#AUTHOR: Mousty, Phillipe and Bertelson, Paul

#TITLE: A Study of Braille Reading:

#ORGANIZATION: Université libre de Bruxelles, Brussels, Belgium

#CATEGORY: Braille, Reading

#PUBLICATION: The Quarterly Journal of Experimental Psychology (1985) 217-233

#ABSTRACT: Considers how reading speed of braille readers is affected by hand usage and degree of contextual constraint. No overall superiority of one hand was observed in one-handed reading, but there were large and reliable individual differences in pattern of hand superiority, which were not related to general performance level.

A Study of Braille Reading:

1. Reading Speed as a Function of Hand Usage and Context

Philippe Mousty and Paul Bertelson

Université libre de Bruxelles, Brussels, Belgium

Evidence regarding the respective functions of the hands in braille reading was sought by considering how reading speed is affected by hand usage and degree of contexual constraint. Twenty-four blind readers read aloud prose, statistical approximations and scrambled words with either hand alone or with the two hands. No overall superiority of one hand was observed in one-handed reading, but there were large and reliable individual differences in pattern of hand superiority, which were not related to general performance level. All subjects read faster with the two hands than with the faster hand alone. The relative gain from two-handed reading was negatively correlated with absolute size of speed difference between the hands in one-handed reading, which suggests that both hands participate in the collection of text information. On the other hand, in each reading condition, reading speed increased with degree of contextual constraint, from scrambled words to prose. The fact that the effect is comparable to the one obtained for visual reading in other studies is inconsistent with the "compensatory processing hypothesis", according to which readers would depend more on context "when the access to textual information is slower. Also inconsistent with the notion are the facts that context effects are not systematically stronger in slow than in fast readers nor for slower than faster hand combinations.

Introduction

The performance of the braille reader is of considerable potential interest for the understanding of cognitive processes involved in all forms of reading, whether visual or tactile. Since the information obtained by the reader is basically the same but is acquired by a perceptual system with widely different characteristics, the comparison of tactile and visual reading may provide a sort of converging operation approach. The hands, for example, explore the text in a serial, letter-by- letter fashion, in clear opposition to the discrete character of gaze fixations, each of which makes a window of text simultaneously avail able. Another important difference concerns the spatial arrangement of the sensory field. In visual reading, the division of function between central and peripheral vision is largely determined by anatomical factors responsible for the differences in acuity between the two regions of the visual field and for their constant spatial relationship, while the hands have, in first approximation, the same tactile capacities, and could in principle adopt any exploration pattern that would prove convenient for understanding the text. The opportunity has, however, not produced the response one could have expected, and there have been few systematic studies of the mechanisms of braille reading, as the recent review by Foulke (1982) clearly shows.

The present study was planned as an exploration into the patterns of hand activity adopted spontaneously by blind readers, with an emphasis on the division of functions between the hands in two-handed reading. Only subjects who reported reading habitually with two hands, or who were known for doing so, were tested. It was decided to use the old- fashioned task of reading aloud rather than one of the more analytical ones preferred in much contemporary work on reading (Marslen-Wilson & Tyler, 1980). We wanted in fact to stay as close as possible to the normal reading situation. Preliminary studies had suggested that two- handed patterns were different when reading technical text and popular novels. To follow up the suggestion, two types of presumably more difficult materials, statistical approximations and disconnected words, were used in addition to prose. The hands of the subjects were recorded on video while they read the different materials aloud, with either hand alone and with the two hands. The results are described in two papers, the first dealing with overall performance and the second with patterns of hand activity.

The main objective in the present paper was to examine whether information on the division of functions between the hands can be obtained simply by analysing the relation of speed of reading to the hand combination used for exploring the text. Two main questions are thus asked: (1) is one particular hand superior in one-handed reading and is the eventual superiority dependent on material? and (2) how does speed of two-handed reading relate to speed of one-handed reading, again with material as a parameter? An accessory objective was to know how speed of reading is affected by contextual constraints.

1. The question of hand superiority in one-handed reading has stimulated more work than any other related to hand movements. Early studies were concerned mainly with the practical question of recommending one particular hand for one-hand reading instruction. In an often quoted study, Smith (1929) asked "which hand is the eye of the blind?" and proposed that it is the motor non-dominant hand. The only direct support for the notion, however, came from the performance of only one right-handed subject. In Germany, Bürklen (1917) and Grasemann (1917) had subjects read texts with either hand alone and classified them on the basis of speed differences into "left-hand (LH)-dominant", i.e reading at least 20% faster with the LH, "right-hand (RH)- dominant" and without hand dominance. They obtained a majority of LH-dominant subjects, which, according to Bürklen, contradicted the opinion prevailing at the time among German braille teachers. In the United States, Holland and Fehr (1942) and Fertsch (1947), using the same procedure, obtained more RH-dominant than LH-dominant subjects. In Fertsch's study, moreover, the relative majority of RH- dominant readers was greater among fast readers than among slower ones. She suggested that the discrepancies between results obtained by different investigators regarding hand superiority might reflect spurious differences of skill level in the subject populations.

In recent studies, the focus has shifted to the relation of eventual hand differences to hemispheric specialization of function. Hermelin and O'Connor (1971a, b) reported that blind children read text faster with the index or the middle finger of the LH than of the RH, and that blind adults read disconnected letters more accurately, although not faster, with the middle finger of the LH than of the RH. They attributed these LH advantages to right-hemisphere superiority in dealing with spatial patterns. LH superiority has, as a matter of fact, been found in several studies where sighted subjects were given tasks involving processing of braille characters which they could not read (Rudel, Denckla and Spalten, 1974; Myers, 1976; Harris, 1980). For blind subjects extracting linguistic information from braille, the evidence is much less clear. Harris (1980) in a study of blind students reading prose with the middle finger of either hand, found a strong superiority of the LH in those who preferred reading with that hand, and no hand superiority in those who preferred reading with the RH or with both hands. Mommers (1981) reports a non-significant LH superiority in children reading aloud disconnected words with the index and middle fingers of either hand. And Bradshaw, Nettleton and Spehr (1982) have recently reported an absence of hand superiority in subjects searching word lists for semantic targets or letter lists for phonological targets. It would thus seem that, for the time being, the question of hand superiority in braille reading admits of no simple answer.

2. The question whether using two hands is advantageous is, of course, fundamental to understanding hand cooperation. The majority of braille readers read with the two hands; 88% of the 50 readers interrogated by Bürklen (1917) and 71% of the 200 surveyed by Lowenfeld, Abel and Hatlen (1969) were described as using the two hands. When, on the other hand, the performance of two-handed readers has been measured for respectively one-handed and two-handed reading, the latter has generally been found superior for the reading of prose (Bürklen, 1917; Grasemann, 1917; Kusajima, 1974)1 as well as for disconnected words and digits (Mommers, 1980).

1Kusajima (l974) writes that the difference is small and non-significant, but a t-test applied to the data in his Table 1 (p.8) comes strongly in favour of two-handed reading [t(ll)=2.75;p=0.01]

Information relevant to division of function between the hands can be obtained also by examining how the speed of two-handed reading relates to pattern of hand superiority in one-handed reading. Bürklen (1917) reported that, of his three subgroups of subjects, those without hand dominance (less than 20% inter-hands difference) were the ones who derived the greater relative gain from two handed-reading. Similarly, Fertsch (1947) found that "subjects with equal hands" were the fastest two-handed readers.

Our data will incidentally provide evidence about the effects of contextual constraints on reading speed. For visual reading, the benefic ial effect of context on speed of oral reading has been well demonstrated in the classical study by Morton (1964), where subjects read aloud prose and statistical approximations from 1st to 8th order. Current discussions regarding the respective roles of textual and contextual information in word recognition make the case of braille especially interesting. It has been suggested that, in ongoing word identification, resort to contextual information is increased in those conditions where textual information pick-up is deficient (Stanovich, 1980). Support for this "compensatory processing hypothesis" is provided, for example, by results showing that word naming times are more affected by sentence context in poor than in good readers (Perfetti, Goldman and Hogaboam, 1979). Since braille reading is much slower than visual reading, presumably because encoding of the characters is a slower process, the compensatory hypothesis predicts that context should be more effective in braille. Predictions that our data make it possible to examine are available also at the level of interactions of context with hand usage (context more effective for the slower than for the faster hand, or for one-handed than for two-handed reading) and with reader's skill level (context more effective in slow readers).

Method

Recording Situation

The subject sat in front of an horizontal tablet, which supported the pages to be read. Two television cameras (Sony AVC-3250CE) were arranged so that one, situated above the tablet and pointing downward, recorded the horizontal movements of the hands on the page, while the other, situated at the level of the tablet and pointing horizontally toward the subject, recorded the vertical position of his fingers. A digital timer giving the time to the nearest 1/10 sec was fixed to the tablet with its display window in the field of the first camera. The output of the two cameras was fed, through a field divider (Sony CMW-1l0CE) to a video recorder (Sony AV-3670CE), together with the voice of the subject.

Reading Material

Three types of material were used: (1) prose: passages from French detective novels by G. Simenon; (2) 7th-order approximations to French obtained through the sequential completion procedure described by Miller and Selfridge (1950); (3) scrambled words: sequences obtained by scrambling the words of 7th-order approximations. The approximations and scrambled-word passages were thus comparable from the point of view of word choice, but such comparability is not guaranteed with the prose passages. There were, however, no substantial differences in mean word length: there were 94.3 words per page of prose against 91.2 words per page of approximations or of scrambled words.

All materials were typed in grade-1, uncontracted braille, up to 40 characters and intervals per line and 14 lines per page, on standard heavy paper. A numbered grid of black lines was traced on each page, so that each character occupied exactly the centre of one cell and could easily be located on the recordings.

For the experiment proper, three passages of prose, each two pages long, and six passages of approximations and of scrambled words, each one page long, were used. Additional pages of each material were used for practice, in the way described in the procedure section.

Subjects

The "main group" of subjects, on whose performance the data to be described will, unless stated otherwise, always be based, comprised 24 blind persons, five female and 19 male, aged 18 to 64 years. They were recruited on the basis of information provided by several institutions. Only persons who were known for reading habitually with two hands and who confirmed that such was the case were tested. Nine of the 24 subjects were considered congenitally blind, 9 had lost their sight between birth and the age of 6, and 6 later on. Braille instruction had begun at 6 years of age or earlier for 8 subjects, at 7 for 7, at 8 or 9 for 6 and above 9 years for the 3 remaining subjects. Seven subjects were students; all the 17 others had a paid occupation. All subjects were submitted to the handedness questionnaire described by Hermelin and O'Connor (1971). Two appeared to be ambidextrous, all others were right-handed. Subjects were questioned about the method by which they had learned braille, but the answers were largely useless.

The study was started with a group of 24 subjects, among whom 3 were later excluded from the main analysis because of extreme asymmetry in the one- handed performances of the two hands (fast hand speed more than double that of slow hand), and replaced by 3 new subjects. A preliminary report on the study (Mousty, Bertelson and Hublet, 1981) was based on the performance of the original group of subjects.

Procedure

Each subject was given the (admittedly fuzzy) instruction to read each passage aloud, at a sustained pace. For each type of material, one passage (i.e. two pages for prose and one page for approximations and for scrambled words) was read with the left hand (LH) alone, one with the right hand (RH) alone and one with the two hands (2Hs), in the same order for the three types of material, this order being varied across subjects in latin square fashion. The prose passages were always read first, and for the remaining tests half the subjects had the approximations and then the scrambled words, and the other half had the opposite order. The passages were allocated in such a way that the subjects who read passages 1-3 for the approximations had the scrambled-word passages derived from passages 4-6 and vice-versa. Every time new material was introduced, its characteristics were first described verbally, and then the subject practised by reading one page in his preferred way, which turned out always to be with the two hands.

Analysis

All the data to be described were obtained by playing back the recordings on a monitor. The present analysis will be based on time to read the 12 median lines of each 14-line page-i.e. the time elapsing between the moment at which the reading finger left the first character of the 2nd line and that at which it left the last character of the 13th line. These basic time scores were then transformed into speed scores expressed in words per mm.

Results

Although, like most braille readers, the subjects were used to reading mainly contracted braille ("braille abrégé"), none reported any difficulty in dealing with the present uncontracted material. Very few errors were made, and these were mainly limited to proper names. The only global measure of performance available for analysis was thus reading speed.

One-Handed Reading

All subjects were able to read with either hand alone. Individual mean speed of one-handed reading of prose with the LH and with the RH are given in scattergram form in Figure 1.

Figure 1 is a graph

#FIGURE:1

Figure 1. One-handed reading of prose: Individual values of mean speed in words/mm for RH (abcissa) and LH (ordinate). Filled circles = fast two-handed readers; empty circles slow two-handed readers; crosses = readers eliminated from main analysis for extreme superiority of one hand. Oblique lines show limiting values of "Relative LH superiority" (see definition in Table I).

Wide individual variations are apparent in pattern of difference between the hands. There is, however, no overall tendency toward the superiority of one particular hand: 14 subjects of the main group read faster with the RH and 10 with the LH, and the mean speed is 75.2 words/mm. for the LH and 76.8 for the RH.

The same absence of overall hand superiority is manifest for non- prose material, as is apparent in Figure 2.

Figure 2

#FIGURE:2

Figure 2. Reading speed in words/mm with the different hand combinations for the different materials.

On the other hand, individual patterns of hand superiority are reliable across materials. The relative left hand superiority coefficient (RLHS, see legend to Figure 1 for definition) was calculated for each subject and each material. Product- moment correlations across subjects range from 0.66 for prose and approximations to 0.76 for prose and scrambled words. The two ambidextrous subjects display no exceptional pattern of hand difference. When the 24 subjects are ordered according to decreasing RLHS, they occupy ranks 12 and 17.

Two-Handed Reading

In Figure 3, speed of two-handed reading of prose has been plotted against speed of one-handed reading with subjects' faster hand. It appears that all subjects read faster with two hands. Large individual differences are observed again in size of relative gain from two-handed reading (defined in the legend to the figure) which ranges from 1 to 62%, with a mean value of 33%.

Figure 3

#FIGURE:3

Figure 3. Speed of two-handed reading of prose (ordinate) as a function of speed of one-handed reading with the faster hand (abcissa). Filled circles = fast two-handed readers; empty circles = slow two-handed readers; crosses = readers eliminated from main analysis for extreme superiority of one hand. Oblique lines show litniting values of "relative two-handed gain" (see definition in Table I).

Table I shows speed of two-handed reading separately for three groups of subjects arranged on the basis of hand superiority in one- handed reading, following Bürklen's and Fertsch's procedure. Although the pattern reported by Fertsch is replicated (median group faster than RH dominant, and LH dominants lowest of all), the differences are small and far below significance.2

2in a preliminary description of the present data (Mousty, Bertelson and Hublet, 1981), Fertsch's effect on speed of two-handed reading was found significant. That was apparently due to a strong influence on group means of the three extreme subjects who have now been replaced, but whose data were still included in the first analysis.

Table I

#TABLE:I

A stronger superiority of the median group is apparent for another index of performance, relative gain from two- handed reading, but nevertheless falls short of significance [F(2, 21)=1.47; p>0.l0]. Another way of analysing the same aspect of the data was through the correlation between two-handed gain and the absolute value of RLHS: a product-moment coefficient of -0.47 (df=23;p=0.05) is obtained. There is thus a definite suggestion that the subjects who benefit most from two-handed reading are those whose slow hand is not too inferior to the fast one.

Type of Material

Figure 2 shows that reading speed is clearly affected by type of material: whether with one or with two hands, approximations are read faster than scrambled words and prose faster than approximations.

In Figure 4, the same data have been arranged in a different way.

Figure 4

#FIGURE:4

Figure 4. Mean reading speed as a function of type of material, with subjects overall two-handed reading speed and hand combination as parameters.

For one-handed reading, data for respectively the faster hand of each subject-i.e. the hand whose mean speed of reading over the three materials is greatest-and for the slower one, have been pooled. On the other hand, the results are given separately for the 12 fast subjects, meaning those with the greater mean speed of two-handed reading for the three materials, and for the 12 slow ones.

The question of interest about these results concerns the variation of effects of materials with hand combination and with subjects' overall speed. It can be considered at the level of absolute effect or of relative ones. Only the analysis based on relative effects is reported here, since it was the one that offered the better chance of supporting the compensatory hypothesis. For each pair of materials taken two by two, relative speed change was calculated for each subject and each hand combination and submitted to a 2-factor (subjects x hand combination) ANOVA. The effect of hand combination is significant for none of the three pair- wise comparisons. The effect of subjects group (fast vs. slow) is significant only for the prose/approximation comparison [F(1, 22)=10.28; p<0.01]. It goes in the direction predicted by the compensatory hypothesis (larger effect in slow readers).

Congenital vs. Late Blindness

The comparison of the nine congenitally blind subjects with the 15 others reveals no strong difference. Mean prose reading speed is higher in the congenitals for the two-handed [123.0 against 106.3 words/mm, t(22)= 1.19; p>0.l0] and for the LH condition [85.2 against 69.2, t(22)= 1.79; p<0.05] and slightly lower for the RH condition [75.8 against 77.4, t(22)=0.13, NS]. Closer examination shows, however, that the inferiority of the late blind for two-handed and LH reading is largely due to the poor performance of the three subjects who, due to late onset of blindness, had started braille instruction later than 9 years of age (two at 10, one at 18). When these subjects are left out of the analysis, the congenital-late difference for LH reading ceases to be significant. It is thus possible that age at onset of blindness affects eventual reading performance only to the extent that it delays the start of braille instruction. Speculation about the fact that the three late starters were slow especially with the LH should of course be suspended until confirmation from a larger sample.

Discussion

Two main findings emerge from the consideration of hand effects in one-handed reading:

No overall tendency toward the superiority of one hand nor any interaction with type of reading material is apparent. Regarding the main issue of hand cooperation in reading, these two findings are inconsistent with all conceptions based on a fixed division of functions between the hands, such as Heller's (1905) notion of a "synthetic" RH opposed to an "analytic" LH nor Hermelin and O'Connor's notion of a systematic superiority of the LH due to direct access to the right hemisphere. Even if negative, the result is important, for with the current interest for hemispheric specialization (Bertelson, 1982) comes a strong bias to expect asymmetries to show up in all sorts of activities with a lateral component.

As we have seen in the introduction, LH superiority has been observed consistently in sighted non-readers asked to discriminate braille characters by touch, which might indicate right-hemisphere superiority for that type of operation. One possibility, which we have examined elsewhere (Mousty, Bertelson and Kurrels, 1982; Mousty and Bertelson, 1984), is that in actual reading this effect is counteracted by another component linked to left-hemisphere superiority for linguistic interpretation. A similar possibility has been considered by Bryden and Allard (1976) in relation to their finding that the direction of field advantage for letter recognition depends on typeface.

There are large and reliable individual differences in pattern of hand superiority. There are several possible sources for the variability. It might reflect variations in the respective efficiencies of right-hemisphere vs. left-hemisphere-based operations involved in braille reading. One particular version of that general notion links the variations to degree of mastery of the skill. The notion would fit easily with the two-process view evoked in the preceding paragraph. One could imagine a right- hemisphere-based character-encoding process that gets progressively automatized and plays a lesser relative role in determining overall reading performance. It would also be consistent with the fact that Hermelin and O'Connor's children showed stronger evidence of LH superiority than their adult subjects (albeit with a different task) and with Fertsch's (1947) data showing a shift from LH dominance to RH dominance with increasing reading skill. Millar (1984) has recently reported a significant shift from LH superiority to RH superiority with increasing reading skill in disconnected letter processing by blind children. In the present study, however, only a vanishingly small difference was observed in two-handed reading speed between RH- dominant and LH-dominant subjects. Degree of mastery, even if it might account for some of the individual differences in hand effects reported by other authors, can thus not explain those observed in the present study. Not much evidence was obtained either for the effective ness of a factor possibly related to mastery, congenital vs. late blindness.

Another possible source of variability is the specific practice received by each hand at school and also during the normal practice of two- handed reading. As Villey (1931) suggested, there are probably import ant differences between the educational practices that prevail in different places. On the other hand, in a same educational environment, different readers may develop individual patterns of two-handed reading that make different demands on the two hands. The lack of reliable information on the methods by which our subjects had learned reading made it impossible to pursue the matter on the present data.

The superiority of two-handed over one-handed reading, which was observed in every single subject, is confirmed beyond doubt. The main question is, of course, how the gain is obtained. To answer it, it is essential to analyze patterns of hand movements. In the meantime, relevant information is brought by the finding, which replicates Bürklen's (1917) theory that the relative gain from two-handed reading is inversely correlated with difference in ability between the hands. In other words, for a given capacity of the superior hand, two-handed performance depends on the independent reading capacity of the slower hand. This suggests strongly that both hands participate in the collection of textual information.

In conclusion, the analysis of overall reading speed has brought no support for the notion that the hands play different functions in braille reading. For the moment, the most parsimonious view is that whatever differences in capacity exist between the hands, these are quantitative rather than qualitative.

Speed of reading was clearly enhanced by contextual constraints. The comparison between prose and non-prose materials may not reflect a pure effect of contextual constraints, since the comparability of the vocabularies used in the two categories of materials is not guaranteed: it is possible that the faster reading of prose is partly due to the use of more familiar words in Simenon's easy prose. But no such problem exists concerning the comparison between scrambled words and approximations, where adequate controls for word choice were applied, so that the observed differences in reading speed can safely be attributed to greater conformity of word order to linguistic constraints in the approximations.

In Morton's (1964) study of visual reading, average speed increased 42% from 1st-order approximations (i.e. scrambled words) (156 words/mm) to 6th-and 8th-order approximations (mean of 221 words/mm). The corresponding relative increase is 26% for two-handed reading. Morton, on the other hand, obtained little further acceleration from 8th- order approximations to prose: only 4%, against 18% in two-handed reading. From scrambled words to prose, the acceleration is 48% in Morton's study and 49% here. So, depending on the particular data one chooses to compare, the effects of context are either comparable in the two studies or slightly smaller in braille. There is certainly no support for the prediction from the compensatory hypothesis that braille readers, having to spend more time encoding the characters, would make greater use of contextual information.

The other tests available in our data also do not produce much support for the compensatory hypothesis. Contrary to the prediction, the effects of material were not stronger in one-handed reading for the slow hand than for the fast one, nor were they stronger in one-handed than in two-handed reading. The other prediction, that poor readers would benefit from context more than good readers, has been confirmed at the level of one only of the three pair-wise comparisons between materials. As explained above, the comparison between approximations and prose is the one that is most likely to be affected by other factors than contextual constraints.

Finally, it should be noted that our results do certainly not support the older notion that good readers are those who depend more on context (Smith, 1971; Goodman, 1976).

References

Bertelson, P. (1982). Lateral differences in normal man and lateralization of brain function. International Journal of Psychology, 17, l73-210.

Bradshaw, J. L., Nettleton, N. C. and Spehr, K. (1982). Braille reading and left and right hemispace. Neuropsychologia, 20, 493-500.

Bryden, M. P. and Allard, F. (1976). Visual hemifield differences depend on typeface. Brain and Language, 3, 191-200.

Bürklen, K. (1917). Das Tastlesen der Blinden-Punktschrift. Beiheft zur Zeitschrift für angewandte Psychologie, 16, 1-66.

Fertsch P. (1947). Hand dominance in reading braille. American Journal of Psychology, 60, 335-349.

Foulke, E. (1982). Reading braille. In W. Schiff and E. Foulke (Eds.), Tactual Perception. Cambridge: Cambridge University Press. Pp. 168-208.

Grasemann, P. (1917). Eme Untersuchung über das Lesen der Blinden. Beiheft zur Zeitschrift für angewandte Psychologie, 16, 67-72.

Goodman, K. S. (1976) Reading: A psycholinguistic guessing game. In H. Singer and R. Ruddell (Eds.), Theoretical models and processes of reading. Newark, Del.: International Reading Association.

Harris, L. J. (1980). Which hand is the "eye" of the blind? A new look at an old question. In J. Herron (Ed.), Neuropsychology of left-Handedness. New York: Academic Press.

Heller, T. (1904). Studien zür Blindenpsychologie. Leipzig: Engelmann.

Hermelin, B. and O'Connor, N. (1971a). Right and left-handed reading of braille. Nature, 231, 470.

Hermelin, B. and O'Connor, N. (1971b). Functional asymmetry in the reading of braille. Neuropsychologia, 9, 431-435.

Holland, B. F. and Fehr, C. A. (1942). The reading of braille music. Outlook for the Blind, 36, 25-29.

Kusajima, T. (1974). Visual reading and braille reading: An experimental investigation of the physiology and psychology of visual and tactual reading. New York: American Foundation for the Blind.

Lowenfeld, B., Abel, G. L. and Hatlen, P. L. (1969). Blind children learn to read. Springfield, Ill.: Thomas.

Marslen-Wilson, W. D. and Tyler, L. K. (1980). The temporal structure of spoken language understanding. Cognition, 8, 1-71.

Millar, S. (1984). Is there a "best hand" for braille? Cortex, 20, 75-87.

Miller, G. A. and Selfridge, J. A. (1950). Verbal context and the recall of meaningful material. American Journal of Psychology, 63, 176-185.

Mommers, M. J. C. (1980). Braille reading: Effects of different hand and finger usage. Journal of Visual Impairment and Blindness, 74, 338-343.

Morton, J. (1964). The effects of context upon speed of reading, eye movements and the eye-voice span. Quarterly Journal of Experimental Psychology, 16, 340-351.

Mousty, P., Bertelson, P. and Hublet, C. (1981). Les roles respectifs des mains dans la lecture du braille: Une étude exploratoire. In A. Harrisson-Covello, H. Herren, G. C. Lairy, P. Oléron and F. Robayc, Les enfants handicapés. Paris: Presses Universitaires de France.

Mousty, P. and Bertelson, P. (1984). Influence of task demands on hand differences in braille reading. 7th European Conference of International Neuropsychological Society, Aachen.

Mousty, P., Bertelson, P. and Kurrels, V. (1982). Effect of reading hand in one- handed apprehension of braille. 5th. European Conference of International Neuropsychological Society, Deauville.

Myers, D. H. (1976). Right- and left-handed counting of braille dots in subjects unaccustomed to braille. Bntish Journal of Psychology, 67, 407-412.

Peffetti, C. A., Goldman, S. R. and Hogaboam, T. W. (1979). Reading skill and the identification of words in discourse context. Memory and Cognition, 7, 273-282.

Rudel, R. G., Denckla, M. B. and Spalten, E. (1974). The functional asymmetry of braille letter learning in normal, sighted children. Neurology, 24, 733-738.

Smith, F. (1971). Understanding reading. New York: Holt, Rinehart & Winston.

Smith, J. M. (1929). Which hand is the eye of the blind? Genetic Psychology Monographs, 5, 209-252.

Stanovich, K. E. (1980). Toward an interactive-compensatory model of individual differences in the development of reading fluency. Reading Research Quarterly, 16, 32-71.

Villey, P. (1931). Psychologie de la lecture tactile. Journal de Psychologie, 28, 214-249.

Revised manuscript received 19 November 1984