Toegepaste Taalwetenschap in Artikelen - Volume 13, Issue 1, 1982

Before giving a survey of reading comprehension theories and their implications for the testing of reading comprehension itself, some text theories particularly with regard to readability and structure are dealt with.The authors are of the opinion that questions should preferably refer to elements which rank highly in the structure of the text and to places in the text where there is a discrepancy between form and structure.After a survey of subskill theories, psycho-linguistic and cognitive theories and functional and communicative ideas on reading comprehension, the following conclusions are drawn with regard to the possibility of constructing tests which are in agreement with the principles of these theories: There is no reason why final reading comprehension tests should aim at specific subskills (provided their existence can be demonstrated), considering the complexity of problems which occur in the normal use of texts. For the time being a central testing of reading operations and strategies seems to be impossible. Choosing texts for which prior knowledge will not play a role is not quite justified. It would be worthwile to construct tests which can show the candidate's ability to use his prior knowledge. This may possibly lead to formulating questions which sometimes indicate an individual (non-uniform) processing of the text. A great many problems will have to be solved before a reliable testing of the use of prior knowledge can be achieved. Demands for a communicative approach, however, are reflected in current practice and will be met within the foreseeable future. The main problem in this area is that the test should be able to demonstrate to what degree the goals of readers with regard to a certain text have been reached.

A report is given of the construction and validation of diagnostic tests for the evaluation of pupils in the first stages of reading instruction. With these tests it will be possible to apply psychological principles of learning to read to the design of interventions that will facilitate the acquisition of desirable levels of reading performance. The tests indicate, which pupils have difficulties in learning to read. They also give an indication of the nature of these problems and of the way these problems can be solved.For each test it was analyzed with the help of information processing task analysis which subskills play a role in reading and writing words and short texts. Also an analysis was made of the errors in the various tests. Then for each test it was investigated what the connection was between the occurrence of certain types of errors in the test and the mastery (or absence of mastery) of subskills that underly the test performance. It turned out that in a number of cases the occurrence of certain types of errors could be related to an insufficient command of the specific subskills which play an important role in the initial stages of learning to read and to write.Thus the tests can be seen as diagnostic evaluation instruments. By analysing the pupil's errors in the tests the teacher will be informed about the degree to which his pupils have mastered the underlying sub-skills. In this way the teacher can himself improve the quality of instruction especially for those pupils who are in danger of falling behind in the teaching process.The construction and validation process is illustrated by means of two of the tests.

Various theorists have proposed different reasons for the usefulness of concrete analogies in written texts. Some stress a concretizing function, others a structurizing function, and still others suggest that analogies induce a more active proces-sing of the text. The main question which we tried to answer in a series of six experiments were:a. Does the addition of concrete analogies lead to higher per-formances and longer reading-times in subjects of different ages?b. What are the effects under restricted time conditions?c. Why are concrete analogies effective?d. Are there aptitude-treatment interactions? In general, the results showed that subjects, when confronted with analogies, not only studied longer, but learned more and different things. However, if the study time was limited, these effects disappeared. Furthermore some aptitude-treatment inter-actions were found, especially in the case of the visualizer-verbalizer dimension. All three of the above mentioned functions of analogies were supported by the data.

Linking and integrating different parts of information in texts have frequently been the object of the psychology of reading. A lot of attention has been spent on how people infer and how they supply incomplete information. Experimental research, however, mostly has been restricted to causal and final connections in short and simple informative or narrative text samples. But in longer and complexer texts of different sorts the possibilities of connections as well as the functions of concealed information may differ, demanding different kinds of reading strategies and processes.In literature the concealing of information and even discon-nectedness have become sanctioned literary techniques. The theory of literary reception as founded by Ingarden and Iser have made the "undetermined spots" ("Unbestimmtheitsstellen") one of its fundamentals and have drawn consequences from it for the reading of literary texts.In an experiment on reading expectations with different age groups between 14 and 22, it could be shown that the perception of informational connections over longer distances and that supplying missing links increased with age. The younger subjects were reading more "linearly", mostly only using the information just read. They were less able to use or modify more remote information or to anticipate later possibilities than the older subjects. This was demonstrated with a detail from one of the short stories used in the experiment.For making such findings available for teaching aims, two types of further experimental research are necessary. It must be traced which effects different variables such as the kind of the relationship between parts of the text, the form of presentation of informational parts, temporal distance, the influence of context, have on the processing of connections. Secondly methods should be designed and tested for effectuating and speeding up the development of the processing activities.

This paper presents an approach of training foreign language reading based on the theory that reading can be described as a cybernetic process of 'bottom up' and 'top down' processing. In this framework we distinguish five fields of knowledge which the reader can use for 'top down' processing:1. Knowledge about the probability of letter combinations2. Knowledge about sentence structures3. Knowledge about the probability of word combinations4. Knowledge about logical structures5. Knowledge about the worldKnowledge in the fields 1, 2 and 3 is language-bound. Therefore, the ratio between the visual information needed and the knowledge already possessed will be more adverse than in the mother tongue, for most foreign language readers. This lack of knowledge can be compensated for by optimalizing knowledge in the fields 4 en 5, which are only slightly language-bound.To enlarge the readers'knowledge in the fields 1, 2 and 3, we have developed a reading programme, providing the student with a rich learning environment in the form of much reading material of an adequate level. We used the CLOZE-procedure to grade the reading material and to determine the appropriate level. To extend the students'ability to use knowledge from the fields 4 and 5, we developed a training programme partly based on Gal'perins theory of the stagewise development of mental activities. One of the most important stages to be completed is the so-called 'verbal stage', in which the prediction strategies used by the reader had to be expressed in words.After 8 months (one hour a week) a qualitative and a quantitative evaluation was completed. In respect of the latter, we found the most significant effect (p < .01) was an increased ability to predict meaning on the basis of a relatively incomplete foreign language knowledge.

In view of the difference of opinion as to what the translation may and should test, it seemed useful to find out what in fact pupils are doing when they are making a translation. For this purpose I interviewed ten pupils of 3 more or less random schools in the week after their final exam. I presented them with a Latin text, which was about half the length of the text in the exam, though comparable as far as degree of difficulty was concerned. They were asked to translate it in the usual manner, the only difference being that they need not write down the result. Everything they were thinking had to be said aloud and was recorded.From these data I concluded that the translation only calls for minimal comprehension and that the degree to which the knowlegde of the language is demonstrably used is sometimes less high than is often supposed. Knowledge of vocabulary plays a very important role because in principle the pupil builds up his translation on the basis of the words that he knows or has looked up. Where the order of the words in the foreign language and the mother language correspond, a correct translation may occur without an apparent effort to understand the structure of the foreign language.The procedure of concurrent verbalisation does not clarify the nature of the processes involved. To find this out a combination of thinking aloud and 'probes' in a less naturalistic setting would seem necessary.

In order to understand how children learn to read, it is useful to consider reading as a language process. Each word has its own characte-ristics. Words are fundamental constituents of the linguistic system. They are conceptualized as abstract units having several different identities: Orthographic, phonological, morphological, syntactic and semantic. One of the most important tasks facing the beginning reader is to amalgamate these different identities.During the last decades various models of reading have been developed: top-down, bottom-up and interactive models. Interactive models provide the best account of the existing data on the use of orthographic structure and sentence context.In the Netherlands the teaching of initial reading has been influenced by different theories of reading. The new edition (1980) of the reading curriculum which is most widely disseminated (in the Netherlands), "de structuurmethode", largely corresponds to an interactive model of the reading process.

"Reading" can be described as a process that derives concepts from visual codes. These codes can be pictural or non-pictural. Non-pictural codes can have graphic-verbal-dialogic backgrounds. The initial reading process by prelingually profoundly deaf toddlers can be developed as "ideo-visual reading", i.e. without any phonologic en- and de-coding. This can be worked out in such a way, that communicative interactions can be represented and read in "visualized conversations".

This research focuses on the comparative speed of reading in a first and in a foreign language. Dutch-speaking subjects with varying degrees of command in French were set to read a number of informative texts in Dutch and French. They were instructed to read through these texts as quickly as they could, and to summar-ize them directly afterwards in a few lines each. When dealing with Dutch texts, subjects were found to read an average of 165 words per minute. When reading the French texts, the speed of reading dropped by 20% for subjects with a completed University education in French. For third-year students of French a slow-down of 25% was found. A third group of subjects - students of various disciplines who had studied French up to Α-level only -needed as much as 150% more time for the French texts. The texts used fell into one of three categories, viz. reports (newspaper items covering everyday events), opinionating texts (newspaper items expressing point of view), and academic texts (passages from introductory academic textbooks on various subjects). When the language was Dutch, reports allowed an average reading speed of 182 words per minute, opinionating texts of 171 words per minute, and academic texts of 141 words per minute. When the results obtained on the French texts were differentiated according to text type a similar pattern emerged.As a further outcome of the experiment the assumption was con-firmed that the correlation between word recognition speed - as measured by a lexical decision task - and reading speed will be lower when the medium in both tasks is a foreign rather than subjects' own language. This is especially the case with subjects whose command of the foreign language is relatively low. These data are in keeping with recent psycholinguistic views on the reading process, and with other experimental findings as to the differences between good readers and poor readers reported from investigations into the initial stages of mother tongue reading.

According to the speech theory of e.g Bühler and Wittgenstein the understanding of speech is more than transforming linguistic forms into semantic structures. The core of the process of understanding has to do with interpreting human activity. Modern research in the field of reading comprehension has demonstrated that knowledge of scripts, frames etc. plays a crucial role in the reading process. This also indicates that speech reflects the world of human ac-tivity. This means that, even when a reader has a good command of the language but has insufficient knowledge of and insight into this background, his reading comprehension will be poor.We take the position that the basic sense of speech is social meaning, which is enclosed in the relation between an utterance and the situation of interpersonal communication and co-operation. E.g. someone wants to sell a house, to inform about a country etc. Within such domains of co-operation language offers the means to indicate and to describe things in a world. Indication and des-cription in speech are always based on social meaning, i.e. they go back to common human activity.When we confine ourselves to informative texts, we think that description of the world we live in is based on common methods of exploring reality. Some of these methods are feature analyses, comparison, classification, process analyses and explanation. In a learning experiment during the third till the end of the sixth class in two elementary schools we taught the pupils to use these methods systematically by engaging them in the formal exploration and description of things, events etc.In this paper the outline of the experimental program is sketched. Some results are reported. It seems acceptable to conclude that reading comprehension is fostered by the experimental program. However, more data are needed for more definite and precise conclusions.