is the case, Welsh-English speakers will tend to choose just one ML, and as
suggested above, it may be the Welsh national identity and Welsh-oriented
social network of the speakers we recorded that determines Welsh as the one
they choose. For Spanish-English speakers, however, the fact that both lan-
guages have SVO word order allows for more flexibility in the choice of ML.
Nevertheless, future research will be needed in order to isolate the role of
structural factors of this kind.
Further research could also pursue the role of speaker-based variables,
using data from more speakers, to determine their role in the specific choice
of English or Spanish ML in a community like Miami where this is variable.
We noted above, with reference to Table 6.5, that the speakers in Zeledon1
use mostly Spanish ML, whereas the other speakers use both English and
Spanish ML. This might be attributed to the lower English proficiency of
AME, who according to Table 6.3 did not arrive in the USA until the age of 19.
Table 6.5 also shows that AME produced fewer bilingual clauses than the
other speakers, which could be attributed to a lesser degree of balanced bilin-
gualism. But a study involving more speakers would be necessary in order to
explore more fully the role of speaker-based variables.
Conclusion
Our linguistic analysis of a sub-sample of Welsh-English and Spanish-
English data showed uniformity in the choice of Welsh ML in Wales con-
trasted with variability in the choice of Spanish vs. English ML in Miami.
Participant questionnaire responses were used to conduct a multivariate
analysis of the Spanish-English data as well as to consider the role of con-
trasting community characteristics in Wales and Miami in the choice of the
ML. While the results showed that more data would be needed to pursue the
multivariate analysis in Miami, specific contrasting community characteris-
tics were argued to be related to the choice of ML in the two communities.
The languages related to social identity, social networks and to education
were argued to influence the choice of ML in the two communities.
The next chapter explores language in education in more depth, and dis-
cusses the potential additive aspects to the relationship between bilinguals’
two languages within the bilingual classroom.
Notes
(1) The support of the Arts and Humanities Research Council (AHRC) is gratefully
acknowledged. The Welsh data were originally collected and transcribed by Marika
Fusser, Elen Robert, Peredur Davies, Jonathan Stammers, Siân Lloyd-Williams and
Margaret Deuchar as part of the project ‘Code-switching and Convergence in Welsh’,
awarded by the AHRC to Margaret Deuchar. They are available via TalkBank
(http://www.talkbank.org) in the BilingBank archive and on Bangor Talk
Factors Influencing Code-Switching 137