Chapter 1 the study of collocations

Table 39.  Nemenyi's multiple comparisons tests of mean rank differences for the blank filling data - Group 1 
TYPES 
1 4  5  11 23 24 27 33 34 36 37 30 31 29 
1. Noun Prep 
 
---  6.68* 5.65* 3.87* .45  2.016 2.89  1.956 6.49* 5.97* 3.65* .115  1.104 .099 
4. Prep Noun 
 
 
---  1.022 2.807 6.23* 4.66* 3.79* 4.72* .187  .709  3.02* 6.56* 5.57* 6.78* 
5. Adj Prep 
 
 
 
---  1.785 5.20* 3.64* 2.769 3.70* .835  .313  2.005 5.54* 4.55* 5.75* 
11. SV(O) Prep O 
 
 
 
 
---  3.42* 1.858 .984  1.918 2.62  2.098 .22  3.75* 2.77  3.97* 
23. SV(O) Adverbial 
     
--- 
1.566 
2.44 
1.506 
6.04* 
5.52* 
3.20* 
.335 
.654 
.549 
24. SV(O) wh-word 
      
--- 
.874 
.06 
4.47* 
3.95* 
1.638 
1.901 
.912 
2.115 
27. Verb Noun (creat.) 
       
--- 
.934 
3.60* 
3.08* 
.764 
2.775 
1.786 
2.98* 
33. Verb Adverb          
--- 
4.53* 
4.01* 
1.698 
1.841 
.852 
2.055 
34. Noun Noun           
--- 
.522 
2.84 
6.37* 
5.39* 
6.59* 
36. Prep Det Noun           
--- 
2.318 
5.85* 
4.86* 
6.07* 
37. Phrasal Verb             
--- 
3.53* 
2.55 
3.75* 
30. Noun Verb 
            
--- 
.989 
.214 
31. Noun1 of Noun2 
             
--- 
1.203 
29. Adj Noun 
              
--- 
*: Significant at the .05 level 
 
 
296

---

Table 40.  Nemenyi's multiple comparisons tests of mean rank differences for the blank filling data - Group 2 
TYPES 
1 4  5  11 23 24 27 33 34 36 37 31 
1. Noun Prep 
 
--- 4.04* 1.362 2.70* 1.559 1.66  1.793 .798  .421  3.64* 1.931 .196 
4. Prep Noun 
   
---  2.68* 1.335 2.48* 2.38* 2.25  4.84* 3.62* .394  2.112 4.23* 
5. Adj Prep 
   
 
---  1.346 .197 .298 .431 2.16 .941 2.287 .569 1.558 
11. SV(O) Prep O 
 
 
 
 
---  1.149 1.048 .915  3.50* 2.287 .941  .777  2.90* 
23. SV(O) Adverbial     
 
 
 
---  .101 .234 2.35* 1.138 2.09 .372 1.755 
24. SV(O) wh-word 
 
     
--- 
.133 
2.45* 
1.239 
1.989 
.271 
1.856 
27. Verb Noun (creat.) 
 
      
--- 
2.59* 
1.372 
1.856 
.138 
1.989 
33. Verb Adverb   
       
--- 
1.219 
4.44* 
2.72* 
.602 
34. Noun Noun   
        
--- 
3.22* 
1.51 
.617 
36. Prep Det Noun 
 
         
--- 
1.718 
3.84* 
37. Phrasal Verb   
          
--- 
2.127 
31. Noun1 of Noun2 
 
           
--- 
 
297

---

 
298
*: Significant at the .05 level 
Table 41.  Nemenyi's multiple comparisons tests of mean rank differences for the blank filling data - Group 3 
TYPES 
1 4  5  11 23 24 27 33 34 36 37 29 28 
1. Noun Prep 
 
---  2.74* 1.217 1.534 1.284 2.66* 1.884 4.61* 1.017 3.73* 2.506 1.339 1.572 
4. Prep Noun 
 
 
---  1.528 1.211 1.461 .078  .861  1.866 1.728 .994  .239  1.406 4.31* 
5. Adj Prep 
   
 
---  .317 .067 1.45 .667 3.39* .2  2.522 1.289 .122 2.78* 
11. SV(O) Prep O 
 
 
 
 
---  .25  1.133 .35  3.07* .517  2.205 .972  .195  3.10* 
23. SV(O) Adverbial 
 
    
--- 
1.383 
.6 
3.32* 
.267 
2.455 
1.222 
.055 
2.85* 
24. SV(O) wh-word 
 
     
--- 
.783 
1.944 
1.65 
1.072 
.161 
1.328 
4.23* 
27. Verb Noun (creat.) 
 
      
--- 
2.72* 
.867 
1.855 
.622 
.545 
3.45* 
33. Verb Adverb   
       
--- 
3.59* 
.872 
2.105 
3.27* 
6.18* 
34. Noun Noun   
        
--- 
2.72* 
1.489 
.322 
2.589 
36. Prep Det Noun 
 
         
--- 
1.233 
2.4 
5.31* 
37. Phrasal Verb   
          
--- 
1.167 
4.07* 
29. Adj Noun 
 
           
--- 
2.91* 
28. Verb Noun (era) 
 
            
--- 
*: Significant at the .05 level 

---

4.2.2.3.4  Implicational Scaling for the Blank Filling Data by Groups 
 
A matrix was compiled for each group, consisting of the types of collocation 
included in the blank filling test, ranked from most accurately answered by all 
subjects in each group to least accurately answered, and the subjects in each group 
ranked in order of their accuracy of response to all types of collocations, from  
subjects responding accurately to the most types to subjects responding accurately 
to the fewest types.  As with the translation data, the 80% accuracy criterion was 
used.  These matrices are given in Appendix O and summarised in Figures 15, 16, 
and 17 below. 
 
For Group 1, the coefficient of reproducibility was .95, and the coefficient of 
scalability was .5.  Even though the coefficient of reproducibility for the blank 
filling data for Group 1 is at the level necessary for this implicational scale to be 
considered valid, the coefficient of scalability is below the recommended level of .6 
(see Andersen 1978; Hatch & Lazaraton 1991:212). 
 
For Group 2, the coefficient of reproducibility was .95, and the coefficient of 
scalability was .61.  The implicational scale for this set of data was found to be 
significant and the items on the scale scalable. 
 
For Group 3, the coefficient of reproducibility was .928, and the coefficient 
of scalability was .68.  The implicational scale for Group 3 was found to be 
significant and the items on the scale scalable. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1 

---

0
.1
.1
.2
.2
.2
.3
.4
.4
.5