• Nie Znaleziono Wyników

PART II: THINK-ALOUD READING COMPREHENSION STUDIES

2. A review of selected studies in the L1 reading context

2.7. Conclusions – reading in L1

Th e review shows that protocol analysis proved successful in eliciting many as-pects of reading ability, not just readers’ attempts to cope with comprehension breakdowns. Below, I look at the relationship between the TA methodology and the results the studies produce. In the next section, I discuss the view of reading that the studies seem to create.

2.7.1. The relationship between the think-aloud method and reading

Th e survey of think-aloud studies presented above shows the “symbiotic” rela-tionship between the think-aloud method and the construct this methodology explores – reading. Th e studies applied various procedures. Diff erences referred to verbal reporting tasks, the degree of intactness of texts and the mode of text presentation. Th ey are presented below.

Table II.4. Th e relationship between the think-aloud procedure and reading as investigated in the L1 reading context

Th ink-aloud procedure Examples of studies

Verbal reporting tasks reading aloud a text (Bereiter and Bird 1985) vs.

silent reading (Lundeberg 1987)

reading texts in an uninterrupted way and reporting on one’s comprehension concurrently (Bereiter and Bird 1985)

vs.

stopping at the end of every sentence and reporting retrospectively (Olshavsky 1976/1977)

Degree of intactness of texts reading complete unmodifi ed texts (Lundeberg 1987)

vs.

reading texts with “problem” areas, identifying and explaining those areas (Baker 1986)

Mode of text presentation reading texts

vs.

listening to texts (Collins, Brown and Larkin 1980)

Source: own study.

In all the studies, the think-aloud method seems appropriate for research purposes.

Let us look at how the studies capitalised on the potential of the think-aloud method-ology. Collins, Brown and Larkin (1980) exposed subjects to the oral version of texts, eliciting uninterrupted retrospective understanding of complete texts. Th is procedure

created an excellent opportunity to observe how readers utilise information from texts and, driven by their prior knowledge, revise and evaluate their understandings of the text in the form of a model. Th is methodology “suits” the theory which serves as a framework for the study (the authors’ progressive-refi nement theory of text, which is a model-based view of reading). An oral presentation of texts to subjects induced the construction of a situation model representation of text, rather than a propositional representation (van Dijk and Kintsch 1983). It created an opportunity to observe readers who “do not simply connect the events in the text into a sequential structure.

[But rather] they seem to create a complex scenario or model within which the events described might plausibly occur” (Bransford and Johnson 1973, cited in Collins, Brown and Larkin 1980: 385). It is worth noting that theories concerning inferencing point to the diff erence between a propositional representation and a mental model representa-tion of text. Research (e.g., Jenkin et al. 1993) suggests that texts can be comprehended in two diff erent manners: reading in L1 may result in creating a mental model repre-sentation of the text, while reading in L2 only in the construction of a propositional representation. I think that the technique applied in this study could have brought diff erent results, had the aim of the study been reading in L2.

Baker (1986, cited in Baker 1996) investigated how readers monitor their compre-hension. Preparing a passage with diffi cult-to-understand parts and asking subjects to identify these areas was a suitable choice to observe this aspect of metacognition.

Let us now compare the results of the study that applied a rather artifi cial tech-nique of stopping reading at the end of clauses (Olshavsky 1976/1977) with the fi nd-ings of the research in which readers were asked to read texts as they would normally have read (Lundeberg 1987). Also in these studies there is a noticeable link between the research method applied and what this method elicits. In the protocols which recorded clause-after-clause reading, Olshavsky identifi ed three types of strategies:

those related to word, clause and story. Slowing down the process of reading and dividing readers’ text processing into clause-protocols shed light on these three levels of text comprehension. However, it is important to emphasise that although 9 strategies were identifi ed within the word and clause types, only one of a very general nature within the story level was noted. Th e clause-by-clause technique did not manage to elicit behaviours refl ecting text processing at a higher level, i.e., the discourse level. In contrast, Lundeberg’s (1987) procedure is less structured, but more natural. Students were asked to read complete, unmodifi ed texts; during the session the subjects were asked questions which were adjusted to their individual behaviours. Th is unstructured procedure yielded protocols which refl ect strategies that readers used at all levels of text processing. For example, the researcher identi-fi ed a word-level strategy of contextually deidenti-fi ning words and a clause-level strategy of adding information. Th e bulk of strategies identifi ed by Lundeberg is related to coping with a more global context, e.g., overviewing the text, synthesising to con-nect facts into a cohesive whole. As the only researcher in this overview, Lundeberg identifi ed a group of strategies referring to readers’ abilities to critically evaluate the facts and opinions presented in the text. It seems that this whole-text-based think-aloud procedure elicited a wide range of behaviours and enabled the researcher to

identify a rich list of diff erences between novice and expert readers. An interesting fi nding is that expert readers “fl ip to the end of the text to preview the text” – a strat-egy impossible to elicit in a sentence-by-sentence procedure.

Th e discussion presented above has demonstrated how each study investigated reading in L1 and what methodology it used. It appears that various ways of organis-ing think-aloud tasks have created various research conditions, which have allowed the researchers to investigate various aspects of reading. Th is has illustrated the potential of protocol analysis as a method providing a rich source of evidence and off ering ample research possibilities.

2.7.2. The view of reading and the reader created by the studies

Th e next issue I discuss is the view of reading all the protocol analyses seem to develop. All the studies analysed above assume that reading resembles a problem-solving activity. Readers aim to understand the text but in this task they encounter diffi culties – problems. Readers attempt to solve problems related to both a local context (the word- and clause-levels of text processing) and a more global context (the whole text-level). Th ey experience diffi culty in understanding the meaning of unknown words or the grammaticality of a sentence as well as comprehending facts presented in a paragraph or the whole text. To solve these problems, readers apply a variety of strategies. Th ey use context to defi ne unknown words, search for synonyms, reread, rephrase to make the text simpler and infer. Th ey know what to do when they cannot comprehend the ideas conveyed in the text. Th ey ask ques-tions to link text ideas with one another and draw on their background knowledge to check whether these ideas are consistent with what they know about the topic.

Sometimes readers do not succeed after the fi rst trial. Asking the wrong questions about the text may lead to constructing a series of intermediate incomplete ver-sions of understanding. Th e process of reaching the fi nal stage of comprehension is similar to constructing a model of text. It is a data-driven process, which means that readers draw on the information given in the text and their prior knowledge.

Based on problem-solving theories, the studies off er insights into how skilled readers diff er from less skilled ones. Good readers seem to use more strategies than poor readers. Skilled readers have suffi cient language profi ciency; they plan their reading, e.g., they prefer to spend more time at the beginning of their reading to overview the text. Th ey are fl exible and adjust their reading rate, recognising parts of text which contain irrelevant details. Th ey use effi cient monitoring skills, i.e., they check the logic and consistency of the model they develop and evaluate the eff ective-ness of their actions. Th ey read critically, i.e., they evaluate facts and opinions from the text and are able to suggest alternative versions of the text message. Th ey know how to draw on their prior knowledge related to the text form; they recognise a type of text and predict from text structure. Th ey are careful while using their knowledge on the topic of the text; they integrate it with the knowledge presented in the text.

As regards reading instruction, the studies imply an encouraging conclusion that strategies are teachable. It is possible to improve readers’ evaluating skills; readers can

be taught how to identify comprehension problems and to engage in active problem solving. Learners can also develop the strategies of rephrasing the text to make it simpler. However, the skill that proved the most diffi cult to teach is being aware of gaps in one’s understanding – an important aspect of metacognition and looking for links between topics – an ability that is an example of higher-level text processing.

Th e studies also point to the relationship between strategies and other learner factors, e.g., reader interest and the style of text. Texts evaluated as interesting elicit more strategies, which seems to imply that readers are cognitively more active while coping with such texts. Abstract texts require more inferencing – the observation indicating that abstract material is a more demanding problem to solve. Th ese fi ndings can prove useful in designing reading instruction.

Th e overview of the reading studies based on problem-solving theories described in this chapter points to a number of merits of this type of research. Th e conceptu-alisation of reading as problem solving has allowed the researchers to:

identify reading strategies used in various types of texts, e.g., expository, –

descriptive and written in an abstract style;

investigate strategies in relation to reading profi ciency and reader interest;

examine diff erences between skilled and less skilled readers;

explore the possibility of teaching less skilled readers strategies used by more –

skilled readers and fi nding out which of the identifi ed strategies may prove most diffi cult to teach.

I think that the advantages enumerated above seem encouraging and can inspire more research into L1 reading.

Notes

Means-ends analysis is a technique that was fi rst used in

1. Newell and Simon’s

(1963) model of problem solving. Solving a problem involves fi rst defi ning the diff erence between the current state and the goal state, and gradually reducing diff erences between the two states. It is crucial to consider at every step in this process the means that one has at his/her disposal to achieve the goals. Th is helps one to defi ne and achieve intermediate states (subgoals) that bring him/her closer to the goal state, i.e., the solution of his/her problem.

It is not clear what Baker (1985) means by propositions. Most probably they de-2.

fi ne propositions as “ideas that can be expressed in words, not the words them-selves” (McNamara, Miller and Bransford 1991). See Part 1, Chapter 3, Section 3.1.1. for more information about proposition-based models of reading.

Although Lundeberg called the strategies described in the study as meta-3.

cognitive, in my review I have deliberately avoided calling these strategies

“metacognitive.” In my opinion, the strategies the researchers identifi ed are mainly cognitive, i.e. they are an integral part of task performance (as defi ned by O’Malley and Chamot 1990).

3. A review of selected studies