What About the Input?: The Diachronic Connection to Early Bilingual Outcome Differences

1
What About the Input? The Diachronic Connection
to Early Bilingual Outcome Differences

• Jason Rothman
• University of Iowa
• University of Newcastle upon Tyne
• March 18th, 2009

2
What I want to do

• MAIN GOAL
• Go over some of some of my and collaborative work
  on the acquisition of Inflected Infinitives in
  different contexts, highlighting the role the
  input (here quality as opposed to quantity) as a
  source of competence outcomes/ differences.
• IN DOING SO
• Problematize the concept of the monolingual
  language X as well as the use of monolingual
  comparative norms for the case of bilingualism.
• Add to a more precise conceptualization of what
  incomplete acquisition is.
• Suggest how the study of child L1 and heritage
  language acquisition is a nice crossroads for
  sub-disciplinary collaboration within linguistic
  inquiry.

3
Background Who are Heritage Speakers?

• Simultaneous bilinguals who speak a family and a
  societal majority languages well as adults (maybe
  even equally)
• Simultaneous bilinguals who are clearly dominant
  in the majority language, but have some (maybe
• good) knowledge of the family
  language.
• Sequential bilinguals who acquire the majority
  language L2 as young children, retaining their L1
  well
• and speaking the L2 well.
• Sequential bilinguals who acquire the majority
  language L2 as young children, loosing most of
  their
• L1(perhaps entirely), but speak the L2 well.
• Sequential bilinguals who acquire the majority
  language L2 as adolescent/young adults, retaining
  their
• L1 well and speaking the L2 well.

4
Introduction Facts (see Montrul 2008 and works
cited within)

• Heritage language grammatical performance/knowledg
  e differs from monolingual norms to various
  degrees and in various domains.
• Heritage Speakers often show incomplete or
  partial knowledge as opposed to an utter lack of
  knowledge.
• Heritage language competences can differ
  significantly from one another, whereby some are
  much more proficient holistically (and in various
  domains) than others.

5
Introduction Why?

• Attrition
• Non-pathological attritionthe erosion of
  previously acquired
• linguistic propertiesis often cited as the
  likely cause of these
• differences.
• Incomplete Acquisition
• Montrul and others offer another explanation of
  ARRESTED
• DEVELOPMENT that, in a way, sidesteps the
  attrition problem
• Heritage speaker children acquire the majority
  language, but
• fail to acquire the home minority language
  completely.

6
Introduction Assessing Previous Whys

• Without longitudinal data, it is very difficult,
  if not impossible, to differentiate a posteriori
  between attrition and incomplete acquisition
  since they both have similar consequences for
  heritage language competence.
• Both attrition and incomplete acquisition make
  explicit and implicit assumptions about the type
  of input to which heritage speakers are exposed.

7
Introduction Emerging Questions

• (a) Is SOME of the incompleteness of heritage
  speaker knowledge due to differences (dialectal,
  contact convergence/leveling, etc.) in the input
  TYPE they receive (i.e. not amount)?
• (b) If so, (how) can this be differentiated from
  outcomes as a result of true incomplete
  acquisition (which subsumes attrition)?
• (c) Should it just be subsumed under the label
  incomplete acquisition?

8

• INCOMPLETE ACQUISTION

9
Incomplete Acquisition A Broad Definition

• In my view, incomplete acquisition and L1
  attrition are specific cases of language loss
  across generations. What I broadly refer to as
  incomplete acquisition (for lack of a better
  term) is a mature linguistic state, THE OUTCOME
  OF LANGUAGE ACQUISITION THAT IS NOT COMPLETE OR
  ATTRITION IN CHILDHOOD. Incomplete acquisition
  occurs in childhood when some specific properties
  of the language do not have a chance to reach
  age-appropriate levels of proficiency after
  intense exposure to the L2 begins.
• (Montrul, 2008).

10
Incomplete Acquisition What is included and what
is not?

• Does the broad definition include
  incompleteness delimited by input differences?
• YES, if its main focus is that of a comparative
  outcome, then any type of divergence from the
  established comparative norm qualifies as
  incomplete. This does not mean that all causes
  of incomplete acquisition are thus assumed to
  be the same. Input differences would merely
  qualify as one contributing factor to the same
  outcome.
• Maybe? . . . but for those who fail to see the
  incompleteness in such acquisition outcomes,
  the label is misleading (for some properties, one
  is talking about the complete acquisition of
  properties that are dialectally different).
  Although this amounts to a terminological
  difference, there are many sociolinguistic
  implications that such a label ignores.
• Incomplete acquisition occurs in childhood, when
  some specific properties of the language do not
  have a chance to reach age-appropriate levels of
  proficiency AFTER INTENSE EXPOSURE TO THE L2
  BEGINS. (Montrul, 2008).

11
Incomplete Acquisition Differentiating the
sources of the same outcome

• Attrition tends to not affect the core grammar
  (e.g. narrow syntactic knowledge).
• Incomplete Acquisition can affect the core
  grammar as well (representationally and in the
  sense that some of the related syntactic
  properties were never fully acquired), but
  vestigial knowledge is still expected (increased
  variability/optionality at least at certain
  proficiency levels). Greatest effects, however,
  at interfaces.
• Input Differences if input does not provide the
  triggers for particular grammatical properties,
  then we should see clear effects in the narrow
  syntax of these properties, resulting in, at
  best, chance knowledge of these properties and no
  production.

12
Incomplete Acquisition A Narrower Definition
(Pires and Rothman 2008 to appear)

• Incomplete Acquisition is the outcome of child
  bilingual language acquisition that is
  comparatively different from normal monolingual
  acquisition when the input provides the triggers
  for full convergence of the properties being
  comparatively examined. This would then subsume
  attrition and much of what is considered already
  considered incomplete acquisition.
• Missing-Input Competence Divergence is the
  outcome of child heritage bilingual language
  acquisition that is comparatively different from
  normal monolingual acquisition when the input
  DOES NOT provide the triggers for full
  convergence of the properties being examined.

13
Incomplete Acquisition The Power of a Label

• Although the outcomes are the same, there are
  important reasons to assign different labels on
  linguistic and sociolinguistic grounds
• For the latter case, the word incomplete is
  misleading as there is nothing incomplete about
  the lack of convergence on properties whose
  triggers are absent from the input available to
  heritage speakers.
• If so, this requires a re-evaluation of why
  monolingual baselines are used when heritage
  speakers do not receive the same input how could
  the outcomes be the same?
• To call this case incomplete acquisition has
  important sociolinguistic consequences it would
  mean that language contact-dialects which are
  emerging, if not established varieties already,
  are somehow inherently incomplete as opposed to
  simply different.
• Since this is not limited to properties in
  contact dialects, but also obtains due to
  differences in monolingual vernacular dialects
  (as opposed to standard) (e.g. Rothman 2007
  Pires and Rothman to appear), such a label would
  also apply to monolinguals who lack particular
  knowledge compared to a normative sample (always
  an educated one), even though that is the only
  variety they speak.

14

• Diachronic Change and Heritage Language
  Acquisition

15
Diachronic Change and HS Acquisition

• Language change is a natural phenomenon.
• In the case of contact situations, the process is
  accelerated and shifts occur in the direction of
  the dominant language (Romaine 1988, 1995 Labov
  1994. 2001).
• In all cases, language change occurs on the basis
  of incomplete acquisition in a sense first at
  the level of the input available to children who
  grammaticalize/solidified the changes as they
  naturally acquire the new forms (see Lightfoot
  1999 and others).

16
Diachronic Change and HS Acquisition

• General Implications for Heritage Speaker
  acquisition
• The emergence of new varieties in contact
  situations means that the input heritage speakers
  are exposed to is likely to be grammatically
  different than that of monolinguals, at least in
  particular domains.
• Such differences are not likely to be uniform
  between Heritage Languages (Spanish vs. Russian,
  Portuguese and Arabic in the US), nor will all
  cases of contact be the same, even between the
  same two languages (Spanish contact outcomes in
  Chicago are likely different than in LA, New
  York, Miami, etc.).
• Under all accounts of acquisition (as logic
  dictates), one can only acquire the variety of
  the language one is exposed to. (This is
  compatible for different reasons with cognitive
  and socio approaches.)

17
Diachronic Change and HS Acquisition

• This scenario makes the study of Heritage
  Language Acquisition a promising locus for
  studying empirically the latent acquisition
  predictions of diachronic proposals that are
  complicated by monolingual recovery of lost
  properties via schooling.
• Additionally, it highlights the possibility that
  SOME of the domains for which Heritage Speakers
  show a lack of knowledge compared to monolingual
  counterparts is best understood as resulting from
  differences in the environmental input they
  receive (as opposed to attrition, for example).
• In such a case, the fact that they are not
  educated in the Heritage Language would mean that
  they have no recourse to recover/acquire some
  properties via exposure to the standard dialect
  (and/or schooling).

18
Diachronic Change and HS Acquisition

• Rothman (2007) sought to test this hypothesis
  examining the syntactic and semantic knowledge of
  inflected infinitives in heritage speakers of
  Brazilian Portuguese, comparing their competence
  to native educated Brazilian monolingual controls
  and advanced L2 learners.
• Pires and Rothman (2008, to appear) looked at the
  same properties (with a modified protocol) in
  monolingual Brazilian Portuguese children from
  the ages of 6-18 in an effort to corroborate
  Rothmans (2007) conclusions.
• Pires and Rothman (in press) used the modified
  methodologies from Rothman (2007) by testing
  European Portuguese heritage speakers, since
  there is no question that all dialects of
  European Portuguese actively have inflected
  infinitives.

19

• TWO EXAMPLE STUDIES

20
Verbal Paradigm of Portuguese
Uninflected Infinitives Personal/Inflected Infinitives Present Indicative
eu falarØ (eu) fal(a)o
(tu) falares (tu) falas
Falar ele ela falarØ você (ele) (ela) falaØ (você)
(nós) falarmos (nós) falamos
(vós) falardes (vós) falais(
(eles) (elas) falarem (vocês) (eles) (elas) falam (vocês)

•  Normal Infinitive

21
Uninflected Infinitives

• Portuguese has two types of infinitives
• Non-finite Morphological Infinitives
• Falar/Beber/Sair
• Have no tense/mood specification nor do they have
  person/number specifications.
• Subject to properties of obligatory control
• Cannot be used as matrix clause predicates,
  incompatible with the complementizer que
• Can appear as embedded questions and relative
  clauses
• Possible with A-movement only (subject raising
  and passives, but NOT wh-movement and
  topicalization)

22
Inflected Infinitives

• Have no tense/mood specification, but they do
  have person/number specifications (Raposo 1987,
  Quicoli 1996, Pires 2006)
• Subject to properties of non-obligatory control
  (Pires 2001, 2006)
• Cannot be used as matrix clause predicates,
  incompatible with the complementizer que
• Cannot appear as embedded questions and relative
  clauses
• Possible with A-movement only (wh-movement and
  topicalization, but not subject raising and
  passives)

23
Non-finite vs. Psuedo-finite

• - tense/mood
• - person/number
• - matrix predicates
• - after que
• A-movement
• - A-movement
• obligatory control
• compliments of PP
• compliments of factive and epistemic predicates
• compliment of volitional predicate

• - tense/mood
• person/number
• - matrix predicates
• - after que
• - A-movement
• A-movement
• - (non)obligatory control
• compliments of PP
• compliments of factive and epistemic predicates
• - compliment of volitional predicate

• Obligatory control restricts interpretation

• Non-obligatory control restricts interpretation

24
Finite vs. Psuedo-Finite

• tense/mood
• person/number
• matrix predicates
• after que
• - A-movement
• A-movement
• - compliments of PP
• compliments of factive and epistemic predicates
• - compliment of volitional predicate

• - tense/mood
• person/number
• - matrix predicates
• - after que
• - A-movement
• A-movement
• compliments of PP
• compliments of factive and epistemic predicates
• - compliment of volitional predicate

25
Why study Inflected Infinitives?

• Pires (2006) (see also Lightfoot 1991 and others)
  maintain that the grammatical properties of these
  forms are no longer part of BP colloquial
  dialects.
• If exposure to/education in the standard dialect
  is a deterministic variable, then studying adult
  HS BP knowledge
• Provides an empirical test case for the
  diachronic proposal
• Can demonstrate nicely how some HS and
  monolingual competence differences are delimited
  by input one might otherwise take for granted is
  available to both.

26
Rothman (2007)

• Rothman and Iverson (2007) tested for the
  syntactic distribution of inflected infinitives
  (i.e. how they differ from finite forms and
  regular infinitival forms) and preformed a
  Context Sentence Matching experiment, which
  tested for properties of control (Hornstein,
  1999 Pires 2001, 2006) that differentiate
  inflected from uninflected infinitives. They
  demonstrated across a GJT (with correction), that
  advanced adult L2 learners of BP and educated
  native monolinguals had full syntactic and
  semantic knowledge of inflected infinitives to
  the same degree.
• Rothman (2007) used the same tests to gauge BP
  Heritage Speaker performance and thus knowledge
  of inflected infinitives.

27
Rothman (2007)

• L2 AS Participants (n17)
• NS Participants (n19)
• (from Rothman and Iverson 2007)
• HS Participants (n11)
• All were highly proficient speakers
• None received ample formal education in BP
  (although one did attend one year of school in
  Brazil)
• 8 of 11 were born in Brazil (all moved to the US
  before 8 most before 4 range 1-8)
• Both parents were Brazilian and BP was the
  preferred language of familial communication

28
GJT with Correction (n5 each type)

• Inflected infinitives as complements of factive
  matrix verbs
• Inflected infinitives as complements of
  declarative matrix verbs
• Inflected uninflected infinitives as embedded
  interrogatives / relative clause
• Inflected infinitives in matrix clauses
• Inflected infinitives w/the complementizer que
  

29
Results
30
Individual Results
31
Context Sentence Matching Task (n10)

• Ellipsis Contexts
• Sloppy reading Uninflected, Inflected
• Strict reading Uninflected, Inflected
• Spilt Antecedent Interpretations
• with Pro not grammatical
• with pro grammatical

32
Results
33
Individual Results
34
Discussion

• For the most part, BP Heritage Speakers have no
  knowledge of the syntax and semantics of
  inflected infinitives. In many cases, it appears
  as if they treat them as finite forms, allowing
  them in matrix clauses, after que and as
  embedded interrogatives.
• So, they do not allow agreement morphology to
  specify person and number independently from
  tense.
• As expected, they have full knowledge of
  uninflected infinitival use (syntactic
  distribution and semantic interpretations).
• These results are consistent with the diachronic
  proposals of loss of inflected infinitives that
  cannot be verified by testing adult BP speakers
  (see Pires and Rothman, to appear, for how this
  bears out in child monolingual BP)
• Provides evidence in favor of the position that
  some competence differences between monolingual
  adults and heritage speakers has to do with input
  type.

35
But..

• Is it possible that inflected infinitives are
  acquired and attrited?
• Is it possible that they are available in the
  input, but for some reason not acquired (due to
  the majority language influence that does not
  allow person/number to be distinct from tense?).
• If so, why do they demonstrate no knowledge as
  opposed to partial knowledge?

36
Pires and Rothman (2008)

• Tested European Portuguese Heritage speakers
  (n16) to a group of native EP monolinguals
  (n10)
• Modified version of Rothmans (2007) tasks.
• GJCT (adjusted for dialect and additional
  properties added such as compliments of PP and
  more contexts with finite morphology to ensure
  they make a three way contrast between finite
  forms, inflected and uninflected infinitives)
• Context Matching Task context for obligatory
  and non-obligatory c-commanding antecedents were
  added.

37
GJT with Correction (n4 each type)
38
Context Interpretation Task (n3)
39
Discussion

• The HS performed exactly like the native controls
  at the group and individual level
• Like the BP HSs they were not educated in
  Portuguese.
• This evidence is further corroboration for
  Rothmans (2007) conclusions
• (a) HSs seem to be a good experimental test
• case for dynamic diachronic change
• proposals
• (b) incomplete acquisition outcomes in
  particular
• domains can result from the
  complete acquisition of
• properties different in the input to
  which HSs are
• exposed.

40

• Implications and Future Direction

41
What About the Input?

• Viewed purely as an outcome of comparative
  difference, the term incomplete acquisition
  already covers this.
• However, this label is not clearly appropriate
  for the present case in that it does not sit well
  with the established idea that..
• Incomplete acquisition occurs in childhood, when
  some
• specific properties of the language do not have a
  chance to
• reach age-appropriate levels of proficiency AFTER
• INTENSE EXPOSURE TO THE L2 BEGINS. (Montrul, in
  press).
• In this case the intense exposure to the L2 is
  not the deterministic factor
• that results in the outcome of comparative
  incompleteness, but HSs who
• largely lack formal training in the heritage
  language have no way to recover
• the properties lost in colloquial dialects that
  form their primary linguistic
• data.

42
What About the Input?

• Everyone grows up hearing many different
  languages. Sometimes they are called dialects
  or stylistic variants or whatever, but they are
  really different languages. It is just that they
  are so close to each other that we dont bother
  calling them different languages. So everyone
  grows up in a multilingual environment. Sometimes
  the multilingual environment involves systems
  are so unlike that you call them different
  languages. But that is just a question of degree
  it is not a question of yes or no.
• (Chomsky, 200059)
• What about literacy/schooling?
• If one way that we come to acquire different
  grammars (stylistic variants, registers,
  dialects) is via exposure to standard dialects
  imparted in schooling, then it stands to reason
  that HS will lack knowledge of the properties
  that form part of those dialects.
• Not differentiating between the monolingual
  dialects we are tapping or assessing, some HS
  competence differences will emerge, not because
  of attrition and incomplete acquisition as a
  result of intense exposure to the L2, but
  because vernacular input does not have all the
  properties exemplified that we expect in the
  competence of monolinguals who are in fact
  multidialectal speakers of their native language.
• So, SOME comparatively incomplete knowledge can
  result in HS not having the same internal
  grammars (not being able to tap a different
  stylistic grammar) at their disposal to
  accomplish some experimental tasks like educated
  monolinguals.

43
What About the Input?

• Pires and Rothman (2008 to appear) highlight how
  using a criteria of outcome difference compared
  to educated monolingual competence expectations
  forces us to use the label incomplete acquisition
  for cases of monolinguals that fail to acquire
  these structures as well (uneducated BP speakers
  should also shown similar patterns, yet they
  speak only Portuguese). They offer the label
  input-delimited competence divergence for this
  case.

44
What About the Input?

• This is NOT to suggest that attrition or true
  incomplete acquisition do not conspire to explain
  many (most?) of the differences observed in
  Heritage Speaker competence.
• Input-delimited competence divergence cannot
  explain obvious optionality and variability. It
  cannot account for gradient knowledge correlated
  to overall proficiency.

45
What About the Input?

• To be fair, Portuguese as a heritage language is
  in a privileged position to tease apart these
  variables (considering the dialectal differences
  within the monolingual forms).
• Although this would be much more difficult for
  Spanish, it is not impossible.
• Trying to map the properties of the emerging US
  dialects that serve as the primary linguistic
  data of Heritage Speakers is a necessary first
  step.
• E.G. Do HSs use subject pronouns differently due
  to individual attrition of discourse pragmatic
  features (for processing or whatever reasons),
  does intense exposure to English arrest the
  pragmatic development of HS Spanish despite input
  that exemplifies otherwise or has English
  influence already caused a shift in US emerging
  Spanish dialects themselves such that the input
  provided to HSs leaves no other possible outcome
  than comparative divergence as compared to
  monolinguals?

46
What About the Input?

• Future research that does just that will benefit
  from tapping resources from different sub-fields
  of empirical linguistic inquiry such as
  diachronic, sociolinguistic and formal
  acquisition methodologies.
• Gathering production corpora from the actual
  sources of HS input (across several generations
  of contact if possible), as well as empirically
  testing the Spanish use/competence of the
  properties we investigate from the input sources
  (not taking for granted that the sources continue
  to provide input that matches what is known of
  monolingual varieties) can help us to determine
  the best source for different instances of
  incomplete acquisition.
• This is especially important for interface
  properties like pronominal subject distribution,
  since we do not fully understand how they are
  grammatically represented or come to be acquired
  even under optimal conditions, they seem to be
  subject to 1st generation attrition ( rapidly and
  pervasively).
• While grammatical attrition for 1st generation
  speakers is largely thought to be rare, if
  occurring at all, for narrow syntactic
  properties, it might be possible (cf.
  Cazzoli-Goeta, Rothman and Young-Scholten 2008
  submitted Domínguez, in press Domínguexz and
  Rothman 2009 Pérez-Leroux et al. to appear ).

47
What About the Input?

• If we can identify the properties that have
  changed in emerging heritage languages, what does
  this mean for intervention?
• Do we attempt to recover these properties through
  education?
• Do we allow the natural process of language
  change to take place and embrace the
  differentness as opposed to incompleteness of
  emerging heritage language forms?

48

• Thank you for enduring!
• Special thanks to
• Martha Young-Scholten for the invitation
• Thanks to colleagues who have offered significant
  feedback and in some cases helped in data
  collection (and whether or not I liked it!)
• Acirsio Pires Silvina Montrul
• Joyce Bruhn de Garavito Elena Valenzuela
• Laura Domínguez Martha Young-Scholten
• Marcela Cazzoli-Goeta Liliana Sánchez
• Masha Polinsky Kim Potowski
• Roumyana Slabakova Pedro Guijarro-Fuentes
• Bill VanPatten Glaúcia Silva
• And to my students without whose help I could not
  possibly do this
• Michael Iverson, Tiffany Judy, Jeff Renaud,
  Lauren Reynolds, Gonzalo Campos, Jennifer
  Cabrelli-Amaro

49
Selected References

•  
• CHOMSKY, N. (2000)The Architecture of Language.
  New Delhi Oxford University Press
• DOMÍNGUEZ. L. (in press). L1 attrition and
  modified input Spanish in contact in two
  different
• communities
• LIGHTFOOT, D. (1991). How to set parameters
  Arguments from language change. Cambridge, MA
  MIT Press.
• LIGHTFOOT, D. (1999). The development of
  language Acquisition, change and evolution.
  Oxford Blackwell.
• MONTRUL. S. (in press). Incomplete Acquisition in
  Bilingualism Re-examining the Age Factor.
  Amsterdam John
• Benjamins.
• PIRES, A. (2001) The syntax of gerunds and
  infinitives Subjects, Case and control. PhD
  Thesis, University of Maryland,
• College Park.
• PIRES, A. (2006). The minimalist syntax of
  defective domains Gerunds and infinitives.
  Amsterdam John Benjamins.