How integrated is

1

• How integrated is
• the integrated
• mental lexicon?

2
L1 and L2 lexicons separation or integration?

• The L2 mental lexicon was formerly seen as
  qualitatively different from the L1 mental
  lexicon and, by implication, quite separate.
• Arguments against such a qualitative difference
  have been put by, e.g., Single-ton.
• Other researchers go further.

3
The case for integration

• Interplay between L2 and L1 mental lexicons is
  demonstrated by research into memory, learner
  errors, learning strategies, bilingualism and
  communication strategies. Such evidence does not
  bespeak full integration.
• Cook (1992) cites the following evidence in
  support of his multicompetence model
• reaction time to a word in one language is
  related to the frequency of its cognate in
  another known language
• morphemic similarities between two known
  languages influence translation performance
• when processing an interlingual homograph,
  bilinguals access its meanings in both their
  languages.
• However, his argument stops at the point of
  talking about intricate links.

4
The case for integration (contd.)

• Dijkstras (2001, 2003) notes that when a
  particular word form is activated similar word
  forms known to the subject are activated also,
  whatever the language affiliation of the words in
  question. The evidence largely derives from
  decontextualized experimental situations and
  often the stimuli were presented in orthographic
  form. These conditions may be favourable to the
  activation of formal neighbours across
  languages. In any case, parallel activation is
  explicable in terms of high levels of
  connectivity between lexicons. Dijkstra accepts
  that languages can be at different levels of
  activation, and proposes language nodes.

5
The case for integration (contd.)

• Franceschini et al. conclude from brain-imaging
  studies that lexical-semantic aspects of the
  processing of all languages known to an
  individual are subserved by the same areas of the
  cortex. Franceschini et al.s inference that
  lexical-semantic processing draws on a common
  system across languages is questionable.
  Neuro-imaging has its limitations. Obler
  Gjerlow (1999 168) comment cortical topography
  is at best the surface component of a
  multidimensional set of systems cortical linked
  with subcortical that enable us to use
  language. Also, every language articulates the
  world differently in terms of lexical/conceptual
  structure, which implies that, if L2 learners
  make use of the L2-specific lexico-semantic
  system, that system must be differentiated from
  that of the L1.

6
The case for integration (contd.)

• Harris argues (a) that both within any given
  individual and within a given speech community
  the continuum of variation is so extensive that
  it makes no sense to talk about a clearly
  demarcated system and (b) that languages blur
  into each other in a myriad of ways. However, the
  individual with more than one language at his/her
  disposal can differentiate between them, even if
  particular elements are sometimes miscategorized
  and or overgeneralized across language
  boundaries, and the psycholinguistic evidence of
  differing levels of activation of different
  lexicons does not sit well with the notion of
  monolithic integration.

7
The case against integration

• The modularity hypothesis sees the mind as
  comprising a number of distinct, specialized,
  structurally idiosyncratic modules (Garfield,
  1987 1), including one devoted to language. At
  least some modularists suggest that part of the
  functioning of the L1 mental lexicon is
  intramodular and at least some hold that any L2
  competence acquired beyond the childhood years is
  extramodular. Taken together, these two positions
  imply that in the case of post-pubertal learner
  of additional languages, the lexical processing
  of such languages proceeds in isolation from that
  of the L1.

8
The case against integration (contd.)

• Another argument relates to formal differences
  between languages. An individual faced with the
  task of working out the morphological structure
  of unfamiliar words refers to the phonological
  composition of familiar items and then
  analogizes. Since languages have highly divergent
  phonological systems, the implication is that the
  search must run through the lexicon of each
  language separately.
• Other evidence comes from studies of language
  loss following brain damage and of aphasia. Lost
  languages may be recovered sequentially. Re
  aphasia, multilinguals may exhibit language
  disorders affecting just one of their languages.

9
The case against integration (contd.)

• A further point is that speakers of two or more
  languages mostly keep their languages apart when
  using them to the extent that where the
  expectation is that language x is being spoken
  but where, in fact, it turns out that language y
  is being used, comprehension may be blocked, even
  where both languages are familiar to the
  individual in question. Many L2 errors seem to
  have little, if any, connection with the mother
  tongue (Duková 1969 19). Cross-linguistic
  influence increases in situations where the
  languages involved are perceived as close but
  this psychotypological dimension runs counter
  to the notion of total integration, because it
  implies selectivity in relation to
  cross-linguistic consultation.

10
The case against integration (contd.)

• Re code-switching, for De Bot Schreuder (1993)
  this provides evidence that both languages are
  continuously activated, though each to a
  different level. They note (1993 212) that
  activation models allow for the possibility that
  words from the non-intended language may always
  slip in. This presupposes intention, selection
  and a degree of separability, while recognizing
  that the system may sometimes break down.
  Code-switching is sensitive to the
  particularities of the languages involved. Such
  sensitivity constitutes further evidence of the
  separation of languages in the mind.

11
The case against integration (contd.)

• Languages may be lost where, e.g, a child migrant
  has an L1 which is different from that of the
  majority of members of the host community, where
  his/her home language receives little/no support
  from the community in question and where his/her
  parents make little/no effort to support it
  either. A more extreme case (Schmid 2002) is that
  of 54 German Jews who emigrated from Nazi Germany
  between 1933 and 1939. The later in the 1930s
  they had left, the more negative their attitudes
  towards German tended to be, and the less
  inclined they were to continue using German. For
  such deliberate rejection of a particular
  language in favour of another to be able to take
  place, the languages in question clearly must
  have separate identities in the individuals
  mind.

12
Different degrees of integration?

• According to Weinreich (1953) in subordinative
  bilingualism L2 word forms are connected to L1
  meanings via primary connections to L1 forms in
  compound bilingualism the L1 and L2 forms are
  connected at the meaning level and in
  co-ordinate bilingualism separate systems of
  form-meaning links exist for each language.
• Different types of relationship between L1 and L2
  may co-exist in the same mind. E.g. De Groot
  (1993, 1995) suggests that where concrete words
  and cognates are stored in a compound manner,
  whereas abstract words and noncognates are stored
  in a co-ordinate manner. Kirsner, Lalor Hird
  (1993 228) suggest that some fraction of the
  second language vocabulary is represented and
  stored as variants of the first language
  vocabulary.

13
Different degrees of integration? (contd.)

• DeGroot and Kroll read the evidence as suggesting
  that subordinative structure is associated with
  low proficiency and compound structure with
  higher proficiency. Woutersens (1997) research
  also suggests that subordinative, compound and
  co-ordinate bilingualism are associable with
  different levels of development. Cieslicka (2000)
  claims that formal-associative and conceptual
  links exist in some measure between the L1 and L2
  mental lexicons in all learners varying in
  strength according to experience. None of these
  positions claims that the L1 and L2 lexicons are
  completely undifferentiated at any proficiency
  stage.

14
Different degrees of integration? (contd.)

• Cook (2003) refers to an Integration Continum
  which favours a single mental system within
  which a balance can be struck between elements of
  a particular aspect of language in a particular
  situation (p.10) and which suggests that the
  relationship of integration versus separation
  varies from component to component (p. 11).
  There is no contradiction between arguing for a
  single mental system while not arguing for
  total integration.

15
Some findings (old and new) from Dublin

• Study I (Singleton Little, 1984) looked at the
  degree to which Anglophone subjects with no
  previous knowledge of Dutch could understand a
  Dutch text and also collected introspective data
  on the process. Some subjects had learned German
  the rest had no knowledge of any Germanic
  language other than English. Subjects with German
  outscored those without German subjects with
  German found the task easier than those without
  a majority of subjects mentioned looking for
  clues from other languages and most who
  mentioned this strategy mentioned only the
  language that was closest to Dutch.

16
Some findings (old and new) from Dublin (contd.)

• Study II (Singleton, 1987) was a case study of a
  beginning adult learner of French whose L1 was
  English and who had some knowledge of Irish,
  Latin and Spanish. Recordings were made of him
  interacting with native speakers of French and
  performing narration/description and oral
  translation tasks. Introspective data were also
  gathered. The subject privileged Spanish as a
  source of transfer over English, Irish and Latin
  he knew Spanish to be typologically related to
  French and was aware of the practical value of
  this.

17
Some findings (old and new) from Dublin (contd.)

• Study III (Singleton Little 1991 Singleton
  1999 Chapter 7) analysed responses from
  university students of French and German to
  C-tests in their respective target languages.
  Subjects had attempted coinages drawing on their
  L1, English, but English-German coinages were
  markedly less frequent than English-French
  coinages, a finding which can be ascribed to
  psychotypological factors considerable numbers
  of English words can after a simple conversion
  process be readily deployed in French
  converting English words into their German
  cognates is typically more complicated.

18
Some findings (old and new) from Dublin (contd.)

• Study IV (Herwig 2004) involved subjects in the
  composition of a story in their L1 and the
  translation of the same story into another
  language. Part of the study had a plurilingual
  dimension. The subjects in this case were 4
  university students 3 of whom had English as L1
  and the 4th, Norwegian who were taking courses
  in German, Dutch and Swedish. These subjects were
  asked to translate their L1 story into German,
  Dutch and Swedish. They used English when
  exploring the semantics of different aspects of
  the translation tasks, but their actual
  cross-lexical borrowings drew predominantly on
  Dutch and Swedish when the translation was into
  German, Dutch and German when the translation was
  into Dutch, etc.

19
Some findings (old and new) from Dublin (contd.)

• All four studies indicate a strong
  psycho-typological factor. Psychotypology
  undermines the notion of a unitary mental
  lexicon, since it implies selectivity in
  cross-linguistic consultation. A possible
  integration-friendly explanation of the data on
  which psychotypological perspectives are based
  might be that a given form simply triggers all
  similar forms available to the subject. However,
  there is evidence that selectivity of
  consultation occurs at a level above that of the
  individual word. A fifth Dublin study yields
  evidence in precisely this direction.

20
Some findings (old and new) from Dublin (contd.)

• Study V (Soufra, 2001) focused on beginning
  English-speaking learners of Modern Greek as a
  foreign language. The instrument comprised
  Greek-English translation tasks and a
  multiple-choice collocation recognition task. One
  finding was that in some instances where the
  relationship between the Greek and English terms
  was very close e.g. galax?aV galaxy,
  auqentik?V authentic learners failed to make
  the connection. Soufras explanation is that
  Anglophone learners of Greek perceive the
  distance between Greek and English as being
  relatively large, and that they do not therefore
  pay as much attention to similarities between
  Greek and English forms as one might expect.

21
Conclusion

• Evidence offered in favour of the notion of an
  integrated mental lexicon supplies powerful
  arguments for a high degree of cross-lexical
  connectivity, but does not entitle one to dismiss
  evidence of differentiation.
• The studies revisited above suggest that when we
  encounter new languages we make judgments about
  their relationship to languages we already know
  and in processing terms exploit the lexical
  resources in those already established languages
  accordingly, prioritizing languages which we deem
  to be most useful. Such prioritization seems
  incompatible with a claim that lexical knowledge
  is radically unitary.