revealed that the Spanish-English community have a more bilingual social
network that includes a balance of Spanish and English. The Welsh-English
participants on the other hand tend to have a predominantly Welsh-language
social network. A Welch two sample t-test was conducted on the partici-
pants’ social network scores (t = 8.38, df = 182.39, p < 0.0001) showing that
there was a significant difference between the mean scores in the two com-
munities. The social network results can be related to ML choice in that the
predominantly Welsh-language network in Wales can be linked to the pre-
dominant selection of Welsh as ML, while the more bilingual network in
Miami can be linked to the more varied choice of Spanish and English ML.
Even though one may assume that the language learned first is closely
related to the matrix language, our data does not support this. However, the
different proportion of bilingual clauses in the three conversations appears
to be related to age of acquisition. Acquiring both languages at an earlier age
seems to be related with a higher proportion of bilingual clauses.
Discussion
In this chapter we set out to identify whether speaker-based variables or
community-wide characteristics would account for the CS patterns found in
Factors Influencing Code-Switching 135Figure 6.8 Social network mean scores for bilinguals