Insights into learners' strategic behaviour while problem-solving: an example using foreign language reading tasks.
Philip Hood
School of Education, University of Nottingham
Paper presented at the British Educational Research Association Conference, Cardiff University, 7-10 September 2000
1.1 Introduction-1: reading in the secondary foreign language (FL) context
Weber (1991) wrote about second language learning in the United States: ‘Learning the spoken form of the second language is primary, learning to read the language is secondary.’ (p101). She added: ‘The possibility that a learner’s knowledge may be confirmed, elaborated or extended through experience with the written language has not been directly addressed.’ (p102). And further, (p108): ‘ The assumption seems to be that reading will follow from knowing the structure of the language and knowing how to read in the first language.’ This view applies also to the context of England where reading during the first three years of FL learning is characterised in a very narrow way. It is seen primarily as a consolidation process for already learned language. Texts are almost invariably very short. Tasks tend to involve formats such as sequencing, true/false statements, grid-ticking. Cognitively more demanding tasks which involve decision-making and problem-solving are much rarer. As a result the motivation to read lacks the stimulus which problem-solving can provide. In addition to this there can be a lack of strategic awareness about reading which results partly from the stigma surrounding the use of English in FL lessons, thereby precluding discussions about learning processes. Such reading approaches might still be elicited by the inclusion of either greater amounts of unfamiliar lexis or a problem-solving task-type with some teacher-modelling. But the reluctance of publishers particularly to move in such directions has made KS3 FL reading in very many classrooms a stagnant process and a wasted learning opportunity.
1.2 Introduction-2: the data
The data for this paper is drawn from a doctoral study which investigated the foreign language reading behaviour of 27 members of a Year Nine comprehensive school French class. The study involved each member of this class in completing two different reading tasks individually and two parallel tasks as part of a group of three or four. As individuals the subjects were asked to complete the reading task in any way they wished, but as far as possible aloud. The class had received some 'think-aloud' training (following the principles outlined by Ericsson and Simon, 1993 and van Someren, Barnard and Sandberg, 1994) before the start of the programme. The sessions were recorded on audiotape. As groups the instruction was to complete the tasks collaboratively but again without any specified approach. In both contexts the researcher played the role of 'live' dictionary, if needed. These sessions were recorded on both audio- and videotape. The study used a range of analysis approaches to investigate the reading behaviour which emerged. These included task performance analysis, strategy use coding, analysis of talk-type and further discourse analysis techniques, including the use of concordancing software.
For the purposes of the paper today I intend to focus principally on strategy use and talk-type analysis with regard to two members of the class (who were together for the group-based task) and one of the task types. This should allow a range of issues to be demonstrated in a more structured format, including the fact that there are qualitative differences between the different analysis methods adopted in the study.
1.3 Introduction-3: FL reading tasks as problem-solving exercises
The second of the two task-types used in the study was a global decision task, (The Four Penfriends task, see appendix 1 for the individual and group variants). In this the reader must determine which of four authors (each with specific writing characteristics) is responsible for the letter. This is allied to a problem-solving process on two levels. Firstly the task itself specifically requires the reader to make a single decision after weighing up evidence. Secondly, because a reader has this aim in mind, the approach to meaning construction (especially in terms of the inference of meaning of unfamiliar lexis and structures) at the word, phrase and sentence level is both more primed and more focused.
1.4 Two subjects
The two subjects selected for this paper are referred to as Subject 3 (male) and Subject 15 (female). A reading test at age 11 had placed them with a reading age of 14+ and 11.6 respectively. A GAP test score obtained a week before the study gave them nearly parallel scores on the same test variant of 38 and 36 respectively. In the individual session Subject 3 subsequently scored very poorly (1 correct out of 8) on Task 1 (a true/false/impossible-to-say [TFI] task) but was the highest scorer (8 details noted) of the 27 subjects on the second task which we are examining in this paper. Subject 15 scored better (3 correct of 8) on the TFI task but noted fewer details (3) on the Four Penfriends (FP) task. During the FP task Subject 3 asked the researcher for 9 vocabulary items, while Subject 15 asked for 6. As further statistical evidence it is perhaps worth noting that in the individual FP task transcripts, Subject 3 had nearly 95% of the word count and 90% of the line count, compared to the researcher, while Subject 15 had 68% of the word count and 74% of the line count. Thus Subject 15 was more reliant on the researcher for vocabulary or procedural advice.
2 Strategy coding
The strategy coding analysis method used was adapted from that proposed by Pressley and Afflerbach (1995), described by McDonough (1999 p5) as having 'enormous implications for the theory of reading in a second language'. Its use gave the facility to compare near-beginner foreign language reading with first language reading in terms of categories of strategies used and not used.
The coding is separated into three or four broad sections (with Monitoring and Activation of cognitive processes both being part of the same section in the original), as in Table 1. Each of the sections listed there is then subdivided into a series of descriptions of individual processing behaviours, as can be seen from the coded transcripts below.
The codes appearing on the left side of Table 1 were entered into one column on the transcripts. This enabled them to be sorted by both code and subject to gain an awareness of the frequency of use of the various categories. The more precise identification of exactly which strategy was in use was entered into a parallel column which was also capable of being sorted for analysis.
Table 1:Key to strategy coding in the transcripts
To demonstrate the information about the reading and problem-solving processes to be gained from the strategy coding analysis, there follows an extract from Subject 3's individual FP task transcript, which can be compared with a similar extract from the transcript of Subject 15 (Extract B). Additionally there is a short extract from the group transcript in which both subjects were involved.
2.1 Extract A - (Subject 3 - FP task lines 37-62)
- 3 And then she's talking about her brother- she says my
- 3 his hair is quite long
- 3 he is very sporty
- 3 He plays football
- 3 for
- 3 the national team
- 3 (( laughs ))
- 3 I think that's what it says
- 3 Um
- 3 um
- 3 What does mais mean, I've forgotten
- R But
- 3 But, but
- 3 he
- 3 something he
- 3 he plays
- 3 something about, I think it's about Londres, what's
- R London
- 3 I thought it was. So it's something about he plays in
- 3 He is
- 3 also
- 3 célèbre pour travail
- 3 à la télévision something about television he likes
- R Devenu means he's become
- 3 He has become also
MC/DR / Paraphrasing text +
Tentative interpretation
- 3 a celebrity on television I think that's what it might
In this extract, focusing on paragraph 2 of the text, Subject 3 uses principally a method of ongoing translation (paraphrasing text) to begin to get an initial reading. He is able to use cognates both to recognise and then infer some probably unfamiliar key words, and is also very prepared to use such inferences in summarising units of meaning before building on them. This constitutes the second common strategy, the use of a tentative interpretation of what he is reading. As part of this process, he stops three times in these 25 lines to summarise, and thereby creates a block of meaning to extend his more global understanding. This process of stating meaning so overtly through text paraphrase and unit of meaning summary acts as a memory spur to allow a level of ongoing meaning construction that he had not managed in the previous (TFI) task. Although we do not have overt evidence of the actual FP task being held in mind, his laughter at line 42 recognises the possible incongruity of a penfriend having a brother who plays for the French national football team. His progression in lines 59-62 is startling from 'he likes watching [ television]' to 'He has become also a celebrity on television', the accurate reading. This would suggest that the meaning construction is being driven by a more global context, which a search for evidence of exaggeration would constitute. We cannot say this is the case, but neither can we rule out the possibility.
On a single word level his four uses of the word something (lines 51-59) in this extract are much more positive than had been the case in his work on the TFI task. Here he is using the word to create a coherent meaning of the rest of a phrase, and therefore as a strategy towards understanding rather than as an indication that the meaning has not been grasped. The notion of pausing to reflect, coded overtly at lines 45/6, also underpins this use of the word something, and signifies concretely continued active involvement with the task.
The strategy coding exercise thus helps us to break down the processing and to see how meaning is constructed. Subject 3 might be expected (considering his reading scores) to manage successfully a task which requires the ongoing sense of a text to be created, section by section, in order that a decision can be made on the problem with a bank of evidence in mind. This analysis shows that the expectations are well-founded and also presents evidence of how this is achieved, ie of what might be termed good reading behaviour.
2.2Extract B - Subject 15 - FP task lines 13-48
Subject 15 does not opt to translate the text as she reads. Immediately before this extract there is an 84 second silence while she reads the majority of the text.
This extract occurs when she has made her decision about the identity of the penfriend author and gives her reasons.
NB (the symbol * = inaudible/incomprehensible speech)
- 15 I'm saying that
+
MC/DRcim / A goal is achieved + Drawing a conclusion
- 15 she's the one who tells all the lies
- R OK [can you give me some reasons ?
- 15 because, yeah ] because
+
MC/DR / Maintaining an hypothesis
+
Explicitly looking for related words concepts or ideas in text
- 15 it says
- 15 something something up there
- 15 I can't remember where it is off-hand *
- 15 It says something
- 15 It says that
- (3)
- 15 her mum
- 15 is having problems
- (3)
- 15 What's these two words mean ?
- 15 That one and that
- R Récemment just means recently
- 15 And what's that mean ?
- R That's that's the word for parliament
- 15 Saying that she's got problems with
+
MC/DR / Restating text
+
Maintaining an hypothesis
- 15 the European Parliament or something
- 15 Making out her family's like dead
- 15 I don't know
- R OK
- 15 And then um
- 15 says her dad is the president of France
- R OK
- 15 and things like that
- R There's two reasons OK. Are there any other
- (7)
- 15 Yeah um
+
MC/DR / Maintaining an hypothesis
+
Explicitly looking for related words concepts or ideas in text
- 15 she
- 15 is really popular
- 15 in Europe
- 15 for something
- 15 there on that line
- R OK
Of the three details she gives to justify her decision that the author is Sacha 2 (who tells outrageous lies), one (her father is French president) is stated unequivocally, while the other two reasons are semi- or very imprecise. This is in contrast to Subject 3 who has assembled meaning much more fully. The fact that he has done this aloud may or may not be significant. (Indeed this is the question for future research which concludes the thesis). We could postulate that if Subject 15 had adopted a similar process, of paraphrasing text aloud, stopping to make tentative interpretations, and attempting to build on meaning as she read, that she would have gained more evidence for her correct task decision. But she may have simply articulated a failure to understand the text. Had she asked for more words, she might have grasped more detail, but we do not know which words she knew and did not know, apart from the six items she requested. In fact we cannot know what her silent reading behaviour consisted of. There is an obvious issue with think-aloud protocols that some subjects do not want to think aloud, especially if this means revealing a lack of understanding.
And here too we see a weakness in the strategy coding system. We have established that the specific range of MC/DR strategies used by Subject 3 are missing here, but the decision-making process is as important as the meaning construction process, and we are seeing sound strategy insofar as Subject 15 is stating a conclusion and then supplying evidence for it. The coding does not differentiate between coherent and less structured argument or reflect obvious complete as opposed to more uncertain understanding of text. This is not to say that strategy coding is not useful. It undoubtedly provides a sound basis for the comparison of reading behaviour. But it needs to be used in conjunction with other techniques as we can show below in Section 3.
2.3Extract C - The opening section of the Group B1 transcript.
This extract is interesting for two reasons. It demonstrates that in the group context there was sometimes much more overtly stated organisational strategy use. A group of four individuals often wished to discuss how they were going to proceed with the reading. Secondly, the reading behaviour already observed in the individual transcripts is immediately replicated here. (The Group session followed five days after the individual sessions). Again the strategy coding system allows this to be compared both quantitatively and qualitatively, and allows also for comparisons between individual subjects' reading behaviour in each context.
- 3 She says hello and then she says I go (2), then
something about the library, (-) I go to
- 2 Shall we start with, like, Sasha One and see if it's ?
- 3 Just read it through because you'll be able to tell (3)
I go (2) / MonPT
+
MC/DR / Own behaviours strategies in processing the text
+
Paraphrasing text
- 15 Well read it to yourself and see what everyone
- 3 What does beaucoup mean ?
Subject 2 had in his individual session tried out a process where he read the FP task text from the point of view of each of the four possible authors in turn (and in this sense was task- rather than text-led, the only one of the 27 subjects to approach the FP task in this way). He suggests this again here at line 2, but is contradicted by Subject 3 who wants to read and translate, and who indeed starts to do this at line 3. Subject 15 suggests the individual silent reading process at line 4 but is ignored by Subject 3 who proceeds to ask for a word. Subsequently if we were to present further sections of the transcript with the strategy coding we would see exactly parallel structures to those of the individual transcript extracts above. Subject 3 (together with Subject 2) continues to try to make sense of what is probably the most difficult of the four texts used. Subject 15 makes a decision early and then only wishes to discuss that decision, not the more exact textual meaning either at word or paragraph level. The lack of a more complete textual understanding is demonstrated by lines 88-102, (See Appendix 2 - B1 Group transcript - FP task), where all four subjects discuss how the letter ought to be framed in terms of prior knowledge or opinion about penfriend letters rather than on the basis of the text before them. But crucial here is that Subject 3's approach which had proved successful for him individually does not predominate in the group context. Subject 15 partially succeeds in enforcing her preferred approach and the group's success-level is more akin to hers than that of Subject 3.
3.1 Types of Talk
Three writers' findings about the nature of group talk were used to give insights into the Group task transcripts in the study.
Mercer (1995, pp104-7) described three categories of talk between peers in classrooms, (disputational, cumulative and exploratory) and three possible levels of analysis, (linguistic, psychological, cultural). He developed the description of each talk category as follows: (ibid, p104-5) 'Disputational talk …. is characterised by disagreement and individualised decision-making. There are few attempts to pool resources, or to offer constructive criticism of suggestions.' ' Information is flaunted rather than shared, differences in opinion are stressed rather than resolved.'
' [In] cumulative talk … speakers build positively but uncritically on what the other has said. Partners use talk to construct a "common knowledge" by accumulation. Cumulative discourse is characterised by repetitions, confirmations and elaborations.' '[In] exploratory talk … partners engage critically but constructively with each others' ideas …. Statements and suggestions are offered for joint consideration. … Challenges are justified and alternative hypotheses are offered.'