an interesting area to explore as they constitute one of the most protracted
aspects of phonological development (McLeod et al., 2001) and are commonly
impaired in paediatric clients with speech impairments (Chin & Dinnsen,
1992). The present study aims to contribute to this line of research by inves-
tigating the acquisition of onset clusters in Welsh-English bilingual children
differing in language dominance.
Consonant clusters and their acquisition in monolinguals
Consonant clusters are sequences of consonants produced in temporal
succession without an intervening vowel. They may occur in syllable-
initial position as onset clusters, as in play [plεɪ], or in syllable-final position
in coda clusters, as in ask [ɑsk]. Some languages, such as Greek (Mennen &
Okalidou, 2007), have a large cluster inventory. Cantonese, on the other
hand, only has two consonant clusters, that is, word-initial /kw/ and /kʰw/
(So & Dodd, 1995). Sometimes a distinction is made between tautosyllabic
clusters, that is, clusters occurring within the same syllable, and heterosyl-
labic clusters, that is, clusters occurring across syllable boundaries, as in
dolphin (Barlow, 2003). Furthermore, clusters may be the result of schwa-
deletion, for example, police as [plis], suppose as [spəʊz] (Heselwood, 2008).
The present study only includes tautosyllabic onset clusters that are not a
result of schwa-deletion.
Children’s acquisition of consonant clusters is a protracted process that
may not even be complete by the age of 9;0 (Smit et al., 1990). Nevertheless,
certain clusters may be mastered by children as young as 2;0 (Lleo & Prinz,
1996). Interestingly, word-final clusters tend to be acquired before word-
initial ones (Dodd, 1995). This asymmetry appears to be attributable to
articulatory factors, rather than structural, morphological or frequency-
based ones (Kirk & Demuth, 2005). As the present study focuses on onset
clusters, no further mention will be made of codas (but see Mayr et al., 2014).
Early attempts to produce onset clusters often result in forms that do not
feature in the ambient language. Dyson (1988), for instance, reports that the
most commonly produced cluster of the 2- to 3-year old English-learning chil-
dren in her study was [fw]. By the age of 3;6, 75% of normally developing
children in Smit et al.’s (1990) study were able to produce clusters containing a
stop + /w/, by 4;6 to 5;6 clusters containing /l/, by 6;0 clusters containing /r/,
and by 7;0 clusters containing /s/ and /θr/. The earliest target-like productions
were /kw/ and /tw/, with /skr/ and /spr/ the last to develop.
Children’s early attempts to produce consonant clusters are characterised
by systematic substitution patterns, or phonological processes (Grunwell,
1987), the most common of which are ‘cluster reduction’ and ‘cluster simpli-
fication’ (see McLeod et al., 2001 for details of rarer substitution patterns, for
example, epenthesis, metathesis and coalescence).
Cluster reduction involves omission of one or more elements from a clus-
ter, for example, brown as [baʊn], and appears to follow broadly the same
6 Part 1: Bilingual Speech