Recurrent word-combinations in contrast 179
So far we have deliberately avoided using the term phraseology. The rea-
son for this is captured in the oft-quoted statement made by Cowie (1998: 210):
“Phraseology is a field bedevilled by the proliferation of terms and by conflicting
uses of the same term ...” In the same vein Granger and Paquot (2008: 27) note
that phraseology is a field with a “highly variable and wide-ranging scope” and is
associated with “a vast and confusing terminology”.
Although we will continue to use the less loaded term (recurrent) word-combi-
nations, we will give a more precise description of the kind of word-combinations
we are interested in by referring to Gries’ definition of phraseologism:
The co-occurrence of a form or a lemma of a lexical item and one or more addi-
tional elements of various kinds which functions as one semantic unit in a clause
or sentence ... (Gries 2008: 6)Gries’ definition continues with a statistical criterion: “... and whose frequency
of co-occurrence is larger than expected on the basis of chance”. Although the
frequency of the word-combinations in original vs. translated text plays an impor-
tant role in the current investigation, Gries’ strictly statistical parameter does not.
Rather, it is the semantic parameter outlined by Gries that is paramount, viz. the
word-combinations must function as one semantic unit, excluding non-phraseo-
logical n-grams such as he said and. However, following Gries (2008: 8), they do
not need to be semantically opaque, i.e. semantic non-compositionality is not a
criterion, which means that phraseological n-grams such as the old man and by
the way may be considered. Both are meaningful and function as semantic units,
but only the latter is semantically opaque.
It follows from the bootstrapping method applied here that the word-com-
binations under study will be uninterrupted strings. However, as will be shown
below, some of them give indications of patterns that approach colligations or
collocational frames (cf. e.g. Renouf & Sinclair 1991).
- Material and method
Traditionally, corpus-based contrastive studies using bidirectional corpora have
used search strings consisting of pre-selected lexical items (including single words
and multi-word combinations) in original and/or translated text. In contrast, our
starting-point is undefined word-combinations yielded by way of the bootstrap-
ping method briefly described in the Introduction. Three-word combinations were
chosen as they have been shown to be frequent enough to yield interesting data
(Altenberg 1998) and were thought to reveal meaningful patterns.