The Reading Rates for Contracted and Uncontracted Braille of Blind Norwegian Adults
J.M. Bruteig
Abstract: This article reports on a study of 35 late-blind Norwegian adults, aged 20-70, in reading equivalent texts and single words in contracted and uncontracted braille. The objective was to register their general reading behavior and to measure their reading rates for both types of Norwegian braille. Although the subjects' reading rates increased when they read contracted versions of both the text passages and single words, their rates were higher with the contracted single words. Furthermore, the slow readers saved more time than did the fast readers when they read texts in which contractions were used frequently.
In recent years, Scandinavian experts in the field of blindness have disagreed about the readability of uncontracted versus contracted braille. One group has maintained that the use of contracted words increases the reading rate, while the others have argued that contracted words disturb the reading process, thereby reducing the reading rate.
The latter group has contended that persons who lose their sight and learn braille as adults ("the late blind") have special difficulty reading contracted braille. This controversy stimulated the author to investigate the readability of contracted and uncontracted braille by adult late-blind persons.
The research, conducted in partial fulfillment of the author's thesis at the Norwegian Institute of Special Education, Oslo, was purely descriptive, without the use of any kind of systematic training of the subjects. The purpose was to obtain a picture of the reading ability of the subjects in their daily lives. The subjects' potential for improving their reading abilities was not tested.
Population and experimental group. Blind and severely visually impaired persons vary widely in their reading skills and techniques. However, the late-blind may be regarded as a group with the following particular characteristics:
1. they have not had braille as their basic natural reading and writing medium;
2. they previously read visually; as tactile readers, their reading ability and the amount they read is greatly reduced; and
3. they have a lower average reading speed than do the congenitally blind and have less general competence in braille.
Some late-blind persons, especially older people, have not managed to adapt. They meet their reading needs exclusively by using the sound media and use braille only for simple tasks such as notes.
In the experiment reported here, the population was limited to those late-blind persons who were using braille to some extent in their daily lives. In Norway, there are no more than 200-500 individuals in this category of blind readers, widely geographically disbursed.
Because great practical (and theoretical) difficulties would be encountered in composing a statistical sample from this small and widespread population, the author decided to contact as many potential subjects as possible in the Oslo area.
This search yielded 35 subjects who seemed to represent a cross section of the population of late-blind persons in Norway in that
a. there were roughly the same number of male and female subjects (20 men and 15 women);
b. the subjects varied widely in age (from 20 to 71 years, with an average of 40);
c. the sample was evenly distributed in relation to the age at which braille was learned, the length of time braille was used, and other relevant types of reading experiences; and
d. both intellectual and practical types of employment were rcpresented in the sample.
All 35 subjects had learned contracted braille; 40 percent read mainly contracted braille, 46 percent normally read both contracted and uncontracted braille, and the remaining 14 percent hardly read any printed braille. There was no systematic correlation between reading experience in contracted and uncontracted braille, respectively, and the test results obtained for the comparable type of braille used during the experiment.
Methods
The ability of the late-blind subjects to read contracted and uncontracted Norwegian braille was studied in two ways: 1) by measuring the reading rate in continuous, normal contracted text and in less-contracted text and comparing these reading rates with the rate obtained reading an uncontracted continuous text and 2) by comparing the rate of identification of contracted single words with the rate of identification of the same words written in full.
It was difficult to choose a method of measuring the subjects' reading rates of passages of text because their reading rate was expected to be as low as 15 words per minute. Therefore, the texts on which they were to be tested could not exceed 180 words if the tests were to be completed within a reasonable time.
In addition, because the subjects had no experience reading braille aloud, the tests had to be conducted through silent reading. However, this meant that it was necessary to find a way of ensuring that the texts were read with sufficient precision and with a reasonable understanding of the content.
These conditions, particularly the need for texts consisting of no more than 180 words, made it essential to formulate the texts with regard to normality and equivalence of language and content. Two criteria were used to assess the normality of the language-the length of words and the frequency of words in the text.
Both the word-length pattern (the distribution of short and long words) and the distribution between frequently used and less common words had to be as similar as possible to those in ordinary Norwegian texts.
Readability in contracted and uncontracted braille could not be compared by reading the same text in contracted and uncontracted braille.
Uncontracted 1 2 3 4 5 6 7 8 9 10+
Contracted 1 2 3 4 5 6 7 8 9 10+
0 % 25% 50% 75% 100%
Figure 1. Word-length pattern in Norwegian braille (continuous text). The strips show the relative distribution of words of 1 to 9 characters and words of 10 characters or more. The upper strip shows the uncontracted braille and the lower strip, the contracted version.
The evaluation had to be made using different passages of text, which thus had to be equivalent in language and content. That is, the texts had to have, as far as possible, the same word-length pattern, distribution of frequently and rarely occurring words, and thematic content.
An earlier study on the length of words in contracted and uncontracted braille and a statistical count of word lengths in Norwegian newspapers (Heggstad 1982) were used as bases for ensuring normality. and equivalence. Figure I is a graphic presentation of the normal percentage distribution of word lengths.
Two simple narrative texts were chosen, each consisting of about 360 words. Each text was divided into two equal parts and written in uncontracted braille, in ordinary contracted braille, and in a less-contracted version that contained only the 37 contractions expressed by a single character in standard Norwegian contracted braille. The texts were then processed linguistically until the word-length pattern was as near to the normal as possible in the uncontracted and the contracted versions. This meant that the texts were also mutually equivalent in relation to the distribution of short and long words (Tables la, lb).
Table la. Percentagedistribution of word lengths in the uncontracted reading text material.
Number Normal WordDistribution
Of Letter (parts of text)
Letters Contents A1 A2 B1 B2
1 4.63 5.1 5.0 5.0 5.0
2 20.52 19.3 20.7 20.7 20.7
3 22.16 23.9 21.8 23.5 22.9
4 10.63 11.9 12.3 12.3 11.7
5 10.00 12.5 13.4 10.6 12.8
6 8.00 11.4 11.2 12.3 11.2
7 5.26 5.1 5.0 4.5 4.5
8 3.68 4.5 4.5 3.9 3.9
9 5.00 2.3 2.8 2.8 3.4
10 or more 10.11 4.0 3.4 4.5 3.9
Total 99.99 100.0 100.1 100.1 100.0
Table lb. Percentage distribution of word lengths in the contracted reading text material.
Number Normal Word Distribution
Of Character (parts of text)
Characters Content A1 A2 B1 B2
1 32.74 32.4 31.3 32.4 31.3
2 18.21 18.2 17.3 16.8 17.9
3 8.89 9.1 10.6 9.5 11.2
4 8.89 11.9 12.8 11.2 10.6
5 7.74 9.7 8.9 12.3 11.7
6 4.53 6.3 5.0 7.8 6.1
7 4.47 4.0 5.0 2.8 4.5
8 3.58 2.3 3.9 2.2 2.2
9 3.26 2.8 2.8 2.8 1.1
10 or more 7.68 3.4 2.2 2.2 3.4
Total 99.99 100.1 99.8 100.0 100.0
The control of frequently used and less common words showed only small deviations among the different texts, which may have been the result of the homogeneity obtained in the content of the contractable contracted words. The most usual contractions represent, after all, the most frequently used words in the language. The purpose of the linguistic treatment was to ensure the greatest possible uniformity in readability of the four parts of the text. Thus, pretests with congenitally blind persons yielded even values for reading rates in all parts of the text.
Test procedures
Continuous text
The texts were written on paper using the Perkins Bmiller and mounted into ordinary braille magazines to provide the most natural physical conditions for the test situation. The subjects were allowed to chose their own reading positions (the height of the table, for example). The reading tests were arranged so that the subjects read the first half (180 words) of a text in contracted braille and, after a short relaxing pause, the other half in uncontracted braille during the first session. In the second session (two or three days later), the first half of the other text was read in uncontracted braille and the second half in the less-contracted version. Lots were drawn to decide the order of the two texts. To reduce the possibility of unpredictable breaks in reading owing to poor reading ability, it was decided that each subject should read each half of the text twice and that the second reading would be taken as the main result.
The subjects' understanding of the content was controlled by means of "word choices" (Madison, 1979). In each passage of text (180 words), at intervals of about 25 words, two short words with the same grammatical function (such as two prepositions) were inserted in brackets, one of which fitted into the context and one which did not.
When the subject reached these brackets, he or she was to read the correct word aloud. (The rest of the text was to be read silently.) Thus, it was possible to record seven times in each passage whether the text was really read and whether the subject had a reasonable and adequate understanding of the content. All 29 subjects who completed the second reading mastered at least six or seven choices and were considered to have an equal level of understanding.
Table 2. Individual reading rates for continuous text.*
Ranking number Contracted Uncontracted Percentage difference
1 15 12 25
2 14 14 0
3 20 16 25
4 20 17 17
5 22 20 10
6 24 20 20
7 25 22 13
8 28 23 21
9 28 24 16
10 30 24 25
11 34 25 36
12 33 27 22
13 3.3 29 13
14 37 29 27
15 44 34 29
16 41 35 17
17 49 36 36
18 41 37 10
19 44 39 12
20 53 39 35
21 53 41 29
22 40 41 -2
23 45 42 7
24 42 43 -2
25 53 43 23
26 51 45 13
27 67 51 31
28 64 53 20
29 72 67 7
M 38.7 32.7 18.5
*Subjects B through F were unable to complete the reading tests for continuous text and hence were excluded om this table.
Single words
The single-word identification test was presented at the second meeting before the reading of continuous text. For that test, the subjects read aloud 28 single character contractions representing high frequency words of two and three letters several times in random order and then repeated this procedure for the uncontracted versions of the 28 contractions. A tape recording was used in scoring the results, and the average rate of identification of the contracted and uncontracted words was calculated for each subject.
The rate of identification of single words was measured using a tactile tachistoscope that could display a line of 15 characters simultaneously (Hunstad, 1981). The exposure time was adjusted individually for each subject so that tne exposure was interrupted before the reading of the line could be completed.
Results
Continuous Text
Six subjects (subjects B through F) were unable to complete the reading tests for continuous text either in contracted or uncontracted braille. The others had an average reading rate of 30-40 words per minute with a distribution of 15-70 words per minute (Table 2). The arithmetic mean for ordinary contracted braille was 38.7 words per minute and for uncontracted braille, 32.7 words per minute. (The less contracted text will be discussed later.)
On the average, the ordinary contracted text was read 18.5 percent faster than the text that was not contracted (p < . 01), but there were large individual variations in this respect. Some read the contracted text up to 36 percent faster than the uncontracted text, and two subjects at the opposite end of the scale read the contracted text 2 percent slower than they read the uncontracted text (Table 2). Thus, it was obvious that contracted braille did not have the same effect on the reading rate of all the subjects. Neither the general background and reading experience of the subjects nor the extent to which an individual subject had used contracted or uncontracted braille in his or her daily life was sufficient to account for this large variation. However, when the subjects were placed into groups according to their rate of reading ("fast," 40-70 words per minute; "medium," 25-40 words per minute; and "slow," 15-25 words per minute), the individual percentage differences tended to be somewhat greater (not significant) for the slow readers than for the fast readers (Table 3). This result was surprising, since it had previously been assumed that the effect of contractions on a person's reading rate was related to competence in reading braille. But in this experiment, the effect seemed to be just the opposite.
Table 3. Percentage differences in the average reading rates of slow, medium, and fast readers for contracted and uncontracted words.*
Type of Number of Percentage
reader subjects difference
slow 10 17.2
Medium 10 23.7
Fast 9 14.0
Total 29 18.5
*Subjects B through F were unable to complete the reading tests for continuous text and hence were excluded from this table.
Single Words
The results of these measurements showed a significant difference of 33.6 percent in the rate of identification of contracted compared to uncontracted words (p <.0l) (Table 4). When the subjects were grouped into fast, medium and slow readers, the difference in the rate of identification of contracted and uncontracted words was 23.0 percent for the fast readers, and 41.2 percent for the slow readers (p<.0l). Thus, the contractions were of decidedly greater importance for the slow readers than for the fast readers (Table 5).
This tendency becomes even more obvious if one looks at the group of subjects who could not adequately read continuous text but who could participate in the identification of single words. For these persons, there was a difference of 56.2 percent in the rate of identification of contracted versus uncontracted words, indicating that, of all the groups, this group derived the greatest benefit from the contractions.
Discussion
The percentage differences in the subjects' reading rates for contracted versus uncontracted Norwegian braille were 18.5 percent for continuous text and 33.6 percent for the identification of single words. The time saved by using contractions was thus much greater for the identification of single words than for the reading of continuous text passages.
However, the contractions in Norwegian braille are moderate; 50-60 percent of the words in a text are written in full without any form of contraction. It should be pointed out, moreover, that if the contractions increase the reading rate, it is not to be expected that the effect of the contractions will be the same when subjects read text passages as when they identify contracted and uncontracted words on a tachistoscope. The result will obviously be influenced by the frequency of the contractions in the text.
Table 4. Individual identification rates of single words.
Ranking number Contracted Uncontracted Percentage difference
1 32 18 77
2 39 22 77
3 35 26 34
4 44 31 41
5 56 31 80
6 46 33 39
7 52 34 52
8 57 41 39
9 63 41 53
10 50 42 19
11 62 42 47
12 66 42 57
13 60 44 36
14 58 47 23
15 59 48 22
16 59 51 15
17 73 55 32
18 74 55 34
19 71 56 26
20 75 56 33
21 88 59 49
22 66 61 8
23 68 61 11
24 65 62 4
25 83 67 23
26 95 68 39
27 102 69 47
28 94 71 32
29 86 72 19
30 92 75 22
31 Ito 81 35
32 80 85 - 6
33 108 98 10
34 119 104 14
(A - 22)
M 70.2 53.4 33.6
Table 5. Average number of words per minute of slow, medium, and fast readers, and subjects B-F, who were unable to complete the continuous text reading.
Number of Contracted Uncontracted Percentage
Type of reader subjects text text difference
Slow 10 59.0 42.2 41.2
Medium 10 75.6 60.6 24.4
Fast 9 93.2 76.6 23.0
29 75.3 59.2 29.8
Subjects B-F 5 40.4 26.2 56.2
34 70.2 54.3 33.7
This presumption seems to be confirmed by the results obtained from the reading of less-contracted text. In that case, the average difference in the subjects' reading rates for contracted and uncontracted braille was only 7.2 percent (p<.0l), probably because there were greater intervals between the contractions than in the ordinary contracted braille (Table 6).
Once again, the effect of the contractions varied considerably from person to person. For the fast readers, there was a nonsignificant difference of only 3.0 percent, but for the slow readers there was a significant difference of 9.5 percent in the lesser contracted text (p<.0l) (Table 7). Thus, as was true in relation to the ordinary contracted text, it was the slow readers who derived the greatest benefit when the braille was contracted.
This finding raises another issue: the participants in the experiment had not previously read this form of lesser contracted braille that contains only a small number of contractions. Therefore, the study also was a test of the flexibility of the subjects in using an untraditional method of contraction.
What seems to have occurred was that the fast readers became confused by this method, as indicated by the uncertain, nonsignificant result for them, whereas the slow, inexperienced readers did not have such a problem and, hence, were able to use this method of contraction to a much greater degree than could the fast readers. In other words, the readability of contracted braille is not always dependent on highly developed reading ability and accurately learned systems of contraction-at least not as far as slow, late-blind readers are concerned.
Nevertheless, it seems that the degree of benefit derived by this group of readers was connected with frequently occurring contractions expressed by a single character.
Thus, there is reason to assume that the increased reading rate in contracted braille is actually due to the fact that many common words are shorter in contracted braille than in texts that are written in full and that this effect is greater than the assumed negative effect of the reader having to interpret the meaning of the contractions.
Obviously, this negative effect will be smallest for the most-common words, and the positive effect of the contractions will be maximal when most of these frequent words are contracted to a single character.
Table 6. Individual reading rates in the less-contracted text.*
Ranking number Less-contracted text Uncontracted text Percentage difference
1 16 14 14
2 17 14 21
3 19 19 0
4 19 19 0
5 23 22 4
6 26 22 18
7 24 23 4
8 25 23 8
9 28 25 12
10 31 27 14
11 27 28 - 4
12 31 29 6
13 37 30 23
14 36 31 16
15 34 33 3
16 43 36 19
17 44 38 15
18 38 39 3
19 43 41 4
20 46 41 12
21 48 45 6
22 53 45 17
23 44 48 -8
24 56 50 12
25 45 51 -12
26 52 53 -2
27 61 60 1
28 62 61 1
M 36.8 34.5 7.2
*Subjects B through F were unable to complete the reading tests for continuous text and hence were exeluded from this table.
Table 7. Percentage differences in the average reading rates of slow, medium, and fast readers for less-contracted text and uncontracted text.*
Type of Number of Percentage
reader subjects difference
Slow 10 9.5
Medium 9 8.8
Fast 9 3.0
Total 28 7.2
*Subjects B through F were unable to complete the reading tests for continuous text and hence were excluded from this table.
The contractions did not lead to negative effects for the slow readers when they read text passages or when they identified single words on a tachistoscope. Therefore, it must be concluded that the positive effect of the contractions on the reading rate is greater for slow readers than for fast readers of braille.
Again, it should be noted that the relatively moderate contractions in Norwegian braille may have accounted for the more favorable results with the slow readers. A different result might have been obtained with a more complex system of contraction that places greater demands on the insight and reading experience of the reader.
Nevertheless, the study strongly indicated that totally uncontracted braille is a relatively poor reading medium for lateblind slow readers of braille. Thus, there is reason to investigate further where the point of balance lies in regard to the readability of braille-that is, the degree of contraction that is optimal for blind persons who are obliged to learn braille as adults.
References
Heggstad, K. (1982). Norsk frekvensordbok. Bergen: Universitets-forlaget.
Hunstad, E. (1981). Lesetreningsapparatet Braillemaster. Heggedal/Oslo: Haaland & Hamre A/S.
Madison, S. (1979). Lasdiagnoser Malmo: Liber Laromedel.
Jan Bruteig, Huseby Utdanningssenter for Synsheminede, G.L. Hovsetervei 3, 0768 Oslo 7, Norway.