Chapter 1 the study of collocations

SV(O) Prep O collocations were also more frequent in the Group 1 
and 2 essays.  The textbook analysis also shows that TWE3, the textbook for Group 
3 (post-intermediate students), contains the least number of collocations of this 
type when compared to the other two textbooks, i.e. the students' production of 
collocations may mirror their exposure to these collocations in their current 
textbook, and not necessarily the incremental growth of collocations from TWE1 to 
TWE3. 
 
There were also collocation types which were used significantly more by 
Group 2 than by Group 1, but they were used less frequently by Group 3.  These 
types are: 12. SV to Inf, 36. Prep Det Noun, 37. Phrasal Verb, 4. Prep Noun, 14. SVV-
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8 

---

ing, and 24. SV(O) wh-word.  Such a phenomenon has been described in previous 
studies as 'backsliding' (Lightbown 1985a).  According to Lightbown, L2 
acquisition is not linear and cumulative, but is characterised by backsliding and 
loss of forms that appeared to be previously mastered.  In this study, Group 3 
subjects are able to use the above collocation types, but they seem to rely less on 
the use of these types than subjects in the lower proficiency levels.  Backsliding has 
been reported in previous developmental studies too (see Hyltenstam 1977; 
Andersen 1978). 
 
There were also collocation types which were not used at all.  These are: 3. 
Noun that, 18. SV possessive V-ing, 22. SVOO, 25. S(it)VO to Inf, 28. Verb Noun 
(eradication), and 32. Adverb Adjective.  The majority of these types are structurally 
demanding and infrequent in everyday English.  According to the BBI, examples of 
these types are:  'We reached an agreement that she would represent us in court', 
or 'it was his desire that his estate be divided equally' (Type 3. Noun that);  'They 
love his clowning', or 'This fact justifies Bob's coming late' (Type 18. SV Possessive 
V-ing);  'It surprised me to learn of her decision' (Type 25. S(it)VO to Inf).  Type 22. 
SVOO collocations consist of a transitive verb and two objects, neither of which can 
be used in the prepositional phrase with to or for, e.g. 'God will forgive them their 
sins', or 'we bet her ten pounds'.  Previous research has also shown that SV 
Possessive V-ing constructions are acquired late (Anderson 1978:97).  Also, SVOO 
constructions were found to be acquired after the more unmarked SVO to O 
constructions (Mazurkewich 1984).  What the above collocation types appear to 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9 

---

have in common is a greater degree of complexity.  Studies in L1 acquisition have 
shown that grammatical complexity is a determinant of acquisition orders (see 
Brown 1973).  Given that collocations in this study are operationalised in terms of 
structurally determined patterns, grammatical complexity could be a factor 
affecting the pattern of results obtained in this study.  Zhang's (1993) study also 
defined collocations in structural terms, and he found that the L2 learners in the 
study avoided, and were unable to produce, the more structurally demanding 
collocations when compared with native speakers (Zhang 1993:126).  These 
collocation types are also structurally different from their equivalent collocations in 
Greek, e.g. Noun that collocations are Noun to [Passive Voice] Infinitive. Laufer and 
Eliasson (1993)  have also reported that L1-L2 difference was the best predictor of 
avoidance in their investigation of the use of phrasal verbs by Swedish and 
Hebrew ESL learners. 
 
Finally, the absence of Type 28. Verb Noun (eradication) and 32. Adverb 
Adjective collocations could be due to the fact that these types are relatively fixed 
(not free combinations) and therefore difficult to acquire.  For example, some Type 
28. Verb Noun (eradication) collocations in the BBI are 'to reject an appeal', 'to 
reverse a decision', 'to rescind a tax'.  The authors of the BBI suggest that 
collocations of this type are arbitrary and unpredictable, i.e. no predictions can be 
made as to why certain verbs combine with certain nouns, therefore L2 learners 
have difficulties acquiring them as they cannot tell why 'make an estimate' is 
acceptable but *'make an estimation' is not (Benson et al. 1986b:258).  For Type 32. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10 

---

Adverb Adjective collocations the BBI includes 'deeply absorbed', 'strictly accurate', 
'sound asleep'.  Previous research has also revealed that adverbs, in particular, are 
difficult for the L2 learner to use appropriately because they typically collocate 
with specific words, i.e. they are fixed (Linnarud 1986:105).  With respect to Types 
3, 18, 22, 25, 28, and 32, it is possible that the subjects in this study have not yet 
reached a proficiency level advanced enough to use such complex, infrequent, 
and/or fixed collocations.  Also, the analysis of the TWE series showed that Types 
3. Noun that, 18. SV Possessive V-ing, and 25. S(it)VO to Inf do not appear in the 
subjects' textbooks, i.e. no tokens of those collocation types were found in any of 
the three textbooks.  Furthermore, only a limited number of Type 22. SVOO, 28. 
Verb Noun (eradication), and 32. Adverb Adjective collocations were found in the 
textbooks (see Table 7, Chapter 3).  It appears that lack of exposure to specific 
collocation types or the low frequency of these collocation types in the subjects' 
textbooks have also contributed to the avoidance of these types by the subjects. 
 
Collocational development across groups was examined by implicational 
scaling analysis of acquisition orders.  The implicational scale for the essay data 
was found to have a significant coefficient of reproducibility which means that a 
subject's performance can be predicted with a high degree of accuracy from that 
subject's position on the scale.  Although previous studies using implicational 
scaling analysis considered a high coefficient of reproducibility as adequate 
evidence for the presence of an implicational scale (see Andersen 1978; Hyltenstam 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
11 

---

1977), in this study the coefficient of scalability was also calculated, to provide 
additional evidence as to the strength of the collocation types as an ordered scale. 
 
The essay data were found to have a low coefficient of scalability (C
scalability
 = 
.33).  It is possible that the large number of items on the scale for this data could 
have reduced the strength of the scale.  Also, the backsliding learning patterns 
which occurred for some collocation types will influence the scalability of the data 
(Hatch & Lazaraton 1991:216).  Even though the statistical validity of the essay 
scale does not reach statistical significance, the relative magnitude of the predictive 
power of the scale cannot be determined, due to the lack of other implicational 
analysis studies in the acquisition of collocations.  According to Davidson (1987) 
"the magnitude of a coefficient of scalability should rightly be judged against 
similar findings in the field" (p. 25).  Since there are no other studies similar to this 
one, it is possible that even scalability of .33 is valid enough as a predictor for the 
order of acquisition of collocation types (Davidson 1987:26).  However, only future 
research and implicational analysis on collocations can verify this.