Spanish Object Expression under Incomplete L1 and L2 Acquisition

Spanish Object Expression under Incomplete L1 and L2 Acquisition

Silvina MontrulUniversity of Illinois at Urbana-Champaign

Harvard University

November 5, 2007

2

Acknowledgements• Center for Advanced Studies, UIUC• Arnold Beckman Award from UIUC Campus

Research Board• RAs and colleagues

Rebecca Foote Melissa BowlesSilvia Perpiñán Brad DennisonDan Thornhill Alyssa MartoccioSusana Vidal Lucía Alzaga

3

Reality of Today’s L2 classrooms (Spanish and LCTLs)

Two Types of Adult Language Learners• Typical L2 learners (late bilinguals)• Speakers of ethnic-minority languages (early

bilinguals)

Increasing numbers of heritage language speakers are turning to typical L2 classrooms to learn, regain, or maintain skill in the heritage language.

4

L2 classrooms• Having learners with different

linguistic/cultural backgrounds in the same class poses serious challenges for teachers.

• How do we make sure that the linguistic and cultural needs and goals of both L2 learners and heritage language learners are met?

5

Heritage Language Practitioners

• Hold the belief that, in general, heritage language (HL) learners are a very heterogeneous group (even within a language)

• L2 learners and HL learners are different• In many ways, HL learners know “more” than L2

learners who start learning the language from zero.

• L2 learners and HL learners should be placed in different classrooms (tracks)

6

Yet• Any pedagogical practice must be informed by a

deep understanding of what L2 learners and HL learners have and do not have in common.

• Basic systematic empirical research on the linguistic and cognitive abilities of heritage language learners using experimental methods is only just emerging.

(HSs have been the domain of sociolinguistics)

7

Lynch (2003), Valdés (1997, 2006)

• So far, research on heritage language teaching and learning has proceeded atheoretically.

• Heritage Language Acquisition is in need of a theory.

8

Valdés (Valdés et al. 2006, p. 119)

“Second language acquisition theories, as well as traditions now guiding traditional foreign language instruction have little to say about these students and what they should be taught. Existing research on incipient or developing bilingualism in foreign or second languages is of little relevance to teachers of heritage students. Moreover, views about second language (L2) developmental sequences and second language (L2) proficiency hierarchies can contribute little to the understanding of the instructional needs of this population. Taking on the challenge of maintaining and developing existing language resources among immigrants, refugees, and their children will involve a dramatic shift in focus by the profession. The dimensions of this shift in orientation can perhaps best be appreciated by comparing the characteristics of traditional foreign language students with those of the new target population of immigrant students.”

9

Second Language Acquisition• Current theoretical views/theories of (L1 and) L2

Acquisition are VERY relevant to approach and explain the nature of linguistic knowledge in both L2 learners and heritage speakers.

• Heritage language learners afford the field of second language acquisition, and linguistics more generally, a unique opportunity to evaluate, from a different perspective current claims about

10

The basic and essential innate and environmental ingredients for successful, complete language acquisition

OR What is a mature “idealized” native

speaker?At what age does one become a mature L1

speaker and under what environmental conditions?

11

Adult BilingualismLate bilingualism or adult L2 acquisition• L2 acquisition after puberty: foundations of the L1 are fully

established

Early bilingualism• Simultaneous L1 acquisition: 2 languages acquired since

birth or before age 3• Sequential or child L2 acquisition: L2 acquisition before

puberty: foundations of the L1 are established

12

Typical Approach to adult SLAComparison of child L1 with adult L2

SimilaritiesL1 and L2 learners must construct a linguistic system based on input

DifferencesOutcome of L1 and L2 acquisition (i.e., endstate of linguistic

competence) are different

Child L1: always uniform and complete Adult L2: typically variable and incomplete

13

The Incompleteness Hypothesis(Bley-Vroman 1989; Clahsen & Muysken 1989; Meisel 1997; Hawkins & Chan 1997; Schachter 1990)

• L1 and L2 Acquisition are Fundamentally Different• L1 acquisition is guided by Universal Grammar• Past a critical period, L2 learners no longer have access to

Universal Grammar• L2 learners use general-problem solving cognitive

mechanisms rather than an implicit linguistic mechanism to build a grammatical representation of the L2

Compatible with other cognitive approaches to SLA (e.g., DeKeyser 2000, 2003; Ullman 2001; Paradis 2004).

14

BottomlineIncomplete acquisition and impaired linguistic representations

in the L2 are due to a late age of onset of acquisition. Late age of acquisition has consequences for the linguistic,

neurological and cognitive mechanisms that subserve fast and efficient language acquisition that typically occurs in childhood.

15

Theories of Full Access (White 2003)

Access to Universal Grammar is not subject to a maturational effect.

• Full Transfer/Full Access Hypothesis (Schwartz & Sprouse 1996)

• Full Access Hypothesis (Epstein, Flynn & Martohardjono 1996)

• Missing Surface Inflection Hypothesis (Prévost & White 2000)

Abstract linguistic knowledge is present but not always accessible due to a production or a processing problem.

16

Incomplete Acquisition1. Developing grammars (L1, L2, bilingual)2. Fossilized grammars (L2)

Incomplete grammars (fossilized) are common in early bilingualism as well

e.g., many bilingual speakers of etnnic-minority or heritage

languages fail to acquire age-appropriate linguistic competence in the heritage language (their L1).

What are the linguistic characteristics of Heritage Language Acquisition?

How does it compare with L1 and L2 acquisition?

18

HS look like L1 learnersEarly exposure to the language

Naturalistic setting (aural input)

Abundant input

Control of features of language acquired very early in life (phonology, some vocabulary, some linguistic structures)

Developmental errors

Outcome is successful and complete

Fossilization/stabilization does not occur.

No clear role for motivation and affective factors to develop linguistic competence

More complex structures, vocabulary, and pragmatic aspects of language developed at school after age 5 when metalinguistic skills emerge.

19

HS also look like L2 learnersLate exposure to the language

Instructed setting (aural and written input)

Varying amount of input

Developmental and transfer errors

Grammar may be incomplete

Outcome is variable proficiency.

Fossilization is typical

Motivation and affective factors play a role in language development

Experience with literacy and formal instruction

20

L1 acquisition L2 acquisition

HL acquisition

21

In GENERAL, HL learners(Valdés 2000, Carreira & Kagan 2007)

• Have good oral comprehension of the language• May be able to speak the language to different

degrees• Have good pronunciation• Have lexical gaps• Make grammatical errors• Poor to minimal reading and writing skills• Self-identify with their ethnic community• Have a strong interest in learning more about

their language

22

Heritage speakers• Received input during the Critical Period• Yet, input in middle-late childhood may have been

insufficient to develop full linguistic skills in the heritage language

(limited literacy)• Cases of L1 attrition or fossilized L1 acquisition

23

Research questions

(1) Do Spanish heritage speakers have some advantages over post puberty L2 learners?

(2) If advantages are found, are these global or selective, i.e., only found in certain grammatical domains and language skills?

24

Au et al. (2002), Knightly et al. (2003)Study of Spanish language overhearers (i.e.,

HS) and typical late L2 learners

Advantages for HS were found in phonology (VOT production) but not in morphosyntax.

25

VOT Results

Native speakers

Heritage speakers

L2 learners

Word initial/p, t, k/

19.3 19.6 *36.2

Word medial /p, t, k/

22.4 23.4 *31.2

26

Accent ratings (max 5)Native speakers

Heritage speakers

L2 learners

voiceless/p,t,k/

4.3a 3.5b 2.2c

voiced/b,d,g/

4.4a 3.3b 2.7c

27

Knightly et al. (2003)

91.8 98.6

63.6

5062.5

52

0102030405060708090

100

GJT Narrative

native speakersheritage speakersL2 learners

No differences

28

Other FindingsAu et al (2002), Knightly et al (2003) Studied very low proficiency Korean and Spanish heritage speakers and L2 learners.Advantages for HS were found in phonology (VOT production) but not in morphosyntax.

Håkansson (1995)Swedish expatriates and L2 learners of Swedish.Swedish expatriates compared to native speakers on V2 order. L2 learners

produced above 80% SV order instead of V2. gender agreement: L2 learners outperformed the Swedish expatriates

Montrul (2005)Studied advanced, intermediate and low proficiency Spanish L2 learners and heritage speakers’

knowledge of lexico-semantic and syntactic properties of unaccusativity (intransitive verbs)Advantages were found for low proficiency HS.

Montrul (2006) Heritage speakers are better than L2 learners with some aspects of the Null Subject parameter

(word order, agreement)

29

3 Features of Spanish Object Expression

1. Clitic pronouns2. Variable word order 3. Differential object marking

30

1. Clitic PlacementObject clitics precede finite verbs

(1) Patricia vio la novela. Patricia saw the soap opera ‘Patricia saw the soap opera.’

(2) Patricia la vio. vs. *Patricia vio la. Patricia it saw ‘Patricia saw it.’

31

Object clitics follow non-finite verbs(3) Ana canta la canción sin entenderla bien. *Ana canta la canción sin la entender bien.

‘Ana sings the song without understanding it well.’

In restructuring contexts, Spanish clitics can climb up to the finite verb or stay low next to the infinitive. (4) Olga lo puede comprar. Olga puede comprarlo. *Olga puede lo comprar ‘Olga can buy it.’

32

2. Word Order

Postverbal Subjects(5) La mujer lo besa. S-Ocl-V Lo besa la mujer. Ocl-V-S ‘The woman kisses him.’

Topicalizations: Clitic Left Dislocations (6) a.Juan tiene las carpetas en la oficina. S-V-O b. Las carpetas las tiene Juan en la oficina. O-cl-V-S ‘Juan has the folders in the office.’

33

3. Differential Object Marking (DOM)In general, Spanish objects that are [+ specific] and [+animate]

are marked with the dative preposition A.

(7) Juan vio a María. [+animate, + specific] Juan saw A Maria

*Juan vio María ‘Juan saw Maria.’

34

Other objects are unmarked

(8) Juan vio el tren. [-animate, +specific] ‘Juan saw the train.’

(9) Juan necesita un abogado. [+ animate, - specific] ‘Juan needs a lawyer.’

(10) El huracán destruyó una ciudad. [-animate, - specific] ‘The hurricane destroyed a city.’

35

Semantic notions like specificity, agentivity, telicity and topicality seem to play a role in explaining the optionality of A with animate and inanimate objects (Aissen 2003; Torrego 1998; Leonetti 2004).

36

Some Theoretical Assumptions• In Romance languages, object clitics head their

own functional projections (Uriagereka 1995).• Clitic Left Dislocations are part of the left-periphery

of the clause (CP and higher functional projections that interface with pragmatics).

• DOM is marked accusative case (Torrego 1998): inherent, semantically based accusative case; the dative preposition is a functional category. Marked objects move outside the VP.

37

Contact Language: English• lacks clitic projections;• has stricter S-V-O order (although it has topicalizations);• does not have DOM.

38

L1 acquisitionClitic pronouns and DOM with animate and specific

direct objects are acquired before the age of 3 (López Ornat 1994; Domínguez 2003; Rodríguez-Mondoñedo 2006)

Clitic-climbing emerges in Spanish speaking children between 2;00-2;8

(Rodríguez-Mondoñedo, Snyder and Sugisaki 2005)

Topicalizations emerge soon afterwards, by the age of 3. (Grinstead 2004)

39

L2 AcquisitionBeginner and Intermediate L2 learners • make clitic-placement errors (especially if L1 is French)

(Liceras 1986, Bruhn-Garavito & Montrul 1996)• have problems interpreting clitics and objects with

alternative word orders (VanPatten & Cadierno 1993)

• do not recognize DOM in Spanish (Farley & McCollam 2004)

Advanced L2 learners• eventually acquire clitics and clitic placement in Spanish

(Duffield & White 1999)• have residual problems with topicalizations (Valenzuela 2006)

40

Spanish Heritage Speakers (Silva-Corvalán 1994, Montrul 2004)

• Robust control of the object clitic system (accusative and dative structural case)

• Omission of DOM in oral production

Group % a-omission (*NP)Control 0Advanced HSs 6Interm. HSs 21.3

41

Hypotheses

Critical Period PositionHeritage speakers should show evidence of parameter setting

in Spanish, whereas L2 learners should show no evidence of parameter resetting in Spanish.

PredictionsEnglish-speaking L2 learners should have problems with

Spanish clitics, clitic placement, word order, and DOM.Spanish heritage learners should bring knowledge of clitics,

clitic placement, word order, and DOM from childhood.

42

No Critical Period PositionParameter resetting is (eventually) possible in L2 learners

regardless of age of acquisitionPredictionL2 learners and HL learners will have knowledge of Spanish

clitics, clitic placement, word order, and DOM. L2 learners may show transfer effects from English as

predicted by the Full Transfer/Full Access Hypothesis

43

MethodParticipants22 Spanish native speaker (baseline group)67 US born 2nd generation Mexican speakers (acquired

English before age 6)72 English-speaking L2 learners of Spanish

L2 learners and HL learners divided into 3 proficiency groups: Low, Intermediate, Advanced (based on a short proficiency test)

44

Instruments

1. Elicited Production (Oral narrative) clitics, word order, DOM

2. Web-based off-line Grammaticality Judgment Taskclitics, ciltic placement, word order and topicalizations, DOM

3. On-line Visual Picture-Sentence Matching Task clitics, word order

45

Proficiency Scores

Native Speakers L2 Learners Her itage Speakers

Groups

20.00

30.00

40.00

50.00

prof

icie

ncy

Mean 48.5

SD 1.00

range 45-50

Mean 36.88

SD 8.17

range 15-48

Mean 35.34

SD 9.24

range 16-50

46

Oral Narrative: Clitic production

51.3 43.521

48.7 56.579

0102030405060708090

100

native speakers heritagespeakers

L2 learners

Cliticsobject NP

47

100 98.5

75.5

9278.1

20

0102030405060708090

100


L2 learners

accusativedative

48

69.864.5 67.7

15.8 13.3 14.714.322.4

16.3

0102030405060708090

100

nativespeakers

heritagespeakers

L2 learners

finitenonfinitefin+nonfinite

49

60 65.8

13.9

40 34.2

86.1

0102030405060708090

100

nativespeakers

heritagespeakers

L2 learners

no climbingclitic climbing

50

1. Clitics and Clitic Placement

GJT: 90 sentences (45 gram, 45 ungram, 18 sentence types x 5 tokens)

Randomized sentences presented with a 5-point Likert scale underneath

1 = totally ungrammatical5 = completely grammatical

51

Grammaticality Judgment Task

ns differences

Clitics cannot be subjects

1.221.63 1.66

1.25 1.43

2.04

1

2

3

4

5

native speakers heritage speakers L2 learners

accusativedative

52

Grammaticality Judgment TaskClitics precede finite verbs

4.69

4.164.37

1.061.39 1.31

1

2

3

4

5


clitiv-V*V-clitic

ns differences

53


Clitics follow non-finite verbs

1.061.47 1.66

4.91 4.694.37

1

2

3

4

5


*Clitiv-VV-Clitic

ns differences

54


Clitic climbing in reestructuring contexts4.94

4.31

3.84

1.041.24 1.32

4.994.6 4.8

1

2

3

4

5


climbing*Mod-Cl-Infno climbing

Differences for clitic climbing

55

2. Word Order (SVO vs. OVS) Grammaticality Judgment Task

4.99 4.78 4.854.57

3.47

2.42

1

2

3

4

5


L2 learners

S-V-OO-cl-V-S

56

On-Line Visual Picture-Sentence Matching Task

Procedure• Participants were presented with two pictures, A and B, on a computer

screen with a sentence underneath. • The pictures depicted the same action, but with the participants reversed

(e.g., a boy calling his parents vs. the parents calling their son). • Participants had to decide as fast as possible which picture the sentence

described, by pressing A or B on the keyboard. • The target pairs of pictures appeared 4 times, with one of the following

sentences with alternative word orders underneath.• (The task included 20 target sentences (5 of each) and 20

distractor/fillers.)

sentence types: Preverbal Subject: S-V-O and S-cl-V Preverbal Object: cl-V-S and O-cl-V-S

57

On-Line Visual Picture-Sentence Matching Task

Preverbal Subject: Juan llama por teléfono a sus padres. (S-V-O)Sus padres lo llaman por teléfono. (S-Cl-V)

Preverbal Object:

Lo llaman por teléfono sus padres. (Cl-V-S) A sus padres los llama Juan por teléfono. (O-Cl-V-S)

58

Example from the online Visual PMT

A B

Lo llaman por teléfono sus padres. (Cl-V-S)

59

Analysis• Two Factorial ANOVAs with Repeated Measures• One for Accuracy, and one for Speed (RTs)• Independent variables: Group (3), Proficiency Level (3), Word Order

(2), Sentences (4)

Accuracy: No difference between L2 learners and HS main effect for Preverbal argumentSpeed: HS faster than L2 learners and no different from Control main effect for preverbal argumentPreverbal subject sentences faster and more accurate than Preverbal

Object sentences

60

Overall Results

Native Speakers L2 Learners Heritage Speakers

Groups

0.60

0.70

0.80

0.90

1.00

Mea

n A

ccur

acy

(%)

Native Speakers L2 Learners Her itage Speakers

Groups

3000.00

4000.00

5000.00

Mea

n R

T (m

s)

Accuracy: no differences between heritage speakers and L2 learners

Speed: heritage speakers faster than L2 learners

61

On-line Visual SPMTMean Accuracy by Sentence Type

96 96 9599 96 949582

67

97

7163

0102030405060708090

100


S-V-OS-cl-Vcl-V-SO-cl-V-S

No effect by group

Main effect by object position

62

On-line Visual SPMTMean RT by sentence type

2500

3000

3500

4000

4500

5000

5500


S-V-OS-cl-Vcl-V-SO-cl-V-S

63

Summary so far• L2 learners and Heritage Speakers know Spanish

clitics and their placement (GJT). But heritage speakers use/produce more clitics than L2 learners.

• HS have an advantage over L2 learners with clitic left dislocations (GJT)

• HS are overall faster than L2 learners when interpreting Spanish sentences with clitics and alternative word order. They do not differ from native speaker controls in speed.

64

3. Differential Object Marking

Main Study: Oral NarrativeGrammaticality Judgment Task

Follow-up study (Montrul & Bowles, under review)

65

Oral NarrativeAccuracy on Differential Object Marking

99

69.1

52.33

100 97.6 94

0

20

40

60

80

100

nativespeakers

heritagespeakers

L2 learners

animateinanimate

66


1.2

3.54

2.77

1.1

2.27

1.58

1

2

3

4

5

nativespeakers

heritagespeakers

L2 speakers

DOM omissiondouble objects

67

Follow-up Study with Heritage Speakers (Montrul & Bowles, under review)

• Written Elicited production Task • GJT (similar to that of main study)• Instructed Intervention (Pretest-Posttest design)• 70 sentences (different sentence types)

DOM sentences Animate object (grammatical, ungrammatical) Inanimate object (grammatical, ungrammatical)

68

DOM with animate direct objects

Animate Objects4.93

4.644.27

3.94

1.2

3.083.5 3.5

1

2

3

4

5

advanced intermediate low

Native Speakers Heritage Speakers

grammatical

ungrammatical

69

DOM with inanimate direct objects

Inanimate Objects4.96

4.64

4.28 4.2

1.1

3.063.21

3.09

1

2

3

4

5

advanced intermediate low

Native Speakers Heritage Speakers

grammaticalungrammatical

70

Main Study HL learners are worse than L2 learners with DOM

sentences in the GJT (better in production).

Follow-up studyProblems with DOM persist with advanced proficiency

in HL learners.

71

Research Questions

(1) Do heritage speakers have an advantage over L2 learners?

YES

(2) Are advantages selective?

YES

72

Are there advantages for HS over L2 learners?

Clitics Word Order

DOM

findings Yes Yes No

support No CPH(FT/FA)

No CPHFT/FA

neither

73

L2 Acquisition

Results are entirely consistent with the Full Transfer/Full Access Hypothesis (White 1989;

Schwartz & Sprouse 1996) although individual results need to be examined closely)

Full Access: Success with clitics (parameter resetting)Full Transfer: Difficulty with word order and DOM

74

Heritage Language Acquisition

Results are consistent with the Critical Period position HL learners have advantages over L2 learners in use

and distribution of clitics, some complex areas of syntax (word order with dislocations), in oral tasks, and in on-line tasks (processing).

75

Input Mode and Frequency

Clitics Word Order DOM

Taught in L2 classroms?

yes mentioned mentioned

Input frequency?

Frequent in oral and written discourse

More frequent in oral than in written discourse

Frequent in oral and written discourse

76

Types of Linguistic KnowledgeCognitive and Neurolinguistic Approaches to SLA (DeKeyser

2000, 2003; Paradis 2004; Ullman 2001)

Implicit vs. explicit acquired knowledge

The critical period affects implicit competence• L2 learners have explicit, metalinguistic knowledge at this stage of

development (GJT)• HS may be using implicit knowledge acquired in childhood (oral task, RT

times)

Support for the CPH position.

77

PROBLEM

But what about DOM? • Early acquired in L1 acquisition• very frequent in input.

78

Linguistic Interfaces (White 2007)

Grammar: Lexicon + computational system

Computational system interfaces with 1) the articulatory-perceptual system (PF)2) The conceptual-intentional system (LF)

PF and LF are external interfaces

79

Internal Interfaces

• Syntax-semantics• Syntax-morphology• Syntax-morphology-phonology• Syntax-pragmatics

80

Interfaces are Vulnerable

Interface areas between syntax and other cognitive systems (i.e., discourse-pragmatics, lexical semantics) exhibit a great deal of developmental instability.

Adult L2 acquisition: fossilization, indeterminate or incomplete representations

Child bilingual acquisition: cross linguistic influenceAdult L1 attrition: instabilityChild L1 attrition: incomplete acquisition/fossilization

81

Linguistic Complexity

Clitics Word Order DOMGrammatical domain

Syntax Syntax/Pragmatics

Syntax/Semantics/ morphology

Grammatical information

Case/Agrreferentiality

Case/AgrSpecificitytopicality

Inherent Acc CaseDefinitenessSpecificityAnimacyTelicityAffectedness

A little word with a lot of meaning!

82

Child L1 Attrition affects DOMSorace (2004): developmental instability in incomplete acquisition is

related to the complexity of the interfaces

Language attrition processSimplification, loss of restrictions, return to the basics (universals)(consistent with Jakobson’s Regression Hypothesis)

Syntax is resilient, interfaces are vulnerable.

Of all the structural domains tested, DOM is the most linguistically complex phenomenon.

83

Loss of Inherent Case

• Happened in the History of English (Lightfoot 1991)• Has also been attested in Russian as a heritage

language (Polinsky 1997, 2006)• Syntactic convergence with English (probably)• Convergence does not introduce new elements into

the weaker language: it biases the grammar toward the selection, retention and strengthening of structures shared by the minority and majority languages (Bullock & Toribio 2004).

84

Conclusions• The nature of Incomplete, fossilized grammars of adult early

bilinguals in a language minority situation is very complex.• Understanding how the linguistic knowledge of heritage speakers is

different/similar to that of L2 learners will require a variety of experimental methodologies.

• Current theoretical approaches to SLA/Bilingualism that emphasize age of acquisition and nature and timing of input are a good starting point but cannot always predict and explain the patterns of incomplete acquisition found in adult early bilinguals, or why these differ or not from those of L2 learners.

• Explanations may be found in a deeper understanding of the structural complexity of the human language faculty and how it behaves during the normal processes of development and change observed in different learning contexts.

85

Current ResearchMontrul & Bowles (in under review a)Do problems with DOM in Spanish heritage speakers generalize to other

instances of inherent case?

Montrul & Bowles (under review b)Does explicit instruction help instructed Spanish HL learners overcome

incomplete acquisition of inherent dative case?

Montrul, Bhatt & Girju (in progress)Is DOM affected in other heritage languages in contact with English, such as

Hindi (dative marker postposition -ko), or Romanian (marker is pe)? Are patterns of incomplete acquisition similar to those observed in Spanish

heritage speakers?Does the acoustic salience of the object marker in Spanish makes it more

vulnerable to language loss in this language?

YES

YES

86

THANK YOU!

Spanish Object Expression under Incomplete L1 and L2 Acquisition

Documents