Chapter 1 the study of collocations

Table 21.  Post-hoc comparisons for the dependent variable: S-Nodes per T-
Unit 
Comparison 
Mean Diff. 
Fisher PLSD  Scheffe F-test  Dunnett t 
Group 1 vs. Group 2  -2.462 
.691* 
24.614* 
7.016 
Group 1 vs. Group 3  -4.747 
.698* 
89.551* 
13.383 
 
242

---

Group 2 vs. Group 3  -2.285 
.693* 
21.085* 
6.494 
* Significant at .05 level 
 
Table 22.  Means and standard deviations for the dependent variable: S-
Nodes per T-Unit 
GROUP COUNT MEAN STD. 
DEV. 
Group 1 
91 
1.119 
.144 
Group 2 
94 
1.366 
.229 
Group 3 
90 
1.594 
.312 
 
4.1.9  Summary of the Results for the Language Proficiency Analyses 
 
The results of the analyses for the language proficiency measures show 
an overall significant main effect for the factor Group.  No difference was found 
between the three groups in the holistic rating of the essays, even though the 
three groups represent three different levels of language proficiency.  This 
could be due to the nature of the holistic rating since it takes into account not 
just the use of language,  but also the structure of the essay, its organisation, the 
expression of ideas, the explanations and arguments provided by the writer, 
etc.  Therefore the ratings based on the holistic scale may obscure differences 
among subjects that are attributable to language proficiency, which is of most 
interest to this study.  However, this lack of significant differences between 
groups using this measure is counter-balanced by the fact that reliable 
differences were found using the other measures, and these are in line with the 
 
243

---

claim that the different groups are composed of subjects at different levels of 
proficiency, and possibly different stages of development.  
 
As can be seen from Table 16, there is a significant drop in Lexical 
Density for subjects in Groups 2 and 3 compared with subjects in Group 1, and 
for subjects in Group 3 compared with subjects in Group 2.  To this extent, 
increases in proficiency appeared to be related to increases in the number of 
grammatical words used in the essays (as the TLU analysis showed).  As a 
result, lower-level students were grammatically less accurate in their essays (as 
the results from the TLU analysis show), and thus the omission of grammatical 
words (e.g. articles) contributed to a higher percentage score for Lexical 
Density.  In Group 3 where the students are grammatically more accurate, as 
the TLU analysis showed, the lexical density is lower.  These results are also 
consistent with the findings of recent research which showed that subjects of 
lower proficiency levels use more content words, while those of higher 
proficiency levels use more function words , e.g. pronouns, articles, and 
prepositions (Ghadessy 1989). 
 
The results for the dependent variable Words per T-Unit also reflect 
different proficiency groupings (see Table 18).  The higher the level of 
proficiency, the more subordination and embedding the student uses in the 
construction of sentences, and thus the longer the sentences they produce.  This 
finding, it must be noted, is in partial agreement with the finding reported by 
Larsen-Freeman and Strom (1977), who found that the mean length of the T-
Units in the writings of the subjects in their study increased steadily with 
 
244

---

proficiency level, but the statistical analysis performed on their data did not 
yield significant differences.  Larsen-Freeman and Strom conclude that length 
of T-Units is still "a viable contender on which to base an index of 
development" (Larsen-Freeman & Strom 1977:132).  
 
The results for the dependent variable Error-Free T-Units do show 
significant differences between the three groups (see Table 19).  Although these 
differences support the claim that the three different groups reflect different 
proficiency groupings, the direction of the difference is in contrast to the 
findings of previous research.  In line with Larsen-Freeman's findings (1978) it 
was expected that more proficient subjects would use more Error-Free T-Units 
than less proficient subjects.  However, the present findings show that subjects 
in Group 1 use significantly more Error-Free T-Units than subjects in Group 2, 
and these subjects in turn use significantly more Error-Free T-Units than 
subjects in Group 3.  The present results could be due to the fact that subjects in 
this  study are simply trying harder to produce more complex syntax than less 
proficient subjects.  It is certainly true that  subjects in the present study are not 
at a sufficiently advanced level to make no mistakes in their writing, since the 
subjects in Group 3, who have had the longest period of instruction in English, 
and who are older by one and two years on average than subjects in the other 
groups, are only at a post-intermediate level.  In Larsen-Freeman's study 
subjects were from a larger range of proficiency levels (5 groups), from subjects 
that were of very low proficiency and needed a great deal of ESL instruction 
(Group 1) to subjects that were advanced enough not to need any more ESL 
 
245

---

instruction (Group 5).  Even though Larsen-Freeman does not report the post-
hoc comparisons for the Error-Free T-Units measure, it is apparent from the 
percentages reported in her paper that it is at the advanced level that subjects 
singificantly use more Error-Free T-Units, e.g. there is a 15% increase in the 
amount of Error-Free T-Units used by the advanced learners in group 5 (see 
Table 23).