is pervasive – ‘sustained over a period of time’ (Paradis & Genesee, 1996: 3)
- not resulting from temporary performance effects (e.g. random instances
of code-switching). Moreover, as a consequence of true transfer, bilinguals’
developing grammars will differ necessarily from those of monolinguals, at
least temporarily. In order to demonstrate true transfer, therefore, one must
also demonstrate differences in bilinguals’ patterns of development as com-
pared with those of monolinguals learning the same language (cf. the
Functional Interference Hypothesis − Sánchez, 2003, 2004, 2006). Similarly, in
the cases of facilitation and delay, one must demonstrate differences in bilin-
guals’ rates of development as compared with those of monolingual (or diff-
erent sets of bilingual controls – see below) in order to identify behaviour as
true instances of facilitation or delay.
What are the necessary conditions for transfer?
Various models of early bilingual development have attempted to explain
the necessary conditions for cross-linguistic transfer. Although no one model
has been shown to fully account for instances of transfer or lack of transfer,
the following have been proposed as potential conditions for transfer:
- Complexity. According to Hulk and Müller’s (2000) account, a structure
must be sufficiently complex, even for a monolingual L1 learner to
acquire, for it to be vulnerable also to cross-linguistic influence. In par-
ticular, the authors identify those structures that are at the ‘interface
between pragmatics and syntax in the so-called C-domain’^1 (p. 228) as
likely candidates. Kupisch and Bernardini (2007) argued that certain, less
complex properties in one language may facilitate the acquisition of more
complex properties in the other language, resulting in accelerated acquisi-
tion. Assuming the transparency of the Welsh determiner relative to the
highly marked determiner system in German, we could argue that since
there is only one, gender-neutral definite article in Welsh, our bilinguals
may acquire this form in Welsh earlier than in German, and that their
acquisition of determiners in German will be quicker than for monolin-
gual German speakers. - Ambiguity/structural overlap. Hulk and Müller’s (2000), Döpke’s (2000) and
Gathercole’s (2007) accounts similarly argue that transfer will occur
across shared commonalities in the surface structure (cf. the Functional
Convergence Hypothesis – Sanchez, 2003, 2004). However, Gathercole
(2007) further argues that transfer is more likely among broad abstract
rules that apply to both systems (e.g. the category of ‘gender’), and is less
likely to occur with lexicon-specific patterns (cf. ‘deep’ vs. ‘surface’ trans-
fer – see, e.g. Sabourin et al., 2006 for discussion relating to gender in L2). - Linguistic typology. A number of studies have suggested that cross-linguis-
tic transfer is highly likely and more pervasive in related language pairs,
Cross-linguistic Influence and Patterns of Acquisition 51