Phraseology and Culture in English

5. Conclusion 
What the present study has shown is that previous work on lexico-gram-
matical variation in international Standard English needs to be complemented 
by the systematic corpus-based study of collocational variability. Empiri-
cally, such work fills an important gap in existing descriptions. Methodol-
ogically, it provides a bridge between traditional descriptive work on varie-
ties of English, which has focussed on phonetic, lexical and morphosyntactic 
variation, and more recent research, which has broadened the scope to in-
clude the language-culture interface (as is done, for example, in Wierzbicka’s 
cognitive-semantic approach to variation). After all, collocational prefer-
ences emerge in text and discourse and are therefore much more direct reflec-
tions of a community’s attitudes and pre-occupations than isolated sounds, 
words or constructions. Ultimately, the systematic corpus-based study of 
collocations should be developed into one central element of a principled 
description of variability in world English on the discourse level – a neces-
sary and long overdue complement to the many existing descriptions with a 
structural orientation. 
With regard to the practicalities of corpuslinguistic research, the present 
investigation has shown that collocations tend to occur in frequency ranges 
that test the limits of current practice. Even for fairly common phenomena 
such as the diathetic uses of see discussed in Section 4, standard one-million 
word corpora such as the various national components of ICE yield barely 
enough material. For the majority of the collocations investigated, even a 
mega-corpus such as the BNC proved inadequate, and web-material had to 
be “domesticated” for the purposes of linguistic research in ways which, 

---

Varieties of English around the world
457
while effective, are not always fully satisfactory in terms of traditional phi-
lological standards in corpus linguistics. 
In its analysis of a small judgment sample of collocations, the present 
study has necessarily remained exploratory. It has demonstrated the impor-
tance of collocational patterns in variety differentiation, but it has merely 
been a first step on the way towards the ultimate goal – the comprehensive 
and maximally automatised profiling of collocational patterns in corpora. 
Such profiling will not only add to our knowledge of variation in English, 
but also lead to breakthroughs in corpus technology. It is, for example, a 
promising strategy to adopt in the automatic identification of types of web 
texts on purely language-internal criteria – one of the current challenges in 
the field, and an area in which the solutions found will have repercussions 
far beyond linguistics. 
Notes
1. I would like to dedicate this paper to my colleague, collaborator and friend 
Günter Rohdenburg, Paderborn, on the occasion of his 65th birthday on 28 
July 2005. 
2. Thus, even the contrast between the “British” participle got and the American 
form gotten (for most non-auxiliary uses), which is the closest one probably 
gets to a grammatical indicator of variety status in present-day English, has a 
characteristic leak. The presence of gotten in a text rules out British origin, 
but the “British” usage can easily be attested in contemporary American texts. 
A search for the strings “have got(ten) + *ed” (with * representing any se-
quence of characters following a space) in the Longman Corpus of Spoken 
American English yielded four instances of got and gotten each. The examples 
do not suggest an explanation to the effect that clear get-passives (e.g. have
got pushed over) prefer got and adjectival participles gotten. The material con-
tains instances of I wouldn’t have gotten arrested (gotten in get-passive) as well 
as you […] have got acquainted (got and adjective). Further grammatical re-
gionalisms usually cited in textbooks – from the mandative subjunctive to vari-
able uses of prepositions in collocations of the type different from/to/than – 
present an even more confusing picture, because statistical tendencies in re-
gional distribution are much weaker. 
3. These have been empirically corroborated in numerous corpus-based studies: 
cf., e.g., Kennedy 1998: 194 or Hundt 1998: 106 for prepositional usage after 
different.
4. Wolf 2001 and 2003 could be cited as two further efforts to integrate the cogni-
tive and corpus-based lines of research. Wolf 2001 proposes a cultural-cognitive 

---

458
Christian Mair 
model of community to account for the “Cameroonianness” of Cameroonian 
English and then moves on to study its supposed reflections in corpus data. The 
problem that I see in his approach is the directness of the posited link. The ar-
gument all but short-circuits observations on points of linguistic detail with a 
cultural model which remains very general and diffuse, being defined by “three 
basic elements of African spirituality: the sanctity of life, the role of spirits 
and ancestors, and the relation between illness, misfortune and sin” (p. 198). 
However, it must be added, if only for fairness’ sake, that like almost any other 
attempt at modelling the language-and-culture interface, Wierzbicka’s has not 
escaped severe criticism either – see Ramson 2001. 
5. The 
OED (s.v. mickey) classifies it as “chiefly British”, gives a first citation for 
1948, and further illustrations from British and Australian sources. 
6. A search for be bothered in the Longman Corpus of Spoken American Eng-
lish yields 12 hits, of which a mere two instantiate the idiomatic use of CAN-
NOT be bothered in question. The corresponding figures for the spoken-demo-
graphic BNC are 133, with c. 9 out of ten instantiating CANNOT be bothered.
7. For a detailed discourse analysis of the use of surely, mainly on the basis of 
BNC data, see Downing 2002. 
8. This is rather late, as the spoken-demographic material of the BNC, which was 
sampled a few years earlier, contains three examples. The roughly contempo-
rary and comparable Longman Corpus of Spoken American English has eight. 
9. See Biber et al. 1999: 990 for a more comprehensive definition and illustra-
tion of the concept. 
10. The term focussing is used here in the technical sense that it has in Le Page/ 
Tabouret-Keller’s 1985 model of standardisation. 
11. The grammaticalisation of see as a passive marker is documented for other lan-
guages in Heine and Kuteva’s (2002) World Lexcicon of Grammaticalization.
English usage in this regard is strikingly similar to French (e.g. quelqu’un s’est 
vu refusé quelque chose), as any web search for phrases such as “s’est vu(e) re-
fusé(e)” or “se sont vus refuses” or even technically non-existent misspelled 
variants such as “s’est vu refusée” or “s’est vu refuser” will testify. Whether 
English copies French, French copies English, or we are dealing with an inde-
pendent development in the two languages cannot be determined here. 
12. Searches for see this happen and see this happening yielded 3 and 5 matches, 
respectively. See happening and see happen were attested 11 times each. 
13. For details of the project see http://www.ucl.ac.uk/english-usage/ice/. Of an 
envisaged total of c. 20 national sub-corpora, six are currently available. 
14. A further two borderline cases illustrating the idiomatic expression see some-
thing coming might have merited inclusion: 
The PCF, Dr Beng said, has seen the demand for nursery places coming, based 
on these two developments: THE birth rate has been going up since 1987 and 

---

Varieties of English around the world
459
is now about 49,000 a year PCF kindergartens have seen a big jump in the num-
ber of nursery places up from 11,893 in 1991 to 20,825 in 1992.  
He and his staff had seen this new demand for nursery places coming, he said. 

They were excluded here because they represent a clear metaphorical exten-
sion of the visual-perception sense of see.
15. This is the impression gained from the present author’s long experience with 
student writing and the International Corpus of Learner English (ICLE), which – 
admittedly – is not a perfect match for the ICE press texts in size and compo-
sition. 
16. See Appendix III for examples.