conclusions: prior to age 2 children have a single system, one lexicon and a primitive

system of rules § Children naturally try to make a single system

§ More recent studies

Ling C140. Fall 2019 G. Silvestri

4

• Garcia (1983): lack of complete adherence to one person/one language does not necessarily impede bilingual development - Sp/En preschoolers exposed to Sp and En: incidence of mixing quite low (1- 15%)

• Goodz (1989) Fr/En bilinguals - parent responded to mixed utterance with mixed utterances - mixing = a pragmatic issue

• Genessee (1989), Meisel (1989) - mixing may reflect input or language dominance

v Tentative conclusions regarding conditions of exposure: • conditions of exposure: (maybe) important consequences for bilingual development (first regular exposure, amount of exposure, or of separation/mixing in input) • une personne-une langue: assumption that the child has no linguistically-based mechanism to keep Ls separate at the beginning may be wrong • In earlier studies mixing = poor acquisition or confusion - More recently mixing = kind of CODESWITCHING 3. Theories of Bilingual Development (2 major ones) 1. Unitary Systems Hypothesis (Leopold 1939; Volterra and Taeschner 1978; Paradis and Genessee 1996) (4) Child initially constructs one grammar and one lexicon that later separate

2. Separate Systems Hypothesis (Meisel 1987; Genessee 1989) (5) Children construct a separate grammar and lexicon for each language

3.1 Unitary System Hypothesis § CODE-SWITCHING: child confused about the two languages

Ling C140. Fall 2019 G. Silvestri

5

- evidence for a unitary system (Bhatia & Ritchie 1999; Genesee 1989; Leopold 1970; Schnitzer & Krasinski 1994; Volterra & Taescher 1978; Vogel 1975)

v 3-STAGE MODEL (Volterra & Taeschner (1978); Lisa and Giulia, 2 German/Italian

bilingual children)

1. Stage I: one lexicon and one primitive system of rules - no synonyms (translation equivalents (TEs)). - words of 2 Ls mixed in 2, 3 word utterances (see examples in (1))

2. Stage II: 2 lexicons, but one set of syntactic rules

- words of 1 L have correspondences in the other L -words from different Ls rarely occur in the same S - child is able to translate from 1 L to the other - same syntax in both Ls (mixed syntax)

- Mixed syntax: Lisa (1;8-2;8), possessive NPs have German syntax (6) miao bua, Paola tita, Guarda Giulia bua, Lisa bicicletta 'cat wound?, P pencil, look Giulia wound?, L bicycle' Lisa (2;9) (correct order for Italian) (7) a penna di Lisa, quetto è di Giulia, la stellina di mama 'the pen of L, this is of G, the star of mom' Lisa (2;9) Adjectives-Noun have German order: (8) a. Dov'è il grosso lupo?

'Where is the big wolf?'

3. Stage III. Separate lexicons and separate rules for each language: (9) a. Giulia Buch

'Giulia’s book' b. Sono i capelli di Lisa

'(They) are the hairs of Lisa' 3.2 Discussion of Unitary System Hypothesis A. Empirical and methodological issues Stage I

• No translation equivalent (TEs) claim is based on 137 words, kids ages 1;10 and 1;6

Ling C140. Fall 2019 G. Silvestri

6

• Children were only speaking with German interlocutor. 4. Quay (1993)

Sp/En bilingual: many TEs - data were gathered in 2 different contexts(Spanish interlocutor and other with English interlocutor, same environment ) • Reexamination of the T &V data shows TEs (Quay 1993); Lisa had 6/54 (10%) TEs; Giulia 14/73 (20%)

Stage II

• confusion concerning the ages of the purported stages (Meisel 1989), eg. Italian adjectives not used until age 2;9 (when stage II is supposed to be over).

B. Conceptual issues

• Unitary system hypothesis: more natural for a child to acquire one language (In contact situations children create creoles = 1 system incorporating elements of different input! ) • If you start with one system, how do you then separate?

C. Conclusions

• Mixing: not conclusive evidence for a unitary system • hard to establish that there is only one system

3.3 Discussion of Separate Systems Hypothesis (J. Meisel 1989) § two separate lexicons and grammars right from the beginning (Ingram 1981; Keshavarz

& Ingram 2002; Meisel 1989; Paradis 1996, 2001; Paradis & Genesee 1996; Schnitzer & Krasinski 1996)

§ differences in constructions in adult Ls same as differences in constructions in children: evidence for separate systems

3.3.1 Rhythm and language discrimination § soon after birth infants from monolingual families discriminate utterances drawn from

languages that differ prosodically, but not between rhythmically similar languages • Bosch & Sebastián-Gallés (2001): language discrimination in 4-month-old

monolingual and bilingual infants simultaneously exposed to 2 Romance languages with same rhythmic category.

- Subjects: - 14 mono Spanish 4mos - 14 mono Catalan 4mos - 28 Spanish-Catalan bilingual 4mos

Ling C140. Fall 2019 G. Silvestri

7

Procedure: - Familiarization-preference procedure - Familiarization phase: infants exposed to sentences in their native language (or mom’s language in the case of bilinguals) until 1.5min of listening - Test: infants exposed to different sentences either in their native language or in the other language Results: - Monolinguals discriminate between the two languages - Bilinguals also discriminated between their 2 languages (10)

- observed effects same size - early capacity to distinguish languages in simultaneous bilingual exposure - challenge to the hypothesis of inexistence or delayed of discrimination capacities are in bilinguals 3.3.2 Visual language discrimination § Weikum, Vouloumanos, Navarra, Soto-Franco and Sebastián-Gallés (2007)

infant’s abilities to discriminate between languages just from viewing silent videos Subjects: - Mono En 4mos, 6mos, and 8mos - En-Fr bilingual 4mo, 6mos, and 8mos. Procedure: - Visual habituation procedure - Habituation phase: infants looked at silent videoclips of bilingual Fr-En speakers reciting sentences in one language until their looking time declined by 60% - Test phase: Infants looked at silent videoclips of the same bilingual Fr-En speakers reciting sentences in either the language they had not been exposed to during the habituation phase (test/switch condition) or the same (control/same condition)

Ling C140. Fall 2019 G. Silvestri

8

Results: - Monolinguals showed a capacity to discriminate between the two languages at 6mo, but not at 8mo. - Bilinguals were able to discriminate between the 2 speakers at 6mo and at 8mo.

o bilingual infants discriminate on visual cues even if they are not exposed to those languages (Sebastián- Gallés, Albareda-Castellot, Weikum & Werker 2012).

v bilingual experience modulates the attentional system

3.3.3 Phonemic discrimination § Sundara, Polka, and Molnar (2008): coronal stop perception in monolingual and bilingual

Fr-En infants § English has alveolar /d/, French has dental /d/

- distinction not phonemic in either language Can bilinguals discriminate between them?

- Subjects: - 6-8mos - 16 mono Fr - 16 mono En - 16 En-Fr biling - 10-12mos - 16 mono Fr - 16 mono En - 16 En-Fr biling

- Procedure: - Visual habituation procedure - Habituation phase: half the infants heard 4 French [dæ] syllables, half heard 4 English [dæ] syllables produced by 2 different speakers, until they lost attention looking time in last 3 trials is 50% of the looking time of the longest 3 consecutive trials) - Test phase: Syllables produced by a new speaker were presented in ABAB order. A = novel, B = familiar. - Results: - 6-8mos were all able to discriminate between dental and alveolar /d/, regardless of lg

Ling C140. Fall 2019 G. Silvestri

9

background - Fr 10-12mos failed - Bilingual succeeded - Mono En 10-12mos also succeeded (Why?) (11)

- If overlapping distributions was a problem, bilingual 10- to 12-month-olds would have difficulty discriminating dental–alveolar variants of /d/: that is not the case! - However, alveolar /d/ and dental /d/ VERY common - Less frequent phonemes acquired more slowly - E.g. The Catalan /e/–/ɛ/ contrast perceived by mono Catalan infants at 4mo, 8mo, and 12mo. Mono Sp infants succeeded at 4mo, but failed afterwards. Bilingual infants succeeded at 4mo, failed at 8mo, and succeeded again at 12mo (Bosch and Sebastian Galles 2003). Why do you think so? 3.3.4 Production: segmental phonology § bilingual children’s segmental phonology similar to monol children wrt substitutions (e.g.

[l] for [r] in the Spanish word ‘cruz’), VOT, and consonant harmony and syllable reduplication.

§ (bidirectional) transference between languages? transference requires by definition two separate systems

§ Paradis (2001): nonword repetition task designed to investigate differentiation and autonomy in bilingual children’s phonological development.

- monol En and Fr-speaking children as well as En-Fr bilingual children aged 2;6 - 10 4-syllable nonsense words in Fr and 12 4-syllable nonsense words in En to test sensitivity to word rhythm and syllable weight - Predictions: - Monol En-speaking children truncate target words following trochaic (SW) production template - Monol Fr-speaking children: truncate target words following iambic (WS)

Ling C140. Fall 2019 G. Silvestri

10

production template - Bilinguals: in the Eng session behave like mono En and in the Fr session they behave like mono Fr Results: - Monol and biling behaved same way in the Fr session - Monoll and biling Eng children behaved same way in all word types but one Discussion: - restricted nature of the crosslinguistic effects: children have two phonological representations - However, crosslinguistic effects appear: directionality of these effects influenced by language dominance. § Fabiano-Smith and Goldstein (2010): spontaneous speech of En-Sp bilingual 3yo

- All biling: phonetic inventories typical of their chronological age - Many biling children did not produce later-developing Sp sounds (flap and trill;Acevedo

1993) - As for En: bilingual group did not produce later-developing sounds such as /θ/ and /ð/ (Shriberg 1993) = monol Eng speakers - Consonant accuracy not different for the monolingual En-speaking - Consonant accuracy slightly better for the monolingual Sp-speaking compared to the Sp bilingual children (but biling not out of the range of the monol peers) - children did slightly better on shared sounds than unshared sounds - Only 25% of children showed any effect of transfer (e.g. deaspirated stops in English) 3.4 Evidence for 2 grammars • If child follows a different developmental trajectory in the two languages i. Some property develops in one but not the other grammar ii. Child goes through a developmental stage in one language but not the other iii. Child shows same behavior as monolinguals in each language • Lisa-type examples (Volterra & Taeschner Stage II errors) due to transfer/interference

from dominant language (and transfer entails 2 systems): - examples are inconclusive - But if you find the converse, evidence for 2 systems

§ Meisel (1989): 2 French/German bilinguals, 1-4 years: word order and SV agreement

• Word order: German SVO/SOV - French SVO, Adult examples: (12) a. Ich schreib einen Brief. (V2)

Ling C140. Fall 2019 G. Silvestri

11

'I wrote a letter' b. ...dass ich einen Brief schrieb. ...that I a letter wrote

(13) a. Notre chat a tout bu 'Our cat has drunk everything' b. Il a tout bu, notre chat 'He has drunk everything, our cat'

• Subject-Verb agreement: - German has rich inflection: (14) ich sage wir sagen,

du sagst ihr sagt er sagt/sie sagt sie sagen

- French is relatively impoverished: (15) je parle nous parlons

tu parles vous parlez il parle ils parlent

§ Results • Children used predominantly SVO word order in both languages, but used it less frequently in German. • SOV order only in German with infinitives (like monolinguals) (eg. Thomas essen) • VOS order only in French (eg. tomber bébé) • fronting of adverbs trigger V2 in German, but not in French, Adv VS (German) (diagnostic of V2 if non-subject above V), ADV SV (French) • SCs acquired early in French • No errors in person marking on German verbs

v Children look like two monolinguals in one head = 2 separate systems developing

§ Paradis and Genesee (1997) - Fr-En bilinguals: Matthieu 1;9-2;11 and Yann 1;11-3;0 - Tracked development of Infl and Det

- IP: frequency of finite verbs, occurrence of modals and auxiliaries (cf. appendix). - DP: use in obligatory contexts (cf. appendix)

Ling C140. Fall 2019 G. Silvestri

12

Table 1. Percentage of finite verbs and age of first occurrence in English and French English French Yann 0 65% (1;11) Matthieu 3% (2:11) 43% (2;3) v Propose 3 criteria for assessing if children show autonomous development in their

languages: o A. TRANSFER o B. ACCELERATION o C. DELAY

(would mean the two systems are perhaps independent but still have an effect on one another)

A. TRANSFER: complete incorporation of property from one L into the other (not just sporadic interference), most likely if one language is more advanced/dominant B. ACCELERATION: a certain property emerges earlier than it otherwise would in monolingual acquisition, again more likely if child has achieved a more advanced level of syntactic complexity in one L than the other C. DELAY: burden of acquiring two Ls could slow down the acquisition process • results parallel those for monol • Fr monol use more finite forms than En monolinguals

v Conclusions

§ Paradis & Genessee (1997) argue against a maturational theory of functional category development

§ - children developed IP in Fr and En at different times = different curves § results support a separate systems hypothesis

3.3.1 Bilingual phonological development § Evidence for separate systems is much weaker § Leopold found no mixing at the phonological level § Deuchar & Clark (1989, 1992) Sp-En bilingual

- initially (1ge 1;7) Manuela no voicing contrast and no difference in VOT between Sp and En; by 2;3 the values are quite different - unitary system or version of the effect with monol in which En VOT contrast is "easier"

§ C. Labelle (2000): prosodic development (En-Fr bilingual)

Ling C140. Fall 2019 G. Silvestri

13

-English is basically trochaic (stress on initial foot) and French is iambic (stress on final foot) -Fr-En bilingual girl age 1;5 to 2;3 produced trochaic stress in both languages (cf. appendix) and truncated words according to a trochaic stress pattern. - evidence for a unitary system or universal trochaic template (Cutler et al. (1989))?

§ To decide, find a case in which there is no interfering and overriding UG influence 3.3.2 Explanations for mixing (F. Genessee 1989) (if not single system) § If children's system undifferentiated, child should use items from both languages

indiscriminately in all contexts - Instead, Genessee (1989): children use language that is being used in a particular context - mixing is code-switching or it is due to other factors: • violation of rules governing CODE-SWITCHING: - Biling adults frequently mix languages according to various non-linguistic factors such as selecting a language according to interlocutor, situational context, etc. - Children code-switch, but no pragmatic or grammatical rules, eg. no switching between a functional category and its complement e.g. the gatto

• INSUFFICIENT LEXICON in one language (retrieval problem) - If child doesn't have the right word in one L, he may choose a word from the other L - Adults do too • AVOIDANCE of phonologically difficult words - Monolingual children selectively avoid words which are hard to pronounce - Bilinguals have a another option • FIRST USE - Children may identify referent with the word that was first or more frequently used to label it - May be hard to acquire new word in second language • MIXED INPUT - input may be more mixed that people realize - Goodz (1989): in 17 Fr-En biling frequency of mixing correlated with frequency of adult mixing - Romaine (1995): children in type 6 bilingual situation do show a lot of mixing for a longer period of time 3.3.3 Differences and similarities with monolingual development

Ling C140. Fall 2019 G. Silvestri

14

• Conditions of exposure (amount of input in each L degree of separation of 2 languages) may matters for bilinguals, though not for monolinguals, but evidence is anecdotal • perfectly balanced bilingualism is difficult, maybe impossible, to achieve • bilingual children construct grammars in the same way as monolingual kids • no significant delay in acquiring 2 languages over acquiring 1 language, except with respect to lexicon [Despite this people have been discouraged from raising their children bilingually. We will return to this when we talk about cognitive and educational effects of bilingualism.] 4. Crosslinguistic influences in bilingual development § balanced bilingualism maybe impossible to achieve:

• Typically balanced = 40-60% • Anything less than 30% of exposure will result in delays or even incomplete

acquisition (Baker 2014; Genesee 2007; Thordardottir 2011, 2015) • delays in bilingual development, due to divided input or due to or influence from

one language to the other • Recent work: which aspects of language are most likely to “transfer” in bilingual

development? common assumption : children DO have two separate systems influencing each other

v Müller and Hulk (2001) discuss the case of object clitics o Monolinguals acquiring Romance languages drop clitics/objects (15a-b). Bilinguals do too, (16c-d). (16) a. Il met dans le bain. (Lou-French) He puts in the bath (cf. Il le mets dans le bain.) b. Anche lui ha. (Marco – Italian) Also he has (cf. Anche lui ce l’ha.) c. Iva repare. (Ivan – French-German bilingual) Ivan repair (cf. Ivan la repare.) d. Prendiamo (Carlotta – Italian-German bilingual) Take-1PL (c.f. La prendiamo.) § But monol children drop objects approx. 11% of the time while biling do so much more

frequently: Ivar 39.5% - Anouk 37.5% - Carlotta 36.4% § Mu ̈ller and Hulk (2001): null object stage (developmental)

- frequency and duration of the stage influenced by the bilingual’s other language (if superficially similar to the developmental phenomenon)

Ling C140. Fall 2019 G. Silvestri

15

v Paradis & Navarro (2003) - examine the case of null subjects in En-Sp bilinguals and mono Sp children.

- Sp = a pro-drop language, En = no-pro-drop language - Will the overt presence of subject in En affect the realization of subjects in Sp? - Participants: - E & L: Mono Sp aged 1;8-2;12 - M: Biling Sp-En aged 1;9-2;8 - Procedure: - CHILDES corpus study - Results: - M, the biling produced more overt subjects than the two monol - However, M’s parents also produced more overt subjects! Perhaps due to a dialectal difference - results could be due to influence from En, or the parent’s input of Sp

Two requirements for cross-linguistic influence

(i) property associated with the syntax-discourse interface (the “vulnerable C domain”)

(ii) surface overlap between the 2 languages

v Yip & Matthews (2000): syntax of a bilingual En-Cantonese child who

Ling C140. Fall 2019 G. Silvestri

16

was Cantonese dominant:

§ Wh-questions - Cantonese has wh- in situ - English requires movement of the wh-element to Spec-CP

(18) a. What did you eat? b. Lei5 sik6-zo2 mat1je5? you eat-PFV what

§ Null objects: Cantonese allows for null objects. En does not and even young children virtually never omit object pronouns in En

(19) A. Gin6 saam1 hou2 leng3 wo3. CL blouse very pretty PRT B. Ngo5 zunglji3 aa3. I like PRT ‘I like (it)’ Participant: - Timmy: 1;6-3;6, lives in Hong Kong. Mom speaks Cantonese, dad speaks British En. - They speak Cantonese between them - Extended family speak Cantonese - Community speaks Cantonese. Procedure: - Timmy’s speech was recorded and transcribed Results: - Wh-questions: Timmy initially asks wh-questions in En with wh- element in situ - These may look like echo questions, but these are very limited in mono child speech (1.6% in Eve vs. 67.6% in Timmy). - Null objects: Timmy produces null objects 12.3-35.3% at a higher rate than mono En children (4-5% in Adam; 2-3% in Valian’s (1991) study, 3.75 in Wang et al.’ (1992) study) § Some studies have found crosslinguistic influences e.g. En-Ita bilin child (Serratrice,

Sorace & Paoli, 2004) and Hebrew-En bilingual child (Hacohen & Schaeffer, 2007) § Some studies haven’t, e.g. English-Catalan child (Juan-Garau & Perez-Vidal, 2000)

v The bilingual has separate systems but development in one language can be influenced/affected by the other language

Ling C140. Fall 2019 G. Silvestri

17

v We will discuss what happens when LA becomes dominant and acquisition of LB is

interrupted or “forgotten” in the Heritage Languages lecture Assigned (**) and suggested reading (*) ** Hummel, K., 2014. Introducing second language acquisition. Chapter 3. ** Paradis, J. and & F. Genesee, 1996. Syntactic acquisition in bilingual children: autonomous or interdependent? SSLA 18, 1-25. ** Paradis, J., 2001. Do bilingual two-year-olds have separate phonological systems? International Journal of Bilingualism 5 (1), 19-38. * Polka, L. & M. Sundara, 2003. Word segmentation in monolingual and bilingual infant learners of English and French. Presented at ICPhS, Barcelona Aug 3-9th. Published in Solé, M. J., Recasens, D., Romero, J. (Eds.) Proceedings of the 15th International Congress of Phonetic Sciences, 1021-1024.