Simplification processes were also found for fricatives: the dental frica-
tive in /θr/ was replaced by [f], the palato-alveolar fricative in /ʃr/ by [s], [ɬ]
or [h].
Three-element clusters in Welsh were not simplified. However, English
three-element clusters were: /spl/ and /spr/, for instance, were realised as
[spw], and /skw/ as [skɹ]. Also, in a few instances, three-element clusters
reduced to double ones involved simplifications. For example, /spr/ was
realised as [pw], /str/ as [sp], [tw] or [ɬt].
Discussion
A total of 41 Welsh-English bilingual children aged 2;6 to 6;0, half of
whom were Welsh-dominant, the other half English-dominant, were
assessed in a picture-naming task. The data were analysed in terms of the
accuracy with which onset clusters were produced in the two languages. The
results revealed significant age and dominance effects on acquisition.
Cluster acquisition and age
Based on Sander’s (1972) criterion for ‘age of acquisition’, that is, 75%
correct in a given age group, even the youngest children in the study showed
evidence of successful cluster acquisition. Thus, the 3-year-old children were
able to produce most stop + /l/, stop + /w/, /s/ + stop, /s/ + nasal, frica-
tive + /l/ and fricative + /w/ clusters accurately in both languages, as well as
the Welsh-specific cluster /kn/. Clusters containing /r/ and three-element
clusters, on the other hand, were acquired late in both languages, with /sgl/,
/gwr/ and /gwn/ the latest in Welsh, and /ʃr/,/θr/ and /skr/ the latest in
English.
The results not only showed a steady increase in production accuracy
across the age range, but also a number of reversal patterns. For example, the
English-dominant bilinguals aged 3;0 to 3;6 produced English /dw/, /gr/ and
/br/ accurately, but these clusters dropped below the 75% acquisition thresh-
old in subsequent age groups. Likewise, /ʃr/ was acquired by 4;6 by the
English-dominant children, but acquisition was not maintained thereafter.
Fluctuating patterns of this kind have been well documented in the literature
(Munro et al., 2005; Smit et al., 1990; Templin, 1957) and may arise from
reorganisation of the children’s phonological systems (Ingram, 1989).
Alternatively, the patterns observed here may be a manifestation of indi-
vidual variation in the cross-sectional sample of this study.
Rate of acquisition
No previous data on Welsh cluster acquisition are available. Interestingly,
in comparison with English monolinguals (Smit et al., 1990; Templin, 1957),
the bilingual children in the present study exhibited a considerably faster
rate of acquisition. Thus, they managed to acquire all clusters, except /fr/,
/skr/ and /θr/ by 5;0. In contrast, Smit et al. (1990) report that a number of
20 Part 1: Bilingual Speech