abandoned. García and Otheguy (1985) report that by 1984 only four bilin-
gual schools remained of the original 14 created in the sixties. Teachers were
usually monolingual English speakers who taught content through the
medium of English, alongside assistants who delivered some Spanish-medium
tuition. More recently the 2001 NCLB ‘No Child Left Behind’ Act is only
concerned with the acquisition of English by immigrants (García, 2009: 421),
thus reinforcing a monolingual approach in schools in both Miami and else-
where in the USA.
The following section addresses previous research that pertains to our
current study on code-switching practices in the two communities.
Previous Research on Uniformity vs Variability
in CS Patterns
There is by now an extensive literature on theories of CS (cf. MacSwan,
2000; Muysken, 2000; Myers-Scotton, 1993, 2002; Poplack, 1980) but only
a small proportion of it deals with the question of uniformity vs. variability
in CS patterns. This section is dedicated to a review of relevant work con-
cerned with: (i) the application of the MLF to the language pairs under study
and (ii) the influence of speaker-based variables and community characteris-
tics on CS behaviour.
Applications of the MLF to data from the two communities
Applications of the MLF to Welsh-English bilingual data
Deuchar (2006), Deuchar & Davies (2009), Davies & Deuchar (2010) and
Davies (2010) have carried out most of the work on Welsh-English CS, which
has so far been conducted within the MLF framework. (For information
about the historical and sociolinguistic setting of this work, see Deuchar
(2005).) Deuchar (2006) presents an application of the MLF to a corpus of
Welsh-English bilingual conversations. Out of a sample of 163 bilingual
clauses, Deuchar was able to identify a ML in 89% of cases. The vast major-
ity of the clauses had Welsh as the ML (141 clauses). It is notable that only
four bilingual clauses had English as their ML in the data. Davies & Deuchar
(2010), analysing a similar dataset (plus two extra recordings) to Deuchar &
Davies (2009), applied the MLF to 1816 clauses (both monolingual and bilin-
gual), and found that Welsh was by far the predominant source of clause ML
(95.43%). Of the 336 bilingual finite clauses analysed, 99.7% (335 out of 336)
were identified as having Welsh ML, with the remaining bilingual clause
identified as showing possible word-order convergence towards English all
these studies, then, Welsh is shown to be the most frequent source of clause
morphosyntax in bilingual clauses produced in informal speech by Welsh-
English bilinguals.
Factors Influencing Code-Switching 117