decision-making. For example, response inhibition upon reading the word car
following the image of a book – as our experimental instructions would
dictate – would imply anticipating that a subsequent adjective (e.g. red or
blue) may be the decisive factor determining the participant’s response to the
trial. However this sequence (car red or car blue) violates the adjective–noun
order in English and monolingual participants fail to form expectations,
despite their knowing that this configuration exists in the experiment.
Unlike the English monolingual group, the Welsh-English bilingual
group was open to the possibility of further information being provided by
an adjective following the noun. Bilinguals anticipated the fact that critical
information may be provided by an upcoming adjective and this led to N2
amplitude increase in the noun-first mismatch condition. This constitutes
evidence of cross-language syntactic activation: as noun–adjective is the
canonical order in Welsh, bilingual participants found it easier than English
monolinguals to anticipate this order in an English sentence.
Our findings are consistent with views on sentence parsing that assign
permeability to the language boundaries in bilinguals (Kroll & Dussias,
2005). A considerable body of research indicates that recognition of linguistic
information in bilinguals may not be language-specific. For example, during
reading in one of the two languages, information from the bilingual’s other
language can also be activated (e.g. De Groot et al., 2000; Van Heuven et al.,
1998). Evidence from studies on lexical and syntactic processing indicating
that both languages are potentially active in bilinguals (see Frenck-Mestre,
2005; Kroll & Tokowicz, 2005) has led to the incorporation of this permea-
bility into models of bilingual processing. For example, the Bilingual
Interactive Activation model of word recognition (Dijkstra & van Heuven,
2002) proposes nonselective activation of orthographic and phonological
properties of a given word in both languages when that word is presented
visually to bilingual individuals.
The present study shows that nonselective activation can also be found
at the more abstract syntactic level. That is, when a bilingual individual is
processing visual language information in a target language, the syntactic
properties of the non-target language are also activated. Although our
research does not directly disambiguate between competing models of sen-
tence processing (e.g. MacDonald, 1994) it gives a new dimension to the
debate of bilingual sentence processing: unlike the vast majority of previ-
ous studies, which focus on second language learners/late bilinguals (e.g.
Bourguignon et al., 2010; Gillon-Dowens et al., 2009; Hahne, 2001; Hahne
& Friederici, 2001; Tokowicz & MacWhinney, 2005), the present study
examined highly proficient, early bilinguals. Consequently, the interaction
between the two languages that was revealed by our experimental results
is free from familiar, persistent questions such as the role of sequential
language learning and relative language dominance (i.e. how properties of
L1 influence processing of their L2). Our results suggest that permeability
Juggling Two Grammars 227