Abstract
Aims
To develop standardised texts for assessing reading speed during repeated measurements and across languages for normal subjects and low vision patients.
Methods
10 texts were designed by linguistic experts in English, Finnish, French, and German. The texts were at the level of a sixth grade reading material (reading ages 10–12 years) and were matched for length (830 (plus or minus 2) characters) and syntactic complexity, according to the syntactic prediction locality theory of Gibson. 100 normally sighted native speaking volunteers aged 18–35 years (25 per language) read each text aloud in randomised order. The newly designed text battery was then applied to test the reading performance of 100 normally sighted native speaking volunteers aged 60–85 years (25 per language).
Results
Reading speed was not significantly different with at least seven texts in all four languages. The maximum reading speed difference between texts, in the same language was 6.8% (Finnish). Average reading speeds (SD) in characters per minute are, for the young observer group: English 1234 (147), Finnish 1263 (142), French 1214 (152), German 1126 (105). The group of older readers showed statistically significant lower average reading speeds: English 951 (97), Finnish 1014 (179), French 1131 (160), German 934 (117).
Conclusion
The authors have developed a set of standardised, homogeneous, and comparable texts in four European languages (English, Finnish, French, German). These texts will be a valuable tool for measuring reading speed in international studies in the field of reading and low vision research.
Keywords: standardised texts, reading performance, low vision
Reading problems are the main complaint of patients with visual field defects involving the visual field centre, the main pathology being age related macular degeneration (AMD). Efficient reading requires not only sufficient visual acuity to resolve the letters, but also a sufficient extent of visual field to process words and plan eye movements.1,2,3,4,5,6 Patients with AMD must learn to use parafoveal retinal areas instead of non‐functional foveal vision, and the text must be magnified to compensate for lower resolution outside the fovea.
Most low vision examinations of patients with AMD involve an assessment of reading performance. Reading speed is a critical component of that evaluation, along with comprehension and endurance. It has been shown by many investigators that letter acuity is not a good predictor of magnifier aided reading speed (see, for example, Ahn and Legge1,4,5,6,7). They showed that a standardised clinical reading test can give a valid prediction of the reading speed a low vision patient is likely to achieve with a magnifier. Many different reading tests have been designed for use in low vision examinations based on random words, sentences, and continuous text. However, most of these tests are not standardised for content and are not available in a variety of languages. In the United States, the Minnesota Reading Test (MNREAD) was devised as a standardised psychophysical test of reading. The test consists of single, simple sentences with equal numbers of characters, with the vocabulary selected from words appearing with high frequency in second and third grade reading material.8,9,10 MNREAD can be used to measure reading acuity, maximum reading speed, and critical print size (the smallest letter size for maximum reading speed). Translations into other languages are under way. Radner and his group created a similar test set in German. The sentences consist of the same number of words, syllables, and characters taken from third grade reading material. All have a similar syntactic structure (for example, all contain relative clauses).11
Both of these test sets consist of single and relatively short sentences. But in order to gain a more accurate measurement of reading speed and information about a patient's reading fluency, it is desirable to assess the patient's reading performance on a longer text passage. The aim of the study was to develop standardised texts for longer paragraph reading as a new tool for reading, rehabilitation, and low vision research.
Methods
Texts
Text sets of 10 passages each were created in British English, Finnish, French, and German. The following criteria were used to ensure that the texts were comparable within and across languages.
Content
The texts were inspired by excerpts from an encyclopedia recommended for children aged 9–11 years and from reading material for sixth grade reading (age 10–12 years). All of the texts were self contained passages of general interest. To ensure that the main factor responsible for difficulties in reading is vision rather than intellectual abilities, the texts were not too demanding in terms of content and prose style.
Length
All of the texts were of equal length (830 (plus or minus 2) characters). The number of characters is used as the measure for text length rather than number of words, since words vary considerably in length across languages.
Word frequency
Several experiments have demonstrated that word frequency affects reading speed. To account for the effect of lexical frequency in the texts, only words with a lexical frequency of 0.0001% or higher were used.12
Syntactic complexity
The time needed to read and comprehend a sentence is also affected by its syntactic structure. Each word read needs to be integrated with text previously read. Considering an example case will help to illustrate this point:
(1) “The cotton clothing is made of grows in Mississippi.”
When reading the sentence in (1), the reader typically stops when he reads the word “grows” and restarts the reading process at the beginning of the sentence. This is because “grows” does not fit into what the reader has understood the sentence to mean so far. Taking the phrase “the cotton clothing” to be the subject of the sentence, one expects a noun phrase specifying the material after “is made of.” Instead, the reader is confronted with the verb “grows.” This conflict is resolved if instead of grouping “clothing” together with “cotton,” one realises that “clothing is made of” is actually a relative clause modifying the noun “cotton.” The sentence in (1) is a reduced version of (2).
(2) “The cotton which clothing is made of grows in Mississippi.”
This example is extreme with respect to the time needed to integrate a particular word and the proportion of the sentence that has to undergo restructuring in order to do so. But similar processes are at work all the time in sentence comprehension.
The use of transformations in generating a sentence increases its syntactic complexity. The syntactic prediction locality theory of Gibson (1998, 2000) provides a metric of structural complexity.13,14 It associates the integration of a new word with a certain number of “energy units.”
Since the goal of the project was to develop routine texts, we could not simply use sentences that have all the same syntactic make‐up. We had to allow for sentences to vary in syntactic structure. Using Gibson's syntactic prediction locality theory, we were able to control for syntactic complexity in our texts by, for example, avoiding split verbs and extraposition of relative clauses in the German texts. We decided to limit the maximal integration cost for each word to four energy units.
Layout
The readability of any text is influenced by parameters such as font, and spacing.10,15 We chose a Times New Roman font, and inter‐line and letter spacing were kept as close as possible to newspaper standard.
The texts were translated by linguistic professionals. Basically, “translation” was conducted by using the content to create a text in the new language that would meet the criteria for text length, word frequency, and syntactic complexity.
Subjects
Normally sighted volunteers aged 18–35 years were recruited for the initial testing of the texts. In a subsequent study normally sighted volunteers aged 60 up to 85 were tested with the text battery. A total of 200 native speaking subjects (25 per language and age group) with minimum visual acuity of 0.6 (0.2 logMAR) and no history of ocular pathology were tested. All subjects gave their informed consent, and the examinations were conducted in accordance with the tenets of the Declaration of Helsinki. Ethics committee approval was secured for the study by the Institute for Ethics and History in Medicine, Tübingen, Germany.
Procedure
The 10 texts were printed in high contrast (98%) with a standard laser jet printer and laminated with matt surface. They were presented in random order. One text was uncovered at a time, and the subjects were instructed to read it aloud as fast as they could without making mistakes and without correcting mistakes. Oral reading times were measured.
Statistical analysis
An analysis of covariance was used to describe the differences between the 10 texts in the four languages. A one way analysis of variance with subsequent Tukey HSD test for the comparison of all pairs of groups was performed to compare reading speed of the young and old volunteers (see table 1).
Table 1 Statistical levels from Tukey HSD.
| Level | Text | Letter | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| English | ||||||||||
| 1 | 6 | A | ||||||||
| 2 | 7 | A | B | |||||||
| 3 | 1 | A | B | C | ||||||
| 4 | 5 | A | B | C | D | |||||
| 5 | 3 | A | B | C | D | |||||
| 6 | 4 | A | B | C | D | |||||
| 7 | 9 | A | B | C | D | |||||
| 8 | 8 | B | C | D | ||||||
| 9 | 10 | C | D | |||||||
| 10 | 2 | D | ||||||||
| Finnish | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 7 | A | ||||||||
| 2 | 5 | A | B | |||||||
| 3 | 3 | A | B | |||||||
| 4 | 2 | A | B | |||||||
| 5 | 6 | A | B | C | ||||||
| 6 | 4 | A | B | C | ||||||
| 7 | 10 | A | B | C | ||||||
| 8 | 9 | B | C | |||||||
| 9 | 8 | B | C | |||||||
| 10 | 1 | C | ||||||||
| French | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 9 | A | ||||||||
| 2 | 4 | A | B | |||||||
| 3 | 5 | A | B | C | ||||||
| 4 | 6 | A | B | C | D | |||||
| 5 | 3 | A | B | C | D | |||||
| 6 | 7 | A | B | C | D | |||||
| 7 | 8 | B | C | D | ||||||
| 8 | 1 | C | D | |||||||
| 9 | 10 | D | ||||||||
| 10 | 2 | D | ||||||||
| German | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 1 | A | ||||||||
| 2 | 2 | A | ||||||||
| 3 | 3 | A | ||||||||
| 4 | 4 | A | ||||||||
| 5 | 5 | A | ||||||||
| 6 | 6 | A | ||||||||
| 7 | 7 | A | ||||||||
| 8 | 8 | A | ||||||||
| 9 | 9 | A | ||||||||
| 10 | 10 | A | ||||||||
Levels are ordered by mean reading speeds, with level 1 read fastest. Levels not connected by the same letter are significantly different.
Results
Comparison among texts
The maximal reading speed difference between texts was 5.4% in English, 6.8% in Finnish, 6.5% in French, and 2.6% in German (fig 1 and table 2). Table 2 lists the means of the reading speeds measured for each text. Capital letters connect the levels with no statistically significant difference (α = 0.05). For future use, sets with at least seven texts connected with the same letter (10 in German, eight in Finnish, and seven in French, and seven in English) can be used interchangeably within one language to measure reading speed.
Figure 1 Reading speeds (mean and 95% confidance interval) listed for all 10 texts read by young observers in English, Finnish, French and German. (The solid line indicates the mean reading speed of all four languages.)
Table 2 List of average reading speeds (SD) and range (characters/minute) for young (A) and old (B) readers for each text in the four tested languages.
| (A) 18–35 years | ||||||||
|---|---|---|---|---|---|---|---|---|
| Text | English | Finnish | ||||||
| mean (SD) | min–max | mean (SD) | min–max | |||||
| 1 | 1247 (161) | 884–1538 | 1216 (112) | 948–1425 | ||||
| 2 | 1201 (137) | 849–1397 | 1281 (148) | 975–1675 | ||||
| 3 | 1236 (126) | 923–1403 | 1286 (138) | 1009–1568 | ||||
| 4 | 1228 (133) | 905–1416 | 1252 (121) | 981–1568 | ||||
| 5 | 1246 (153) | 874–154 | 1291 (157) | 990–1669 | ||||
| 6 | 1270 (147) | 904–1499 | 1255 (129) | 1018–1518 | ||||
| 7 | 1257 (155) | 785–1516 | 1304 (166) | 1009–1764 | ||||
| 8 | 1223 (158) | 799–1447 | 1241 (136) | 988–1595 | ||||
| 9 | 1227 (159) | 862–1451 | 1250 (152) | 935–1622 | ||||
| 10 | 1207 (152) | 900–1440 | 1254 (150) | 995–1561 | ||||
| General | 1234 (147) | 785–1540 | 1263 (142) | 935–1764 | ||||
| Text | French | German | ||||||
|---|---|---|---|---|---|---|---|---|
| mean (SD) | min–max | mean (SD) | min–max | |||||
| 1 | 1188 (132) | 989–1478 | 1139 (96) | 946–1290 | ||||
| 2 | 1177 (153) | 927–1486 | 1135 (112) | 952–1406 | ||||
| 3 | 1222 (161) | 957–1603 | 1135 (135) | 826–1362 | ||||
| 4 | 1240 (141) | 1024–1568 | 1132 (107) | 947–1368 | ||||
| 5 | 1228 (143) | 949–1519 | 1131 (110) | 896–1368 | ||||
| 6 | 1222 (162) | 988–1567 | 1128 (107) | 935–1294 | ||||
| 7 | 1216 (167) | 936–1528 | 1122 (98) | 899–1261 | ||||
| 8 | 1207 (158) | 990–1499 | 1114 (109) | 877–1258 | ||||
| 9 | 1258 (157) | 1038–1584 | 1111 (95) | 931–1256 | ||||
| 10 | 1180 (155) | 951–1458 | 1110 (81) | 943–1243 | ||||
| General | 1214 (152) | 927–1603 | 1126 (105) | 826–1406 | ||||
| (B) 60–85 years | ||||||||
|---|---|---|---|---|---|---|---|---|
| Text | English | Finnish | ||||||
| mean (SD) | min–max | mean (SD) | min–max | |||||
| 1 | 967 (118) | 781–1230 | 987 (182) | 626–1388 | ||||
| 2 | 941 (87) | 783–1099 | 1023 (184) | 611–1355 | ||||
| 3 | 952 (91) | 780–1135 | 1032 (174) | 693–1328 | ||||
| 4 | 950 (97) | 808–1132 | 1010 (188) | 633–1352 | ||||
| 5 | 971 (91) | 812–1095 | 1026 (183) | 663–1349 | ||||
| 6 | 975 (90) | 790–1111 | 1018 (173) | 671–1305 | ||||
| 7 | 969 (113) | 809–1173 | 1030 (183) | 668–1369 | ||||
| 8 | 915 (89) | 774–1067 | 995 (181) | 614–1317 | ||||
| 9 | 946 (87) | 779–1117 | 998 (189) | 608–1369 | ||||
| 10 | 918 (94) | 751–1184 | 1017 (182) | 610–1353 | ||||
| General | 951 (97) | 751–1230 | 1014 (179) | 608–1388 | ||||
| Text | French | German | ||||||
|---|---|---|---|---|---|---|---|---|
| mean (SD) | min–max | mean (SD) | min–max | |||||
| 1 | 1124 (162) | 789–1413 | 956 (113) | 802–1200 | ||||
| 2 | 1084 (148) | 810–1382 | 923 (109) | 761–1188 | ||||
| 3 | 1119 (151) | 857–1417 | 931 (107) | 772–1209 | ||||
| 4 | 1169 (163) | 884–1480 | 953 (125) | 727–1248 | ||||
| 5 | 1143 (171) | 771–1409 | 925 (124) | 745–1229 | ||||
| 6 | 1148 (171) | 838–1446 | 940 (124) | 782–1216 | ||||
| 7 | 1139 (169) | 859–1378 | 950 (127) | 764–1213 | ||||
| 8 | 1107 (152) | 814–1375 | 903 (111) | 733–1141 | ||||
| 9 | 1155 (156) | 893–1416 | 935 (127) | 763–1216 | ||||
| 10 | 1119 (165) | 785–1367 | 926 (114) | 750–1182 | ||||
| General | 1131 (160) | 771–1480 | 934 (117) | 727–1248 | ||||
Comparison among languages
Mean values (SD) in characters per minute are: English 1234 (147), Finnish 1263 (142), French 1214 (152), and German 1126 (105) for young readers. There is a statistically significant difference among the reading speeds measured in the languages (α = 0.05) with a maximum difference of the mean reading speeds of 10.8% (between Finnish and German).
Comparison between young and elderly readers
The comparison of young and elderly readers shows significantly (α = 0.05) lower reading speeds in the older population. Differences between young and elderly readers were 19% in English, 25% in Finnish, 7% in French, and 19% in German (fig 2, table 2).
Figure 2 Young and old subjects̀ reading speed (char/min) in all four languages. Points are the individual values (jittered), the horizontal lines indicate the means and the vertical extremities represent the 95% confidence intervals.
Discussion
Continuous text is a much better predictor of real world reading performance than visual acuity. However, standardised texts are required to achieve reliable and valid measurements of reading speed. Mansfield et al standardised the MNREAD reading test according to the criteria of constant character count and controlled word frequency. They allowed variable word length and syntactic structure across the sentences.16
Radner et al developed a set of sentences that were highly constrained to have similar lexical difficulty, word length, and semantic structure. Reading speeds for normally sighted observers were similar across all of the sentences. They reported that the reading speeds for the sentences were highly correlated with reading for text passages consisting of about 250 words (r = 0.89).11,17
Both the MNREAD and Radner tests use single sentences in a range of letter sizes. This is useful for measuring reading acuity and critical print size, but the passages are rather short for making accurate measurements of reading speed. We have developed a set of longer text passages in four European languages (English, Finnish, French, and German) that can be used to accurately measure reading speed. The texts conformed to the criteria of equal length and comparable linguistic complexity. The homogeneity and comparability of the texts was verified in experimental testing in a group of 100 readers aged 18–35 years and 100 readers aged 60–85 years without vision disorders. We identified sets of texts that are statistically homogeneous. This resulted in a minimum of seven passages in English and French, eight Finnish, and 10 texts in German that were comparable. For many purposes, the maximal difference of 7% in mean reading speed between texts can be tolerated; in these cases all texts can be used. If an even higher accuracy is needed, the normative data of table 2 can be used to calculate corrective factors. This is especially recommended for inter‐language comparisons, since the reading speed differences among the languages are partially caused by the structural differences of the languages.
In testing we found a difference in reading speeds across subjects. However, within each subject the reading times for different texts are very homogeneous. This is a further indication that the texts require the same amount of processing capacity from a subject to read them.
Clinical studies evaluating the reading performance of low vision patients, in particular those with central scotoma and eccentric fixation, underline the need for a reading test to assess visual performance especially in the context of low vision rehabilitation.18 We recommend text reading as a tool for the evaluation of reading performance rather than single sentences.
Summary
The newly developed texts close a gap in reading diagnostics, whereas reading acuity tests, like MNREAD and the Radner test, in which single sentences are arranged in a decreasing order of size, are useful for quantifying reading acuity and magnification need. Paragraph reading texts, like the one developed here, are crucial to judge reading performance as a task of everyday life to lay the foundation of visual rehabilitation. Comparing reading speeds measured in a patient under different conditions requires several comparable text passages to assess the patient's visual performance. The reading texts we compiled are well suited to test the reading ability of patients in a realistic task. Texts in additional languages will be available soon.
These new texts are designed to be used in the field of reading and low vision research for a variety of purposes, for example:
as a diagnostic tool in vision or reading disorders
for follow up testing of reading performance
for comparison of different magnifying aids
for international comparative studies assessing reading performance.
Abbreviations
AMD - related macular degeneration
MNREAD - Minnesota Reading Test
Footnotes
Support: European Commission, Key action No 6, AMD‐READ‐Project, QLK 6‐CT‐2002‐00214.
The authors have no commercial interests.
This study was presented in part at the ARVO meeting 2004 in Fort Lauderdale: Weismann M, Hahn GA, Gehrlich C, et al. New standardised texts in four European languages for assessing reading performance. ARVO abstract 2004:4347/B808.
References
- 1.Aulhom E. Fixation width and fixation frequency of the contours presented in reading. Pflügers Arch Physiol 1953257318–328. [DOI] [PubMed] [Google Scholar]
- 2.McConkie G W, Rayner K. The span of the effective stimulus during a fixation in reading. Percept Psychophys 197517578–586. [Google Scholar]
- 3.Rayner K, Well A D, Pollatsek A. The availability of useful information to the right of fixation in reading. Percept Psychophys 198231537–550. [DOI] [PubMed] [Google Scholar]
- 4.Trauzettel‐Klosinski S. The significance of the central visual field for reading ability and the value of perimetry for its assessment. In: Wall M, Heijl A, eds. Perimetry update. Amsterdam, New York: Kugler, 1997417–426.
- 5.Trauzettel‐Klosinski S, Laubengaier C, Sadowski B.et al The significance of visual acuity and magnification need for the reading ability of low vision patients. Z Prakt Augenheilkd 200021529–533. [Google Scholar]
- 6.Trauzettel‐Klosinski S. Reading disorders due to visual field defects—a neuro‐ophthalmological view. Neuro Ophthalmol 20022779–90. [Google Scholar]
- 7.Ahn S J, Legge G E. Psychophysics of reading—XIII. Predictors of magnifier‐aided reading speed in low vision. Vis Res 1995351931–1938. [DOI] [PubMed] [Google Scholar]
- 8.Ahn S J, Legge G E, Luebker A. Printed cards for measuring low‐vision reading speed. Vis Res 1995351939–1944. [DOI] [PubMed] [Google Scholar]
- 9.Legge G E, Ross J A, Luebker A.et al Psychophysics of reading. VIII. The Minnesota Low‐Vision Reading Test. Optom Vis Sci 198966843–853. [DOI] [PubMed] [Google Scholar]
- 10.Mansfield J S, Legge G E, Bane M C. Psychophysics of reading. XV: Font effects in normal and low vision, Invest Ophthalmol Vis Sci 1996371492–1501. [PubMed] [Google Scholar]
- 11.Radner W, Willinger U, Obermayer W.et al A new reading chart for simultaneous determination of reading vision and reading speed. Klin Monatsbl Augenheilkd 1998213174–181. [DOI] [PubMed] [Google Scholar]
- 12.Foss D. Decision processes during sentence comprehension: effects of lexical item difficulty and position upon decision times. J Verbal Learning Verbal Behavior 19698457–462. [Google Scholar]
- 13.Gibson E. Linguistic complexity: locality of syntactic dependencies. Cognition 1998681–76. [DOI] [PubMed] [Google Scholar]
- 14.Gibson E. The dependency locality theory: a distance‐based theory of linguistic complexity. In: Miyashita Y, Marantz A, O'Neil W, eds. Image, Language, Brain. Cambridge, MA: MIT Press, 200095–126.
- 15.Chung S T, Mansfield J S, Legge G E. Psychophysics of reading. XVIII. The effect of print size on reading speed in normal peripheral vision. Vis Res 1998382949–2962. [DOI] [PubMed] [Google Scholar]
- 16.Mansfield J S, Ahn G, Legge A.et al A new reading acuity chart for normal and low vision. Opt Soc Am Tech Digest 19933232–235. [Google Scholar]
- 17.Radner W, Obermayer W, Richter‐Mueksch S.et al The validity and reliability of short German sentences for measuring reading speed. Graefes Arch Clin Exp Ophthalmol 2002240461–467. [DOI] [PubMed] [Google Scholar]
- 18.Cummings R W, Whittaker S G, Watson G R.et al Scanning characters and reading with a central scotoma. Am J Optom Physiol Opt 198562833–843. [DOI] [PubMed] [Google Scholar]