These cases indicate that the given learners were able to determine the word boundary
between ‘side’ and ‘down’. The majority of responses also indicated the learners were able
to establish the onset syllable of ‘side’, but frequently perceived a different offset (see also
Field, 2004). They had activated a range of potential word candidates with this correct onset,
but had preferred one of them in favour of ‘side’. While this suggests a process problem, i.e.,
‘side’ was known vocabulary but not recognized in connected speech, it may also have been
a text problem. The idiom ‘let the side down’ is very low frequency (spoken frequency of
0.01 tokens per million (Davies, 20082009)), and it was highly unlikely the learners had
encountered it before. Alternatively, learners were unaware of the potential for ‘side’ and
‘down’ to occur sequentially, which is an uncommon combination. Notably, the learners
persisted with their incorrect choice without regard for semantic appropriacy. The most preferred version of ‘side down’ was ‘size down’ (N=16). Apart from the
possibility that process or text problems were responsible for this decoding error, it is also
very possible that learners’ preference for ‘size down’ was due to an intrusion problem. That
is, ‘size down’ is a loanword collocation which is used in Japanese to refer to a decrease in
something (typically with respect to the downsizing of the workforce). It fitted the given