國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
the instrument (e.g., the program’s static estimate of MyNewPos (FNR-1597)) or the source (e.g., manufacturing’s share of production (G08-460)).
4.5 Results of Covarying Collexeme Analysis
The actual usage of of -constructions is attested by incorporating a statistical methodology introduced earlier, covarying collexeme analysis (CCA) developed by Stefanowitsch and Gries (2005). CCA identifies the most frequently found instances of of -constructions in the British National Corpus (BNC) academic subcomponent. It is based on the computation of the collostruction strength of each N1-N2 covarying collexeme pair with Chi Square tests and the Fisher-Yates Exact tests. The results for each pair of N1-N2 found in the 591,000 concordance lines of of -constructions were produced (see Appendix B for top 150 instances). As mentioned earlier in Section 3.2.1, collostruction strength is the log transformation of p-values based on Fisher-Yates exact test. There exist both positive and negative values of collostruction strength, indicating, respectively, attraction and repulsion to the construction. The higher the value the stronger the association between the N1-N2 pair and of -construction. A significance level is reached at p<.001 when the collostruction strength is above 3.000. Table 4.4 shows the top 50 most strongly attracted N1-N2 pairs in the of -construction and a word-to-construction delta p (△P attraction) value greater than 0.50. This value informs us about the probability of observing the words at N1 and N2 in the presence of an of -construction. While the△P ranges from zero to one, the higher the △P value indicates a higher chance of the words being entrenched in the of -constructions (Schmid & Küchenhoff, 2013: 551-2). For the present study, an arbitrary lower-bound limit was set at 0.50, which is the mid-point of the range of
△P.
‧
Table 4.4 Top 50 covarying collexeme pairs attracted to the of -construction in the order of△P attraction
N1 N2 Frequency △P Attraction Collocational Strength
1 Brandon Oakbrook 11 1 55.9
2 Morris Borth-y-gest 10 1 51.2
3 Thietmar Merseburg 11 1 54.8
4 Goff Chieveley 41 1 185.5
5 Jauncey Tullichhettle 29 1 133.7
6 Keith Kinkel 43 1 192.4
7 Bridge Harwich 48 1 207.6
8 Bay Biscay 33 1 142
9 Slynn Hadley 23 0.96 108.9
10 element bounty 21 0.95 61
11 elders Zion 14 0.93 64.6
12 John Gaunt 18 0.9 71.5
13 office censuses 42 0.89 138.1
14 director prosecutions 100 0.88 312.5
15 tip iceberg 21 0.87 80.7
16 state destination 58 0.86 141
17 isle Wight 47 0.85 169.6
18 law gravitation 22 0.84 56.9
19 William Malmesbury 13 0.81 50.4
20 synthesis ltb 29 0.81 94.5
21 analysis variance 92 0.79 195.2
22 breach warranty 44 0.78 105.1
23 defence volenti 18 0.78 51.9
24 expression pre-s1 25 0.78 65.2
25 Donaldson Lymington M.R. 70 0.78 282.3
26 power attorney 35 0.78 86.3
27 republic Somalia 34 0.74 112.4
28 cup coffee 14 0.74 57.1
29 duty fidelity 24 0.73 67.3
30 University Edinburgh 24 0.73 67.2
31 oxides nitrogen 45 0.73 180.6
32 infusion caerulein 21 0.72 67.7
33 rule thumb 32 0.69 89.3
34 number occasions 49 0.68 70.5
35 king Mercians 19 0.68 53.7
36 length stay 38 0.68 96.9
37 chancellor exchequer 40 0.66 154.3
38 sides Atlantic 55 0.64 166.9
39 point departure 39 0.62 81.8
40 incidence discharges 27 0.61 66.9
41 archbishop Canterbury 57 0.6 194.3
Continued on next page
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 4.4 – continued from previous page
N1 N2 Frequency △P Attraction Collocational Strength
42 judgment L.J. 51 0.59 139.8
43 east Suez 16 0.59 52.9
44 Æfgifu Northampton 14 0.58 61.5
45 office Mufti 19 0.58 55.9
46 college physicians 24 0.57 68
47 balance probabilities 35 0.56 89.8
48 Master Rolls 25 0.56 84.7
49 culture-ideology consumerism 26 0.53 109.7
50 Police Metropolis 16 0.52 64.7
For the top eight collexemes, all of which are titles, the value of△P attraction is 1.00, meaning that these pairs exhibit absolute specificity to the of -construction. Such a high value suggests that neither N1 nor N2 can be found elsewhere in another of -construction with another nominal head. These eight collexeme pairs are all proper names for lords and barons. For example, the first pair, Brandon-Oakbrook refers to Lord Brandon of Oakbrook, who was a British judge.
Table 4.4 shows that the CCA results display a variety of combinations ranging from proper names of personal titles (e.g., Lord Brandon of Oakbrook (line 1), King of Mercians (line 35)), organization titles (e.g., University of Edinburgh (line 30), Office of Mufti (line 45)), positional titles (e.g., the Chancellor of the Exchequer (line 37), the Master of the Rolls (line 48)), locative titles (e.g., the Bay of Biscay (line 8), both sides of the Atlantic (line 38) to quantity (e.g., cup of coffee (line 28), a number of occasions (line 34), length of stay (line 36)) and idiomatic expressions (e.g., the tip of the iceberg (line 15), a rule of thumb (line 33), point of departure (line 39)). The table also shows a number of discipline-specific technical jargons such as law of gravitation (line 18), analysis of variance (line 21), and breach of warranty (line 22). In contrast, expressions of quantity fewer instances of everyday language such as cup of coffee (line 28), number of occasions (line 34), and length of stay (line 36) were found from the list. Overall, the results of the CCA analysis suggest that proper names are the most prototypical instances for the of -construction.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
To investigate which N1-N2 pairs are most commonly found in the of -construction, the CCA results were ordered by collocational strengths. Table 4.5 presents the CCA results of the top 50 N1-N2 pairs. The major difference between the previous and current results lies in the fact that the current list measures the degree of association between N1 and N2, whereas the previous one compares the degree of association between the N1-N2 pair and the construction. Instead of being flooded by specific proper names as demonstrated earlier, the constituents of this second set appear to be different. First, proper names are still present but they are mainly organizational titles (e.g., the Court of Appeal, House of Lords, House of Commons, School of Economics, and Department of Health) and positional titles (e.g., the Secretary of the State, Head of the Department and the Director of Public Prosecutions). Second, some of -constructions specify time like towards the end of the 20th century and half of the 19th century. In addition, technical jargons, particularly in the fields of social sciences and law, are plenty including division of labour, the breach of duty/contract, the mode of production, a/the duty of care, the rule of law, and a cause of action. A number of the technical terms involve an action relation as exemplified by breach of duty/contract/peace, exercise of discretion, eradication of H pylori, and analysis of variance. Quantity relations like over a long period of time, a period of 100 years, and the age of 10 years are found in the top 50, whereas other quantity relations like a number of people/ways fall out of the top 50 but can be found within the list of top 150 (see Appendix B). Finally, there are only a handful of fixed expressions like one’s point of view and the state of affairs and some not-so-fixed expressions like quality of life, the desired way of life, and in a variety of ways in the top 50.
Table 4.5 Top 50 covarying collexeme pairs in the of -construction in the order of collocational strength
N1 N2 Frequency △P Attraction Collocational Strength
1 Court Appeal 1335 0.77 Inf
Continued on next page
‧
Table 4.5 – continued from previous page
N1 N2 Frequency △P Attraction Collocational Strength
2 point view 879 0.67 Inf
3 House Lords 876 0.92 Inf
4 Secretary State 817 0.28 Inf
5 House Commons 494 0.88 Inf
6 division labour 336 0.43 Inf
7 end century 334 0.19 Inf
8 breach duty 249 0.42 Inf
9 half century 224 0.13 Inf
10 mode production 215 0.16 Inf
11 state affairs 215 0.43 Inf
12 breach contract 210 0.20 Inf
13 head department 190 0.40 Inf
14 rule law 193 0.06 317.830
15 Director Public Prosecutions 100 0.88 312.470
16 duty care 162 0.13 311.750
17 facts case 132 0.15 293.683
18 Donaldson Lymington M.R. 70 0.78 282.337
19 decision court 159 0.11 279.406
20 years age 167 0.19 278.442
21 exercise discretion 116 0.42 267.457
22 quality life 212 0.08 261.874
23 circumstances case 128 0.14 259.345
24 period time 203 0.11 256.622
25 cause action 165 0.12 251.637
26 colleges education 115 0.06 250.139
27 means production 198 0.14 248.630
28 relations production 125 0.09 243.553
29 breach peace 129 0.35 243.389
30 School Economics 99 0.35 234.830
31 age years 146 0.13 233.026
32 Bank England 101 0.13 230.635
33 sociology knowledge 114 0.07 226.299
34 provisions act 134 0.09 224.288
35 Department Health 135 0.22 222.975
36 course business 150 0.18 221.553
37 eradication H pylori 70 0.34 221.489
38 variety ways 138 0.29 220.386
39 exercise power 149 0.07 219.630
40 provision services 147 0.11 214.247
41 burden proof 78 0.45 210.839
42 Bridge Harwich 48 1.00 207.610
43 way life 219 0.08 207.283
Continued on next page
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 4.5 – continued from previous page
N1 N2 Frequency △P Attraction Collocational Strength
44 period years 156 0.13 207.262
45 contract sale 79 0.26 202.854
46 judgment court 113 0.08 195.636
47 analysis variance 92 0.79 195.151
48 middle century 110 0.06 195.107
49 Department Economics 103 0.36 194.745
50 archbishop Canterbury 57 0.60 194.333