awareness of words as phonological entities: the rôle of

10
Applied Psycholinguistics (1987) 8, 223-232 Printed in the United States of America Awareness of words as phonological entities: The r ôle of literacy R ÉGINE KOLINSKY, LUZ CARY, and JOSÉ MORAIS Université libre de Bruxelles and Universidade de Lisboa ADDRESS FOR CORRESPONDENCE Régine Kolinsky, Laboratoire de Psychologie Expérimentale, Université libre de Bruxelles, Avenue Adolphe Buyl 117, 1050 Bruxelles, Belgium ABSTRACT lUiterate, unschooled adults were tested on their notions of word length. Experiment 1 showed that only about half of them performed very poorly on a task requiring the production of a long/short Word. They were clearly inferior to formerly illiterate, unschooled adults. The illiterate group also broke up neatly into two subgroups, one performing perfectly or very well, the other failing com pletely or almost completely, when required to match the written and the oral form of long/short words. Similarly, Experiment 2 showed that about half of the illiterates were unable to choose the longest of two names when presented with drawings of objects. The results suggest that leaming to read, though not strictly necessary, plays a décisive rôle in the development of the ability of many individuals to focus on phonological length. The présent study concems the ability to reflect on words as objects. Several researchers have shown that young children are not aware of words as pho nological forms and confound words with the objects or actions the words dénote or with the act of speaking itself (Piaget, 1929; Vygotsky, 1962; Papandropoulou & Sinclair, 1974). BerthoudPapandropoulou (1978) concluded that it is not until âge 6 or 7 that words are seen as labels that have an independent existence. According to several authors (Francis, 1973; BerthoudPapandropoulou, 1978; Januschek, Paprotté, & Rohde, 1979a, 1979b; see discussion in Valtin, 1984), characteristics of the written language are crucial to the development of awareness of words as phonological forms. Beginning readers frequently refer to thèse characteristics when tested on their knowledge of words as objects, even if sometimes they find compromises indicating that the formai properties of the word are still not completely detached from its réfèrent (a typewriter may be a long word "because it has a lot of letters," BerthoudPapandropoulou, 1978). Donaldson (1976, 1978) and BerthoudPapandropoulou (1978) suggested that written words, because of their permanent and objective qualities, lend them selves better to contemplation and study than spoken words. They are not lim ited, as spoken words are, to one speaker and to the here and now of the speaking act. Unfortunately, studies on word awareness in children often confound âge and © 1987 Cambridge University Press 01427164/87 $5.00 + .00

Upload: others

Post on 16-Jan-2022

1 views

Category:

Documents


0 download

TRANSCRIPT

Applied Psycholinguistics (1987) 8, 223-232 Printed in the United States of America

Awareness of words as phonological entities: The rôle of literacy

RÉGINE KOLINSKY, LUZ CARY, and JOSÉ MORAIS Université libre de Bruxelles and Universidade de Lisboa

ADDRESS FOR CORRESPONDENCE Régine Kolinsky, Laboratoire de Psychologie Expérimentale, Université libre de Bruxelles, Avenue Adolphe Buyl 117, 1050 Bruxelles, Belgium

ABSTRACT lUiterate, unschooled adults were tested on their notions of word length. Experiment 1 showed that only about half of them performed very poorly on a task requiring the production of a long/short Word. They were clearly inferior to formerly illiterate, unschooled adults. The illiterate group also broke up neatly into two subgroups, one performing perfectly or very well, the other failing com­pletely or almost completely, when required to match the written and the oral form of long/short words. Similarly, Experiment 2 showed that about half of the illiterates were unable to choose the longest of two names when presented with drawings of objects. The results suggest that leaming to read, though not strictly necessary, plays a décisive rôle in the development of the ability of many individuals to focus on phonological length.

The présent study concems the ability to reflect on words as objects. Several researchers have shown that young children are not aware of words as pho­nological forms and confound words with the objects or actions the words dénote or with the act of speaking itself (Piaget, 1929; Vygotsky, 1962; Papandropoulou & Sinclair, 1974). Berthoud­Papandropoulou (1978) concluded that it is not until âge 6 or 7 that words are seen as labels that have an independent existence.

According to several authors (Francis, 1973; Berthoud­Papandropoulou, 1978; Januschek, Paprotté, & Rohde, 1979a, 1979b; see discussion in Valtin, 1984), characteristics of the written language are crucial to the development of awareness of words as phonological forms. Beginning readers frequently refer to thèse characteristics when tested on their knowledge of words as objects, even if sometimes they find compromises indicating that the formai properties of the word are still not completely detached from its réfèrent (a typewriter may be a long word "because it has a lot of letters," Berthoud­Papandropoulou, 1978). Donaldson (1976, 1978) and Berthoud­Papandropoulou (1978) suggested that written words, because of their permanent and objective qualities, lend them­selves better to contemplation and study than spoken words. They are not lim­ited, as spoken words are, to one speaker and to the here and now of the speaking act.

Unfortunately, studies on word awareness in children often confound âge and

© 1987 Cambridge University Press 0142­7164/87 $5.00 + .00

Applied Psycholinguistics 8:3 Kolinsky et al.: Words as phonological entities

224

literacy. It would thus be useful to examine the compréhension of the formai properties of words in illiterate adults. We did this specifically for the property of phonological length, in Expenment 1 by asking the adults both to produce a long and a short word and to match the written and the oral form of a short and a long word (the "mow-motorcycle" task devised by Rozin, Bressman, & Taft, 1974). In Expenment 2, we showed them two pictures and asked them to choose the one with the longer name. In the matching task of Experiment 1 as well as in Experiment 2, the relationship between word length and size of the denoted or depicted object is varied. This procédure allows us to examine to what extent illiterates can focus on formai rather than semantic properties.

The production task of Experiment 1 was presented to both illiterate and formerly illiterate adults. The latter are people with the same social origin as illiterates, but who leamed to read and write in spécial classes as adults. The two matching tasks were presented only to illiterates. Ail the adults tested in Experi­ment 1 had been tested before on several tasks of speech analysis (the corre-sponding results are reported in Morais, Bertelson, Cary, & Alegria, 1986), thus allowing for the examination of the relationships between différent meta-phonological abilities.

EXPERIMENT 1

Method

Subjects. The experiment was run in an agricultural area of south Portugal and in a Lisbon shantytown. The adults, 21 illiterate people and 20 former illiterates, were ail of peasant origin. Most of them were contacted through local trade unions, and ail were paid for their participation.

The illiterate group consisted of 3 men and 18 women, aged 25 to 60. Ten were employed in agricultural coopératives; the others were servants or manual workers. None had received any reading instruction at any time.

The formerly illiterate group consisted of 5 men and 15 women, aged 17 to 60. They began leaming to read at âges ranging from 18 to 40 years by attending elementary instruction evening classes for adults. Some of them had been per-suaded to do it by their own literate children. Others had been motivated by the fact that leaming to read would enhance their chances to find a better job. Twelve had received some kind of certificate after completing the literacy course. Five were still attending class. Thirteen were servants, four were industrial workers, and three were agricultural workers.

Both groups were tested on the production task. Only illiterates were tested on the matching task. Unfortunately, two of them were not available for the latter task.

Procédure. In the production task, the participants were asked to give one example of a short word and one of a long word ("dizer uma palavra pe-quenalgrande"), and then to justify their response.

In the matching task, the participants were shown two written words, told "one of thèse words is X, the other is Y , " and then asked, "Which one is X ? "

Applied Psycholinguistics 8:3 Kolinsky et al.: Words as phonological entities

225

They had to justify their choice. In five trials the relationship between word length and size of the denoted object was incongruent: the longer word corre-sponded to the smaller object (example: aviào 'plane', gafanhoto 'grasshop-per'). In five other trials this relationship was congruent: the longer word corre-sponded to the bigger object {uva 'grape', hipopotamo 'hippopotamus'). The shortest words had between 3 and 5 letters, the longest between 8 and 10. Ail denoted common objects.

Results and discussion

Production fas/c. The percentages of participants giving responses of a particu-lar kind are presented in Table 1. We judged response category by taking into account the two responses made by each person.

The responses of the former illiterates show that rudimentary literacy is associ-ated with correct understanding of the task. The illiterates as a group performed worse than the former illiterates. About 80% of the illiterates were unable to produce responses consistent with the instructions. For about 30%, however, the failure was apparently due to a lack of knowledge of the term " w o r d " rather than to the inability to manipulate phonological length. For example, one person gave as a long word "Gostava de saber 1er'' T d like to be able to read' and as a short word "Gostava" i ' d like'. The fact that most of the illiterates able to respond in terms of phonological length produced clauses or sentences instead of single words seems to contradict Barton's (1985) claim that illiterate adults demonstrate sophisticated awareness of the word unit. This claim was based on the observation of very good performance by illiterate adults on tasks of seg-menting a sentence into words or judging whether an expression was a word. The

Table 1. Experiment 1 production task: Percentage of participants giving a response of a particular type in each group

Illiterates Former illiterates

Correct answer 19 75 Response based on the phonological length of

express ions (many words /a f ew words) 2 9 5 Response potentially correct but ambiguous

(correct phonological length différence be­tween words, but also congruent s ize dif férence) 0 20

Response based on semantic content (based on the s ize of the denoted objects or

events) 24 0

(based on the emotional importance of the denoted objects or events) 14 0

Response based on the characteristics o f the speaking act itself (différences in speed or intensity of prononciation) 14 G

Applied Psycholinguistics 8:3 Kolinsky et al.: Words as phonological entities

226

discrepancy between Barton's results and ours is probably due to the fact that he was actually studying low-level literates, not true illiterates.

Among the illiterates who, like preliterate children, gave évidence of lack of awareness of the property of phonological length, most confounded this property with characteristics of the denoted object or event (long word: nogueira 'walnut tree', because it is a big tree; short word: galinha 'chicken', because it is smaller). The others confounded phonological length with another, pho-nologically irrelevant characteristic of the speaking act (e.g., the person's name, first said loudly and at a normal rate and then whispered at a faster rate). To sum up, responses to the production task show that about half of the illiterate adults are able to focus on the phonological length of expressions, even if most of them do not understand the term "word" . A few focus on some other formai charac­teristic of speech, but référence to semantic properties is the main source of mistakes.

It is interesting to relate thèse results to the scores obtained by the same illiterate adults on other tasks (cf. Morais et al., 1986). Among the 10 who responded phonologically to the word-length task, 5 scored at least 80% correct in deleting the initial syllable from an utterance, and an equal number performed well in a rhyme détection task. Among the 11 who did not respond on a pho­nological basis, the 80% correct level was reached by 7 of them in the syllable deletion task and by 5 in the rhyme détection task. Thèse results suggest that, before the onset of literacy, compréhension of the notion of phonological length may have little in common with other metaphonological abilities such as analysis of speech into syllables or rhyme appréciation.

Matching task. On the matching task, 93.7% of responses were correct in the congruent condition and 47.4% correct in the incongruent condition. Pattems of responses in the incongruent condition indicate a high degree of both interin-dividual variability and intraindividual consistency: Ten respondents based their match on the physical size of the réfèrent, and eight did so on phonological length, on at least four out of five trials.

Among the successful respondents, four reported that they recognized the target on the basis of letter knowledge: For example, one word was aviâo because it contained the letter " v " . However, this strategy may be misleading. One of thèse people, asked to point to the word mesa, pointed to alfinete because - as he said - it had the letter " e " . The explanations of the remaining four successful respondents referred to the number of letters. One of them also indi-cated rapidity of pronunciation. Among the 10 subjects who systematically made the wrong choice in the incongruent trials, 7 paradoxically justified their re­sponses on the basis of number of letters (e.g., aviâo was indicated to contain more letters than gafanhoto). Only 2 made explicit référence to the physical size of the object (e.g., saying that carruagem 'carnage' is the longest word because "i t has more space").

Thèse results show that literacy is not strictly necessary to understand that length of spoken and written words are correlated. As a matter of fact, almost half of the illiterate adults tested here could match spoken and printed words for length, even when the correct choice required resisting semantic countersugges-tion. One of the reasons for success is that illiteracy in the context of a literate

Applied Psycholinguistics 8:3 Kolinsky et al.: Words as phonological entities

227

Society does not necessarily mean complète ignorance of the alphabetic code. On the other hand, living for many years in a literate environment may not be sufficient to be able to disregard réfèrent size and attend to word length. About half of the illiterate adults actually made their choices according to semantic rather than formai properties.

Comparison with preliterates' behavior in the same task is made difficult by *' the apparent inconsistency among studies on children. Testing four-year-olds,

Lundberg and Toméus (1978) found an average of 58% and 50% correct re-sponses (i.e., near chance level) for congruent and incongruent trials, respec-

I tively, while Bialystock (1986) found 80% and 29%, thus suggesting strong reliance on the semantic criterion at this early âge. The inconsistency between the two studies, as well as the observation of différent pattems of behavior among illiterate adults, support the notion that there is no strict developmental séquence and that environmentally determined expériences with written language are much more influential than cognitive maturation.

The corrélations between number of correct responses in the incongruent trials and scores for rhyme détection and deletion of the initial syllable from an utter-ance (cf. Morais et al., 1986) were 0.18 and 0.17, respectively (both nonsignifi-cantly différent from zéro). The ability to detect rhyme is not necessarily anté­cédent nor conséquent to the compréhension of length correspondences between phonology and orthography: Among the 12 people who scored at least 50% correct responses on rhyme détection, 5 were poor on the incongruent trials. On the other hand, among the 8 subjects who were good on thèse trials, 2 performed at chance level on rhyme détection. The same conclusions can be drawn concem-ing the links with the ability to analyze speech into syllables: Among the 12 people who scored at least 50% correct on syllable deletion, 5 were poor on the incongruent trials. On the other hand, among the 8 subjects who were good on thèse trials, 2 scored poorly on syllable deletion.

Finally, no apparent relationship was observed between the matching and the production tasks. Among the 19 illiterates tested on both tasks, 8 produced short and long words consistent with the criterion of phonological length. However, 4 of them responded to the incongruent trials of the matching task on a semantic basis. On the other hand, 4 scored 100% correct on the latter trials but responded in the production task according to the referent's physical size. Two of them justified their matching responses on the basis of the number of letters, the other two were able to recognize some letters of the written words. On the whole, only 16% of the subjects performed correctly on both tasks; 31% failed both, and 53% were good in one but poor in the other. This apparent inconsistency between the

» two tasks may be linked to the fact that among illiterates the individual pattems of expérience with written material (e.g., letter knowledge and ability to read some words as ideographs) are probably highly diverse.

(. . •

EXPERIMENT 2

In Expenment 1 participants were asked either to produce short and long words or to match oral and written words, without any further explanation. Experiment 2 offered the participants several opportunities, via training trials, to understand what is meant by the notion of word length. The task consisted in silently

Applied Psycholinguistics 8:3 Kolinsky et al.: Words as phonological entities

228

evoking names for a pair of object drawings and indicating the one with the longest name. This task is différent from the matching task of Experiment 1 in several important aspects. First, the response is iess Hkely to be influenced by any available knowledge of letters or of word orthography. Second, acoustic eues are completely absent. Third, while in the matching task of Experiment 1 the réfèrent of the word need not be identified and, if identified, its physical size is not immediately apparent, the naming component of the task used in Experi­ment 2 implies a more direct confrontation with the réfèrent and may, therefore, enhance the tendency to rely on properties of the réfèrent rather than on pho­nological properties. For ail thèse reasons, this task provides a more stringent assessment of the ability to focus on phonological length. Moreover, we used a neutral condition (i.e., the two words, one short and one long, denoted objects of about the same size) in addition to the congruent and incongruent conditions. This permitted us to estimate the amount of interférence and facilitation provided by the semantics of the words.

Method

Subjects. A fresh group of 10 unschooled illiterate adults (8 women and 2 men), aged 46 to 69 years, was tested in Lisbon shantytowns. Most of them worked as servants or manual workers. They were paid for their participation.

Stimuli and procédure. The test consisted of 42 pairs of standardized drawings (from Snodgrass & Vanderwart, 1980), each pair inciuding one drawing whose name was short and one whose name was long. Short words were mono- or bisyllabic and included two to five phonèmes. Long words included three to five syliables and five to nine phonèmes. Within each pair, the words differed by one to three syliables and by one to six phonèmes.

For each trial, the examiner presented the participant with a pair of drawings and asked him or her to point out the one whose name was longer. For half of the trials, the target was on the left, for the other half it was on the right. Names were not provided, and participants were asked not to speak before the response. After the response, we controlled for correct naming.

There were three types of trials, each corresponding to a particular relationship between word length and size of the denoted objects. For neutral trials, only the phonological length varied, denoting objects of approximately the same size (example: pato-galinha 'duck-chicken'). For congruent trials, the longer word denoted the bigger object (example: camelo-olho 'camel-eye ') . For m-congruent trials, the longer word denoted the smaller object (example: casa-televisâo 'house-television'). There were 12 expérimental trials corresponding to each pair type. The order of trials was fixed and controlled to prevent the occurrence of more than two consécutive correct responses on the same side, and more than two consécutive trials of the same type. The same drawing ne ver appeared two times consecutively.

The task was introduced by a set of 12 neutral training trials. Six of them involved proper names differing by word length and referring to two identical photographs of the same person but presented as twins. In thèse trials, the

Applied Psycholinguistics 8:3 2 2 9 Kolinsky et al.; Words as phonological entities

examiner attributed orally a name to each " tw in , " for example Ruth and Mar-garet. Six training trials on the standardized drawings were then given, using common names of objects or animais, as in the expérimental trials. However, unlike in the latter, the names were uttered by the examiner. Moreover, for both types of training pairs the examiner provided corrective feedback, whereas no feedback was given during the expérimental pairs.

Results and discussion

Only trials where the participant had given the correct name were taken into account. Table 2 présents the individual and mean percentages for each type of item. Analysis of variance showed an effect of item type (F(2,18) = 16.96, p < .0005). The comparison between neutral and incongruent trials showed an interférence effect ( F ( l - 9 ) = 19.92, p < .005); that between neutral and con-gruent trials showed facilitation (F(l ,9) = 10.07,p < .025). So, while illiterates were able to perform the task above chance in the neutral condition (t{9) = 5.6, p < .0005), they were strongly influenced by the irrelevant semantic infor­mation.

Moreover, individual différences were important. Four participants performed very poorly on neutral trials (around 50-70% correct), whereas three of them obtained very high scores (80-90%). The positive corrélation between the neu­tral and incongruent scores (r = .87, p < .001) shows moreover that the amount of interférence is stronger for those who performed poorly on neutral tiials. Corrélations between neutral and congruent, and between congruent and in­congruent trials were nonsignificant (—.20 and —.34, respectively).

In the neutral condition almost half of the participants responded at or only slightly above chance level, thus showing again that many illiterate adults lack

Table 2. Experiment 2: Mean percentage of correct responses for each participant and pair type

Pair type

Participant Neutral Congruent Incongruent

1 90 86 90 2 67 100 09 3 67 92 17 4 90 80 64 5 75 83 58 6 50 83 17 7 75 83 58 " 8 75 100 58 9 83 92 75

10 58 92 08

Mean 73.0 89.1 45.4

Applied Psycholinguistics 8:3 Kolinsky et al.: Words as phonological entities

230

the ability to focus on phonological length. However, 30% performed at a relatively high level, thus indicating once more that literacy is not a strict prereq-uisite to judging phonological length. On the other hand, only one successful person was not disturbed by the semantic interférence présent in the incongruent condition. The results thus converge with those of Experiment 1 to indicate the trouble illiterates have in ignoring semantic content when asked to attend to formai properties.

On average, illiterate adults performed at about the same level as preliterate kindergartners. Using materials and procédures similar to those of Experiment 2, we found that preliterate children (38 five-year-olds) produced correct responses 67% of the time on neutral trials (Content, Kolinsky, Morais, & Bertelson, in préparation). Among them, 24% obtained very high scores (around 80-90%), this proportion being only slightly inferior to the one found in adult illiterates (30%). The mean score obtained on incongruent trials was also about the same: 44% of correct responses.

CONCLUSIONS

The fact that in both experiments approximately half of the illiterates could respond on a phonological basis implies that the ability to appreciate pho­nological length does not dépend crucially on having leamed to read. Other metaphonological abilities also seem to develop before the onset of literacy. For instance, both the ability to appreciate rhyme and the ability to analyze speech into syllables are présent in many illiterate adults (cf. Morais et al., 1986). Leaming to read plays a much greater rôle in the development of the apparently more complex ability of segmentai analysis (cf. Morais, Cary, Alegria, & Ber­telson, 1979; Morais et al., 1986; Read, Zhang, Nie, & Ding, 1986).

On the other hand, the fact that a nonnegligible number of illiterates failed the présent tasks indicates that cognitive maturation and expérience of language are not enough to guarantee the ability to appreciate phonological length. The results of the production task of Experiment 1 also show that thèse factors are insuffi-cient to guarantee the exact compréhension of the metalinguistic term " w o r d . " Despite their low level of literacy, ail the former illiterates tested here produced responses consistent with the phonological length criterion, and most of them showed an accurate compréhension of the term " w o r d . " Thus, leaming to read does elicit awareness of words as phonological entities in those people who had not achieved such awareness before.

We have found no clear relationship between tasks that require subjects to focus on phonological length, to grasp the length correspondence between pho-nology and orthography, to segment speech into syllables, and to appreciate rhyme. The lack of corrélation between the two tasks of Experiment 1 may be due to the dues provided by the expérience with written material, which proba-bly differs largely within the illiterate population. The lack of corrélation be­tween each task of Experiment 1 and the other metaphonological tasks (rhyme détection and syllable segmentation) suggests that ail thèse abilities develop independently of each other in illiterate people. Before the onset of alphabetic literacy, progress in one metaphonological skill may not generalize to the others

Applied Psycholinguistics 8:3 Kolinsky et al.: Words as phonological entities

231

because the différent skills rely on spécifie opérations and awareness of the segmentai structure of speech is still missing. But once segmentai awareness is acquired, an efficient analytic strategy may replace ail those spécifie opérations. Consistent with this view, Morris (1983) found, with beginning first graders, high corrélations between the ability to map spoken to written words in a memo-rized text and either segment counting or spelling ability, as well as between thèse two abilities.

ACKNOWLEDGMENTS This work bas been partially subsidized by the "Ministère de la Politique sc ient i f ique" (Act ion de Recherche fondamentale concertée "Processus cognit ifs dans la lecture") . The first author holds Research Assistant grants from the "Fonds national de la Recherche Sc ient i f ique" ( F . N . R . S . ) . Thanks are due to the Editors and two anonymous référées for their help in improving the présent report. Address requests for reprints to Régine Kol insky, Laboratoire de Psycholog ie expérimentale, 117 Avenue Adolphe Buyl , 1050 Brussels , Be lg ium.

REFERENCES Barton, D. (1985). Awareness of language units in adults and children. In A. W. Ellis (Ed.),

Progress in the psychology of language. Vol. 1. New York: Eribaum. Berthoud-Papandropoulou, I. (1978). An expérimental study of children's ideas about language. In

W. J. M. Levelt, A. Sinclair, & R. J. Jarvella (Eds.), The child's conception of language. Berlin: Springer-Verlag.

Bialystok, E. (1986). Children's concept of words. Journal of Psycholinguistic Research, 15, 13-32.

Content, C , Kolinsky, R., Morais, J., & Bertelson, P. (in préparation). Correlates of speech segmentation skills in kindergaitners.

Donaldson, M. (1976). Development of conceptualization. In V. Hamilton & M. D. Vemon (Eds.), The development of cognitive processes. New York: Académie.

Donaldson, M. (1978). Children's minds. London: Fontana. Francis, H. (1973). Children's expérience of reading and notions of units in language. British

Journal of Educational Psychology, 43, 17-23. Januschek, F., Paprotté, W., & Rodhe, W. (1979a). The growth of metalinguistic knowledge in

children. In M. van de Velde & W. Vandeweghe (Eds.), Sprachstruktur, individuum und gesellschaft. Akten des 13. Linguistischen Kolloguiums, Gent, 1978, Vol. I (pp. 243-254). Tubingen: Niemeyer.

Januschek, F., Paprotté, W., & Rodhe, W. (1979b). Zur ontogenèse sprachlicher handlungen. Osnabrucker Beitrage zur Sprachtheorie, 10, 37-69.

Lundberg, I., & Toméus, M. (1978). Nonreaders' awareness of the basic relationship between spoken and written words. Journal of Expérimental Child Psychology, 25, 404-412.

Morais, J., Cary, L., Alegria, J., & Bertelson, P. (1979). Does awareness of speech as a séquence of phones arise spontaneously? Cognition, 7, 323-331.

Morais, J., Bertelson, P., Cary, L., & Alegria, J. (1986). Literacy training and speech segmentation. Cognition, 24, 45 -64 .

Morris, D. (1983). Concept of word and phonème awareness in the beginning reader. Research in the Teaching of English, 17, 359-373.

Papandropoulou, T., & Sinclair, H. (1974). What is a word? Expérimental study of children's ideas on grammar. Human Development, 17, 241-258.

Piaget, J. (1929). The child's conception of the world. New York: Harcourt Brace. Read, C , Zhang, Y., Nie, H., & Ding, B. (1986). The ability to manipulate speech sounds dépends

on knowing alphabetic spelling. Cognition, 24, 31-44 .

Applied Psycholinguistics 8:3 Kolinsky et al.: Words as phonological entities

232

Rozin, P., Bressman, B., & Taft, M. (1974). Do children understand the basic relationships between speech and writing? The mow-motorcycle test. Journal ofReading Behaviour. 6, 327-334.

Snodgrass, J. G., & Vanderwart, M. (1980). A standardized set of 260 pictures: Norms for name agreement, image agreement, familianty, and visual complexity. Journal of Expérimental Psychology: Human learning and memory, 6, 174-215.

Valtin, R. (1984). The development of metalinguistic abilities in children leaming to read and write. In J. Downing & R. Valtin (Eds.), Language awareness and learning to read. Berlin: ' Springer-Verlag.

Vygotsky, L. S. (1962). Thoughl and language. Cambridge, MA: MIT Press.

4

1