these were considered shared clusters because bilingual adults may use trilled
and approximant realisations of rhotics in both languages (Jones, 1984).
The patterns found in the study include /r/ realised as [ʋ] in two-element
clusters. This pattern only affected English rhotics, however, yielding [bʋ],
[tʋ], [dʋ], [kʋ] and [fʋ] for /br/, /tr/, /dr/, /kr/ and /fr/, respectively. Rhotics
in two-element clusters were also substituted by /l/ in a few instances, again
only in English, yielding [pl], [kl] and [ɡl] for /pr/,/kr/ and /g r/, respectively.
Only Welsh /sdr/ was realised as [t], only English /str/ as [sp], [tw] and [ɬt].
However, in both languages, [st]-realisations predominated. Similarly, only
Welsh /sgr/ was realised as [s], only English /skr/ as [kr], with [sk] the pre-
dominant pattern in both languages. Together, these results indicate that the
Welsh-English bilingual children have differentiated phonological systems.
Cross-linguistic interactions
As mentioned previously, many studies have revealed cross-linguistic
interactions (Fabiano-Smith & Barlow, 2010; Holm & Dodd, 1999; Paradis,
2001; Yang & Hua, 2010). In terms of the use of language-specific elements
in the ‘wrong’ language, there is little evidence of such patterns in the pres-
ent study. This is probably because the consonant inventories of Welsh and
English are very similar. A possible example of a cross-linguistic effect in the
study is the use of Welsh-specific /ɬ/ in realisations of English /ʃr/ and /str/.
However, this pattern was marginal. Moreover, /ɬ/ is well attested in mono-
lingual English-speaking children (Grunwell, 1987), for example, as a form
of feature synthesis in the production of /sl/.
Evidence for cross-linguistic interactions was also found in connection
with the correlation analyses carried out across shared clusters. The results
revealed significant positive correlations for nine of the clusters tested, and
ceiling effects elsewhere. Hence, overall the more accurately the children
produced the English clusters, the more accurately they produced their Welsh
counterparts. It may well be that these interaction effects underpin the accel-
erated rates of acquisition reported above.
Conclusion
The findings reported here provide a first extensive account of onset
cluster acquisition in bilingual children, and raise interesting questions about
the nature of the developing phonologies in children growing up with more
than one language.
The study also has implications for speech and language therapists as it
provides a benchmark for normal consonant cluster acquisition in Welsh-
English bilingual children, differing in age and dominance. As such, it con-
stitutes an extension of existing normative data on singleton consonant
acquisition in the same population (Ball et al., 2001; Munro et al., 2005).
Although the data provided here should prove useful to clinicians working
with Welsh-English bilingual children, further research is required on other
Speech Learning in Bilinguals 23