Based on the previous results in Section 4.1.3, several phenomena are observed.
First, among the three types of cohesive devices identified in the think-aloud protocols, all the four participants relied most heavily on the lexical cohesion for confirming answers in taking the DS test. Also, among the four subcategories of lexical cohesion identified in the protocols: derivatives, synonyms/antonyms,
superordinates and repetition, repetition was the cohesive device most frequently used by all the four participants. This phenomenon can be easily explained with Morgan and Sellner’s (1980) “content coherence” that the source of coherence lies in the content, which inherently contributes to repeated occurrences of certain words and occurrences of related vocabulary or expressions. From this, a pedagogical
implication could be drawn: In teaching reading comprehension, it would be helpful if the teacher could put emphasis on the identification of lexical relations.
Second, it is found that the percentage of dependence on lexical cohesion for confirming answers in taking the DS test increased as the English proficiency decreased; in other words, readers of lower proficiency tended to depend more on lexical devices, especially repetition, for confirmation of answers in taking the DS test.
Similar observation is also made for excluding alternatives and for the misuse of
cohesive ties as clues. Of the three types of cohesive ties used by the participants to exclude alternatives, the Low participant used lexical ties, all of them repetition, most frequently, and Mid used only repetition. As to the misuse of cohesive devices, the three misuses committed by Mid were all repetition, and seven out of the eight misuses by Low were repetition. As Morgan and Sellner (1980) state that coherent content inevitably contains repeated occurrences of certain words, lower proficiency readers, who are not proficient enough to recognize other types of cohesive ties or to judge the exact function of repeated words in the coherence of the text, may have no choice but to be misled or confused by a string of repeated words. This result suggests that poor readers probably do not focus on the meaning of the text as they read and therefore may fail to discover the misjudgment of cohesive devices, or vice versa, that they may turn to seemingly apparent cohesion, especially repetition, because they cannot figure out what the text is about.
Third, while the low-proficiency readers were more likely to depend on lexical ties for coherence, the proficient readers tended to be more well-rounded and sophisticated in recognizing critical cohesive ties. It is found that the types of cohesive devices identified by Highest and High show greater variety than those by Mid and Low. This result confirms Kavale and Schreitner’s (1979) finding that above-average readers are able to use alternative reasoning strategies more effectively than average readers. In this present study, this difference is most obvious with
References. For confirming answers, Highest identified three types of References, High two, Mid one, and none for Low. The frequencies of References identified by individual participants decreased as the proficiency level decreased; that is, the more proficient a reader is, the more likely references are recognized as significant
cohesion. This finding further echoes Chen’s (2001) claim that the different manifestation of References in English and Mandarin Chinese could result in
Taiwanese students’ inability to detect references as significant cohesion in reading comprehension. Also, the result that Highest and High identified more subtypes of conjunctions as well as greater numbers of conjunctions than Mid and Low could be explained by the results of some studies showing that conjunctions, which usually mark logical relationship in text, are the most difficult type of cohesion (Bridge and Winograd, 1982; Nunan, 1993). These results have suggested, as McCarthy (1991) points out, that references and conjunctions require special attention in ESL
classrooms.
Fourth, the number of cohesive ties recognized as clues to confirm answers roughly corresponded to the proficiency level of the participants, but the same situation did not apply to the exclusion of alternatives, in which case there was a greatest number for Low. One explanation for this discrepancy is that proficient readers are more confident in their discovery of relevant cohesive devices and the answers they decide on. Therefore, there is less need to turn to the technique of exclusion. This redundancy nature of exclusion is also reflected in the fact that both Highest and High were coded five Unspecified items of cohesive devices, as they claimed those to be obvious. Also, the reason why Mid pointed out only three cohesive ties as clues to exclude alternatives might be, as she stated in the response questionnaire, that she had been so used to “thinking-aloud” that redundant
information could be unconsciously omitted in her think-alouds.
Finally, it is found that each of the participants made at least one error by misjudging a cohesive device, and that there is a great difference in dealing with the errors between proficient readers and less proficient ones as well as in the number of errors they made. The number of errors made is found to be in negative proportion with proficiency level. Each of Highest and High made only one error, and they soon corrected the errors by discovering the incoherence as they read along, namely, when
they found out “It doesn’t make sense with that alternative here.” However, Mid made three such errors and had one corrected; Low made eight errors and corrected only four. This result echoes Kavale and Schreitner’s (1979) finding that lower-proficiency readers are more likely than above-average ones to produce decoding errors that violate passage meaning and fail to correct them. In this study, both Mid and Low failed to correct some errors because their awareness of cohesive devices was not sophisticated enough to detect other potential, and even crucial cohesive devices. The inability of the lower-proficiency readers to discover their misjudgment of cohesion prevented them from figuring out the coherence of the text, i.e. making sense out of the text, which resulted in their poorer performance on the DS test. This result has demonstrated that precise detection of cohesive ties can play an important role in good performance on the DS test. As a result, it is highly recommended that students be taught cohesive devices explicitly to perform well on the DS test, as many
researchers have suggested that cohesion mechanism be taught to students, especially poor and ESL readers, to enhance their reading comprehension (Johns, 1986; Nunan, 1993; Staddord, 1991; Wang,1998).