Phraseology and Culture in English

 .uk
.au
.nz
.ie
.za
.edu
.us 
.ca 
overall mean 
32.9
12.6
2.9
2.3
1.7
29.6
6.9 10.2 
6 adjective + noun 
31.0
12.8
3.5
2.3
1.8
29.3
8.0 10.0 
4 adverb + V-ed 
35.7
12.3
2.0
2.3
1.5
30.0
5.3 10.5 
overall span 
26-48
8-19
2-7
1-4
1-3 18-41 2-11 7-13 
span: adjective + noun 
26-37
8-19
2-7
1-4
1-3 22-41 7-11 7-11 
span: adverb + V-ed 
27-48 11-13
2
1-4
1-2 18-41 2-11 7-13 
* 
Figures are given as rounded percentages. Percentages sometimes do not 
add up to precisely 100 due to rounding. The total, 100 per cent, does not 
cover the entire web but refers to the total amount of text provided in the 
eight top-level domains investigated. 
While British English, Irish English and the British-derived New Englishes 
were easy to access through the respective country domains (.uk, .au, .nz, .ie, 
.za, for Britain, Australia, New Zealand, Ireland and South Africa), Ameri-
can English, as the de facto default variety of the Web, is somewhat more 
tricky to home in on. The country domain .us, which would seem to be the 
obvious choice, is a little used one and yields insufficient material. Possible 

---

444
Christian Mair 
candidates to fill the gap are the .gov (US government) and .edu (academic) 
domains. The latter was used here because it provides a better mix of texts 
and an initial worry that non-US institutions of higher learning included in 
it might distort the results proved unfounded. Canadian English is again 
represented straightforwardly through its country extension. 
The distribution reported in Table 1 is visualised in Figure 1: 
Figure 1. Baseline distribution of neutral collocations 
The middle line represents the averages calculated for Table 1, that is a 
baseline distribution to be expected in all those cases in which colloca-
tional preferences do not vary across varieties of English. The upper and 
lower lines represent the maximal deviations in either direction observed 
for the ten neutral collocations. The middle band of the diagram therefore 
represents a range of expectation into which a given variety-neutral collo-
cation is very likely to fall. Conversely, collocations falling outside this 
range are likely to be variety-specific, at least in a statistical if not an abso-
lute sense. 
General imponderables concerning web size, composition and growth (see 
Mair 2005) mean that only the most robust statistical distributions permit 
meaningful linguistic interpretation. Table 2 shows how well-established col-
locational markers of Britishness are distributed in the material. (The two 
final bracketed forms are the neutral alternative for one British collocation 
and a further neutral idiom included for control purposes):