Bilingualism and Child L2 acquisition

1
Bilingualism and Child L2 acquisition

• Linguistic Research on Second Language
  Acquisition Development Lecture 8
• sharon.unsworth_at_let.uu.nl

2
Todays lecture

• introduce and define some terminology
• brief overview of (some of) literature on
  acquisition of two languages from birth
• single or separate systems?
• methodological issues
• same as monolingual L1 acquisition? ?
  crosslinguistic influence/transfer/interference?

3
Todays lecture

• brief overview of (some of) literature on child
  L2 acquisition
• why look at child L2?
• comparison with adult L2 acquisition
• comparison with monolingual/bilingual L1
  acquisition

4
Definitions

• Whats a bilingual?
• Whats bilingual acquisition?

5
Whats a bilingual?

• Individual bilingualism vs. societal bilingualism
• The ability to produce complete and meaningful
  utterances in two languages (MacLaughlin 1978)
• What does complete mean? Native-like?
• What counts as meaningful?
• Probably too inclusive

6
Whats a bilingual?

• A person with native-like control of two
  languages (Bloomfield 1933)
• The two monolinguals in one person viewpoint
  (Grosjean 1989) too simplistic
• What does control mean?
• Who should be considered a native?
• Bilinguals use different languages in different
  contexts
• Bilingual ability vs. bilingual use or degree
  vs. function of bilingualism
  (Baker 1996 Mackey 1968)

7
Whats a bilingual?

• Bilingual a person who uses two (or more)
  languages (or dialects) in their everyday life
  (Grosjean 1998132)
• two languages are complementary
• equal fluency in both languages is rare
• dynamic, i.e. changes over time
• interaction with both monolinguals and bilinguals
  ? different behavioural patterns

8
Whats bilingual acquisition?

• simultaneous vs. successive

the acquisition of two languages from birth
the acquisition of one language before the other
as an adult
as a child
9
Whats bilingual acquisition?

• vs.

Bilingual First Language Acquisition (2L1A)
Child and Adult Second Language Acquisition
10
Bilingual First Language Acquisition

• (De Houwer 19903, following Meisel 1989)
• BFLA refers to those situations in which
• a child is first exposed to language B no later
  than a week after he or she was first exposed to
  language A
• a childs exposure to languages A and B is fairly
  regular, i.e. the child is addressed in both
  languages almost every day

11
Bilingual First Language Acquisition

• Why such a strict criterion?
• (Padilla Lindholm 1984, cited in Romaine
  1995)
• whats already known in one language might have a
  subsequent effect language acquired later
• to ensure strict comparisons with monolingual
  children

12
Bilingual First Language Acquisition

• single or separate systems?
• methodological issues
• same as monolingual L1 acquisition?

13
Bilingual First Language Acquisition

• single or separate systems?
• methodological issues
• same as monolingual L1 acquisition?

14
Single System Hypothesis

• (Volterra Taechner 1978 Redlinger Park 1980
  Vihman 1985)
• children start out with one system only before
  differentiating their two languages
• evidence mixed utterances, i.e. utterances
  containing elements from both languages

15
Volterra Taeschners (1978) stages

• one lexical system with words from both
  languagese.g. NL-ENG bilingual uses meisje and
  boy but not girl and jongen
• separate lexical systems but same syntactic
  systeme.g. NL-ENG bilingual produces (ik) wil
  dat niet and want that not
• separate grammatical systems
• e.g. NL-ENG bilingual produces dat doe ik
  niet and Im not going to do that

16
Against Single System Hypothesis

• Genesee (1989), Paradis Genesee (1996)
• data not presented/analysed in context so we
  dont know whether children use lexical items
  from two languages differently
• no data on input mixing in child utterances may
  result from mixing in input
• decline in mixing over time may result from
  increasing linguistic repetoires rather than
  separation of two systems

17
Against Single System Hypothesis

• Genesee (1989), Paradis Genesee (1996)
• data/analyses are often incomplete e.g. data
  from one of two languages are missing
• questionable whether code-mixing is a valid
  measure of single underlying system (because its
  part of sociolinguistic/pragmatic competence)
• circularity mixing evidence for single system
  and single system explanation for mixing

18
Against Single System Hypothesis

• Köppe Meisel (1995)
• code-switching acquired very early (age 20) (see
  also Lanza 1992)
• sociolinguistic constraints children choose
  language according to addressee
• structural constraints dependent on properties
  of two languages, imposed on intra-sentential
  switches

19
Against Single System Hypothesis

• ? language mixing reflects sophisticated language
  use rather than properties of underlying system

20
For Separate Systems Hypothesis

• Meisel (1989)
• two bilingual French/German children
• boy (27-37) and girl (111.14-30)
• predominant word order in two-word utterances
  containing objects is VO in French and OV in
  German

21
For Separate Systems Hypothesis

• evidence for early use of syntactic categories
  used in target-like ways in each language, i.e.
  separate systems

22
For Separate Systems Hypothesis

• More evidence Meisel (to appear) summarises.
• bilinguals with V2 language and non-V2 language
  consistently produce target-like utterances, i.e.
  finite verb in 2nd position in V2 language and in
  3rd position in non-V2 language

23
For Separate Systems Hypothesis

• Meisel (to appear) summarises.
• similar differentiation for negative
  constructions in Basque-Spanish and subject
  omission in combinations with one null-subject
  language
• these differences appear as soon as multi-word
  utterances appear (approx. age 110, MLU 1.75)

24
For Separate Systems Hypothesis

• Meisel (to appear) summarises.
• the available evidence favours very strongly the
  dual system hypothesis
• (see also e.g. Kaiser 1994 Meisel 1990 Parodi
  1990 for more evidence of early differentiation
  also papers in Meisel 1990, 1994))

25
Bilingual First Language Acquisition

• single or separate systems?
• methodological issues
• same as monolingual L1 acquisition?

26
Methodological issues

• Many common to monolingual L1 acquisition, e.g.
  pitfalls of spontaneous data
• Mode in which subject is recorded (Grosjean
  1998)
• bilingual language mode when both languages are
  activated
• monolingual language mode when one of languages
  is totally active and other is inactive
• continuum from monolingual to bilingual mode

27
Methodological issues

• How to analyse utterances (De Houwer 1998)
• utterances in Language Alpha
• utterances in Language A
• mixed utterances
• utterances that can be categorised as either
  Language Alph or Language A
• utterances whose linguistic membership is unclear
• utterances that are not relatable to any language
• Which utterances to include in analysis (depends
  on research question)

28
Methodological issues

• Which properties of language to examine should
  be aspects of two languages which are
  structurally different but functionally similar
  (De Houwer 1990, Meisel 1989)
• Bilingual errors vs. developmental errors (i.e.
  those found in monolingual L1 acquisition
  literature)

29
Methodological issues

• Need enough information regarding (De Houwer
  1998 see also Grosjean 1998)
• subjects sociolinguistic context (when, where,
  with whom subjects speak the two languages)
• input conditions (which languages child hears
  from whom, how much, etc.)

30
Methodological issues

• How to compare bilingual childs two languages
  (similar to problem of how to compare L1
  acquisition across languages)
• Is MLU an appropriate measure?
• Vocabulary size?

31
Bilingual First Language Acquisition

• single or separate systems?
• methodological issues
• same as monolingual L1 acquisition?

32
2L1A vs. monolingual L1A

• Does 2L1A ressemble monolingual L1A?
• Is 2L1A qualitatively or quantitatively different
  from monolingual L1A?
• Do the systems interact in the course of
  development?
• Autonomous or interdependent? (Paradis Genesee
  1996)

33
2L1A vs. monolingual L1A

• Any observed delays found for 2L1A are within
  range of normal rate for monolingual L1A (Meisel
  to appear), i.e. no quantitative differences
• Although measuring this isnt always easy!
• Separate systems similar to monolingual 2L1A
  (i.e. one source of possible qualitative
  difference is eliminated)

34
Paradis Genesee (1996)

• Do bilingual childs grammars interact over
  course of acquisition?
• Are they autonomous or interdependent?
• autonomous i.e. independent
• interdependent

35
Paradis Genesee (1996)

• interdependent i.e. systematic influence of
  Language Alpha on Language A, leading to
  differences in patterns and rates of development
  in comparison to monolingual L1A

influence at the level of representation or
competence, sustained over time
36
Paradis Genesee (1996)

• Interdependence may manifest itself as
• transfer incorporation of a grammatical
  property into one language from the other
• acceleration a certain property emerges in the
  grammar earlier than would be the norm in
  monolingual acquisition
• delay the burden of acquiring two languages
  could slow down the overall progress in
  grammatical development

37
Paradis Genesee (1996)

• Transfer
• has been reported in literature (e.g. Vihman
  1982) but not clear how systematic this is
  because of too little relevant information
• De Houwer (1990) no instances of transfer in
  Dutch-English bilingual 3-year-old, but subject
  may be too old

38
Paradis Genesee (1996)

• Acceleration
• few instances reported (but see work by Kupisch
  2004)
• Maiwald Tracy (1994) something comparable to
  acceleration
• German-English bilingual
• German functional elements in English when these
  werent available in that language

39
Paradis Genesee (1996)

• Delay
• evidence for delay Murrell (1966) Swain (1972)
  Vihman (1982)
• evidence against delay Padilla Liebman (1975)
  Nicoladis (1994) De Houwer (1990)
• problems sometimes no monolingual comparisons,
  data inadequate, lack of grammatical analyses

40
Paradis Genesee (1996)

• Looked for evidence of autonomy/interdependence
  in English-French 2L1A
• Aspects of language where two languages differ
• (i) emergence and use of finite verbs
• (ii) developmental stages of negation
• ((iii) distribution of pronominal subjects)

41
Paradis Genesee (1996)

• Emergence and use of finite verbs
• French verb raising
• Cécile I (ne) boiti NegP pas VP ti du café
• Cécile NegP (ne) pas VP boit du café
• English no verb raising ? affix-lowering
• Melinda I does NegP not VP drink coffee
• Melinda I drinksi NegP not VP ti coffee

42
Paradis Genesee (1996)

• Emergence and use of finite verbs
• Monolingual French L1A finite/inflected verbs
  in most utterances as young as 2 years
• Monolingual English L1A non-finite/uninflected
  verbs in most utterances until age 3 years

43
Paradis Genesee (1996)

• Emergence and use of finite verbs
• Analysis of monolingual L1A (Déprez Pierce
  1993, 1994) no movement in initial child
  syntax (no verb-raising in French, no
  affix-lowering in English)

44
Paradis Genesee (1996)

• Developmental stages of negation
• Early monolingual child French negator pas in
  preverbal position with non-finite verbs and
  postverbal position with finite verbs Pas
  chercher les voitures (Philippe 21)Ça tourne
  pas (Philippe 21)

45
Paradis Genesee (1996)

• Developmental stages of negation
• Early monolingual child English negative markers
  always preverbalNo Leila have a turn (Nina
  21)Me no go home (Peter 21)

46
Paradis Genesee (1996)

• Possible points of interdependence?
• transfer incorporation of a grammatical
  property into one language from the other
• acceleration a certain property emerges in the
  grammar earlier than would be the norm in
  monolingual acquisition
• delay the burden of acquiring two languages
  could slow down the overall progress in
  grammatical development

47
Paradis Genesee (1996)

• Possible points of interdependence
• early acquisition of verb-raising in French,
  coupled with evidence of some verb-raising in
  English (be and have) ? transfer of verb raising
  for English main verbs
• greater number of finite utterances in English
• postverbal negators in both English and French

48
Paradis Genesee (1996)

• Possible points of interdependence
• knowledge of finitenon-finite distinction
  present in French might accelerate emergence and
  use of marking of this distinction in English
  (i.e. affix-lowering)
• differences between two languages wrt movement
  and INFL ? all aspects of grammars involving
  these properties emerge later than they do for
  monolinguals, i.e. delay

49
Paradis Genesee (1996)

• Examination of spontaneous data from three
  bilingual English/French children
• No evidence for
• transfer (no raising of main verbs in English)
• acceleration (acquisition of finiteness in French
  precedes acquisition of finiteness in English
  i.e. comparable to monolinguals)
• delay (children were all within relevant ranges
  for monolingual children)

50
Paradis Genesee (1996)

• Conclusion
• separate and autonomous acquisition

51
Müller Hulk (2001)

• Indirect transfer of Language Alpha on Language
  A, i.e. crosslinguistic influence
• (see also Hulk Müller 2000 Müller 1998
  Müller, Hulk Jacubowicz 1999 cf. Unsworth
  2003)

52
Müller Hulk (2001)

• Crosslinguistic influence occurs if two
  conditions are met
• (1) interface between pragmatics and syntax,
  i.e. the C-domain, is involved
• (2) there is surface overlap between the two
  languages

53
Müller Hulk (2001)

• Why the C-domain?
• because this area has been claimed to be
  problematic for monolingual L1 acquisition (and
  adult L2 and impaired acquisition)
• Whats surface overlap?
• language Alpha has a syntactic construction which
  may seem to allow more than one syntactic
  analysis and, at the same time, language A
  contains evidence for one of these analyses

54
Müller Hulk (2001)

• Object drop in Romance/Germanic bilinguals
• Dutch/German topic dropDat doe ik niet
• French/Italian non-topic drop

55
Müller Hulk (2001)

• Object drop in monolingual L1 children
• Dutch/German frequent object dropDat heeft
  mevrouw de Wachter gemaakt (Joost 208.19)
• French/Italian object drop, but less
  frequentil le met dans le bans (Lou)

56
Müller Hulk (2001)

• Early stages of acquisition children universally
  license empty objects as empty topics via a
  (default) discourse licensing strategy
• ? object drop meets condition (1) because it
  involves the C-domain

57
Müller Hulk (2001)

• Adult Dutch/German input contains utterances with
  topic-drop
• ? supports validity of discourse licensing
  strategy which child initially adopts

58
Müller Hulk (2001)

• Adult French/Italian input is potentially
  confusing because
• the canonical object position is empty in certain
  constructions
• topicalised object Ça jai vu EC (that Ive
  seen EC)
• implicit object Je sais EC (I know EC)

59
Müller Hulk (2001)

• ? supports validity of discourse licensing
  strategy which child initially adopts

60
Müller Hulk (2001)

• BUT preverbal object clitics are very frequent
• Jean le voit EC (John him sees EC)
• ? evidence against discourse licensing strategy
• ? so much evidence that this strategy is quickly
  abandoned

61
Müller Hulk (2001)

• Both languages provide some evidence for
  legitimacy of empty object positions
• some surface overlap between the two languages
• object drop meets condition (2)

62
Müller Hulk (2001)

• Predictions
• crosslinguistic influence from Germanic to
  Romance
• Romance/Germanic bilinguals should exhibit more
  object drop in their Romance language than their
  monolingual counterparts

63
Müller Hulk (2001)

• Results
• Romance/Germanic bilingual children use object
  drop in their Romance language to a much higher
  degree than their monolingual counterparts, i.e.
  quantitative but no qualitative differences

64
Summary of 2L1A

• Early differentiation between two systems
• No evidence for qualitative differences, i.e.
  bilingual children fall within normal range for
  monolingual children
• Some evidence for quantitative differences
• ? crosslinguistic influence one of hot topics
  in current literature!

65
Whats bilingual acquisition?

• simultaneous vs. successive

the acquisition of two languages from birth
the acquisition of one language before the other
as an adult
as a child
66
Child L2A Definitions

• Child L2er L2er whose first exposure to L2 is
  between ages ?? years and ?? years

lower bound
upper bound
67
Child L2A Definitions

• Lower bound i.e. differentiating child L2 from
  child L1
• McLaughlin (1978 adopted by Lakshmanan 1995)
  first exposure at 3 years old
• arbritrary (as McLaughlin himself admits)
• not all aspects of first language are acquired at
  this point (Lakshmanan 1995 fn. 5) ? definition
  might differ depending on whats to be acquired
• Meisel (to appear) 5 years old
• Schwartz (1992 2003a 2003b) 4 years old

68
Child L2A Definitions

• Upper bound i.e. differentiating child L2 from
  adult L2
• Johnson Newport (1989) 7 years old
• Meisel (to appear) 10 years old
• Schwartz (1992 2003a 2003b) 7 years old
• this issue is part of critical period debate (see
  Singleton 1989 and Long 1990 for overview)

69
Child L2 acquisition

• (i) why look at child L2?
• (ii) comparison with adult L2 acquisition
• (iii) comparison with monolingual/bilingual L1
  acquisition

70
Child L2 acquisition

• (i) why look at child L2?
• (ii) comparison with adult L2 acquisition
• (iii) comparison with monolingual/bilingual L1
  acquisition

71
Why child L2?

• Lakshmanan (1995)
• on the assumption that child L2 is successful,
  child L2 grammars constitute testing ground for
  correctness of specific theoretical proposals
• child L2ers are more cognitively mature and may
  therefore inform the debate between
  linguistic-based vs. cognitive-processing-based
  accounts of developing grammars

72
Why child L2?

• Schwartz (1992, 2003a, 2003b)
• implications for L1 acquisition whether
  linguistic maturation can explain development
• implications for L2 acquisition whether adult L2
  acquisition is constrained by UG
• teasing apart acquisitional patterns across
  various components of language (i.e. syntax vs.
  morphology)

73
Child L2 acquisition

• (i) why look at child L2?
• (ii) comparison with adult L2 acquisition
• (iii) comparison with monolingual/bilingual L1
  acquisition

74
Child L2 vs. adult L2

• Does child L2 share characteristics of adult L2?
• general lack of nativelike ultimate attainment
• initial L1 transfer
• developmental sequences

75
Child L2 vs. adult L2

• General lack of nativelike ultimate attainment
• evidence that child L2ers can become nativelike
  Felix (1985) Johnson Newport 1989, 1991)
  Krashen, Long Scarcella (1979) Larsen-Freeman
  Long 1991)
• evidence that child L2ers do not necessarily
  become nativelike MacDonald (2000) Snow
  Hoefnagel-Höhle (1972) Weerman, Bisschop Punt
  (2003)

76
Child L2 vs. adult L2

• But ultimate attainment might not be right
  criterion for comparison
• .. because with certain L1-L2 combinations,
  necessary input might not be available because of
  L1 transfer

77
Child L2 vs. adult L2

• Initial L1 transfer
• Unsworth (2003, to appear)
• L1 English, L2 Dutch
• SVO utterances
• Nijntje gaat niet plukken de bloem
• produced by L2 children and adults

78
Haznedar (1997)

• spontaneous data from 4-year old Turkish-speaking
  boy acquiring L2 English
• arrival in UK at 311
• (mostly) Turkish only during first two months
• first systematic exposure to English at 41 (2.5
  hrs per day, 5 days per week at nursery)
• first recording at 43

79
Haznedar (1997)

• VO vs. OV
• L1 Turkish OV i.e. objects/adverbials appear
  before verbs
• L2 English VO i.e. objects/adverbials appear
  after verbs

80
Haznedar (1997)

• Results
• Erdems initial utterances were OV
• Investigator Are they playing? Erdem yes,
  ball playing
• Investigator Where are we going now? Erdem
  Newcastle going

81
Haznedar (1997)

• Results
• Clear evidence of switch from OV to VOsamples
  1 8 91 (21/23) XV utterancescf. samples 9
  22 98.6 (861/873) VX utterances

82
Child L2 vs. adult L2

• Developmental sequences
• evidence for similarities between child L2 and
  adult L2 Cancino, Rosansky Schumann (1978)
  work cited in Schwartz (1992) Unsworth (2003, to
  appear) Weerman et al (2003)
• evidence for differences between child L2 and
  adult L2 McDonald (2000) Weerman et al (2003)

83
Child L2 acquisition

• (i) why look at child L2?
• (ii) comparison with adult L2 acquisition
• (iii) comparison with monolingual/bilingual L1
  acquisition

84
Child L2 vs. child L1

• Does child L2 share characteristics of
  monolingual child L1?
• nativelike ultimate attainment
• developmental sequences/patterns

85
Child L2 vs. child L1

• Nativelike attainment
• evidence that child L2ers can become nativelike
  Felix (1985) Johnson Newport 1989, 1991)
  Krashen, Long Scarcella (1979) Larsen-Freeman
  Long 1991)
• evidence that child L2ers do not necessarily
  become nativelike MacDonald (2002) Snow
  Hoefnagel-Höhle (1972) Weerman, Bisschop Punt
  (2003)

86
Child L2 vs. child L1

• Similar developmental sequences/patterns?
• Haznedar (1997) child L2 ? child L1 because of
  L1 transfer
• Unsworth (2003, to appear) L2 developmental
  stages L1 transfer stage child L1 stages
• Hilles (1991) child L2 child L1 because both
  use Morphological Uniformity Principle cf.
  Lakshmanan (1989) child L2 ? child L1 because
  MUP is inoperative in child L2

87
Child L2 vs. child L1

• Similar developmental sequences/patterns?
• Grondin White (1993) evidence for DP, IP and
  probably CP in child L2 but whether this is
  (dis)similar to child L1 crucially depends on
  analysis of child L1 data
• Gavruseva (2000) child L2 ? child L1 because
  distributive and interpretive patterns found for
  Root Infinitives in child L1 are not found in
  child L2
• see also Prévost (2003)

88
Child L2 vs. child L1

• Summary
• evidence is inconclusive either way
• need more child L2 data
• need carefully controlled child L2child L1
  comparisons

89
2L1A vs. child L2

• vs.

Bilingual First Language Acquisition (2L1A)
Child Second Language Acquisition
90
2L1A vs. child L2

• Could be used to
• disentangle crosslinguistic influence from L1
  transfer
• determine role of cognitive/problem-solving
  factors in (L1/L2 acquisition)
• determine cut-off point between simultaneous and
  successive bilingual acquisition, thereby making
  it less arbitrary

91
2L1A vs. child L2

• Problematic aspects of such a comparison
• individual variation in input/sociolinguistic
  conditions as well as (problem in 2L1A as it is)
• finding enough subjects with correct language
  combination and background
• very few studies WE NEED MORE DATA!!!