Toegepaste Taalwetenschap in Artikelen - Volume 43, Issue 1, 1992

Two divided attention experiments investigated whether graphemes and phonemes can mutually activate each other during bimodal sublexical processing. Dutch subjects reacted to target letters and/or speech sounds in single-channel and bimodal stimuli. In some bimodal conditions, the visual and auditory targets were congruent (e.g., visual A, auditory /a:/), in others they were not (e.g., visual U, auditory /a:/). Temporal aspects of cross-modal activation were examined by varying the stimulus onset asynchrony (SOA) of visual and auditory stimulus components. Processing differences among stimuli (e.g., the letters A and U) were accounted for by correcting the obtained bimodal reaction times by means of the predictions of an independent race-model. Comparing the results of the adapted congruent and incongruent conditions for each SOA, it can be concluded that (a) cross-modal activation takes place in this task situation; (b) it is bidirectional, i.e. it spreads from grapheme to phoneme and vice versa; and (c) it occurs very rapidly.

When readers understand a text they construct a mental representation of the text. This text representation contains not only the information that is explicitly expressed in the text, but also information that is inferred by the readers during reading. A text representation can be described in terms of relations between the elements in the representation, but also in terms of relations of the representation with (a model of) the world. For example, readers can determine whether a particular sentence is logically consistent with the text, as well as whether it is true or false, whether it is plausible or possible, or whether it is completely new for her or him.The distinction between relations within the structure of the text representation and relations that refer to the world has hardly ever been made in empirical research on inferences. In general inferences have been investigated from the point of view of their contribution to the coherence of the representation. The consequence is that the question whether inferences are made during reading has generally been answered in terms of the internal coherence of the representation: Inferences that contribute to the coherence of the representation are made during reading, other inferences are not.The central question in the present research is to what extent inference processes are determined by the relations of the inferences with a model of the world and, accordingly, to what extent inference processes are determined by the reader's knowledge of the world. Experiments are discussed on inferences that contribute to the coherence of the representation and, accordingly, are supposed to be made during reading. In this research the knowledge of the reader with respect to the topic of the text was varied by having expert economists and non-experts reading texts on economics. The results indicate that the prevalent conclusion with respect to inferences is an overgeneralisation. It was demonstrated that inference processes are controlled by the reader's knowledge.

In the present study a detailed analysis was made of the oral and written language skills of 60 native Dutch and 120 non-native third and fourth grade students at Dutch primary schools. First of all, the achievement on a variety of oral language and reading and writing tasks were examined as a function of ethnic group and grade level. In addition, a principal component analysis was conducted on the test scores in the two groups of children. Finally, the prediction of reading comprehension achievement in terms of oral language and decoding skills in the two groups was determined.The results showed that the non-native children leave behind their native Dutch peers on most language tasks administered. Especially on vocabulary measures and reading comprehension tests the second language students attain relatively poor achievements. With children's progression of grade the differences between first and second learners tend to diverge rather than converge. With respect to principal component analysis it was found that four factors underlie the language proficiency in either group: oral proficiency, reading comprehension, decoding efficiency and writing ability. To a large entent the children's achievement in reading comprehension in both groups could be explained in terms of their oral proficiency and decoding efficiency. However, for the second language learners the level of reading comprehension ability turned out to be relatively more dependent on their oral proficiency level in Dutch.

In the Netherlands, as well as in other Western European countries, many minority children are unsuccesful in their school careers. Their lack of achievement is, among other things, due to a language problem which is most probably related to literacy. In vocational education they are faced with the specific task of text comprehension in the content areas, which is a reading task and a learning task at the same time.The study on which will be reported, concerns the assessment of text comprehension in relation to vocabulary and knowledge about some topics of the texts used in the same text-comprehension test. Dutch (L1) and foreign (L2) students were submitted to these tests in the first year of their vocational schooling. It was found that the L2-readers tend towards a top-down way of reading, which implies a stronger reliance on conceptual guidance than in L1-reading.

In most testing practices not only the ability or knowledge the test aims to measure is assessed, but also the reading ability of the tested person. The level of his reading ability interferes with the ability to be measured. Several 'communication-problems' between teacher and student reinforce the effect of this interferance. For non-native students with little or no experience with the test- and teaching customs of their host-country the effect is stronger.In this article the author discusses the three main categories of this miscommunication:- the question lacks an explicit instruction, causing uncertainty for the student as to what the teacher expects him to do;- the way in which the question is stated makes that it does not ask what the teacher aims to ask, often caused by a lack of accuracy in the formulation;- the question elicits information about an ability other than the one the teacher wants to get information about, caused by a wrong operationalization of the construct.

The final exams for English in Dutch secondary schools contain a multiple choice component for text comprehension. The questions in this component are mainly about factual content but progressively also about communicative aspects such as opinions and style. It is usually assumed that the latter add to the level of difficulty of the test in general. But is this really the case? In a (limited) project caried out by the University of Groningen this assumption has been tested in six pre-final forms at three grammartype schools, by setting the same test in Dutch and in English within a six-week interval. At the same time the influence of vocabulary command was mapped out against the overall test results. The communicative questions did not turn out to be more difficult in English than in Dutch and there appeared to be a strong correlation between the overall (not: single item) vocabulary command and general test score. The average test results were considerably higher (4.0) for the Dutch test than for the equivalent English one.

This article is a plea for more research into the interaction between linguistic expectancy and acoustic information. Even if under normal circumstances priority should be given to the acoustic signal, it is evident that linguistic expectancy is a prerequisite for understanding speech. Second language students in the Netherlands generally have a good command of the phonotactic properties of English. In developing the students' listening proficiency teachers therefore experience the law of diminishing returns if they keep focusing on phonetic and phonological properties of the foreign language. However, the students' linguistic familiarity with English may still give rise to interpretative problems. Notably collocations, false friends (faux amis), homonymy and polysemy are categories of words prone to misunderstanding. In research on listening proficiency more attention should be paid to how linguistic expectancy interacts with acoustic information in the perception of these categories of words in particular.It is demonstrated that slight uncertainty with respect to acoustic information may have disproportionate repercussions on understanding speech. A slight decrease in redundancy may make the difference between understanding everything or nothing, as the hearing-impaired are well aware. In this respect minor hearing losses - not uncommon nowadays at early ages due to overexposure to noise and loud music - may have serious effects on information processing. If reduction of redundancy results from hearing loss, there is little teachers can do to help their students improve their listening proficiency. Teachers should be aware that the inherent differences in sound perception of their students may go unnoticed in the mothertongue, but will inevitably surface in the second language, where linguistic expectancy is by definition less well developed. It would therefore seem unwise to relate listening proficiency in the native language as a test-measure to listening in the foreign language, as has been suggested by some researchers in the past.

The use of the Kalman filter in the Dutch pupil monitoring system LISKAL makes it possible to reduce with 30 to 50% the estimation error variance resulting from Bartlett's well-known cross-sectional factor score estimator. The PC-program LISKAL is based on a longitudinal LISREL model of reading achievements of pupils in primary education. It combines past and present information from educational test scores to estimate optimally individual pupil's achievements in reading skills: decoding speed, reading comprehension and spelling.It is explained how LISKAL works and how it can be used by the teacher in the class room. Most LISKAL results are given in graphical form. Different kinds of information are provided by LISKAL: 1) the individual pupil's achievements are compared at successive time points with those of a normative group, 2) the pupil is compared with itself at previous points in time, 3) on the basis of the model predictions can be made of future achievements to enable the teacher to take timely intervention measures, 4) by comparing predicted development with actual development the effect of intervention measures can be evaluated, 5) in addition to individual developmental curves the program also generates mean curves for groups of pupils as, for example, classes or schools.In the present version of LISKAL achievement variables are present as standardized variables in the form of z-scores, having mean 0 and standard deviation 1 at each point in time. This implies relative assessment in relation to a normative group. However, teachers are often interested in absolute developmental curves, especially for low achieving pupils. The problem of estimating absolute developmental curves can also be solved by means of the Kalman filter. The LISREL model on which it has to be based is more complicated, however. Absolute assessment will be added as an option in a new version of the LISKAL program based on a new research project with a nationwide sample of 3430 pupils.