Using Rotated Text to Investigate Cross Cultural Differences in Reading

Lead Research Organisation: University of Reading
Department Name: Sch of Psychology and Clinical Lang Sci


Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.


10 25 50
Description When children learn to read, there are many clues they can use to identify words - they can sound out the letters (graphemes) of the words and then blend these sounds (phonemes) to form the word (grapheme-to-phoneme correspondence) or they can use a 'look-and-say' method to remember the visual form of the whole word. The size of the unit (number of letters) used to read varies in these two approaches, with small units being used in grapheme-to-phoneme correspondence and larger units being used for the look-and-say method. In English, children have to use look-and-say at least some of the time since many words in English do not follow conventional letter to sound correspondences (e.g. cough, through, bough) and so these have to be read as whole words. English is therefore described as having an irregular spelling system and requires a large grain strategy. By comparison, in languages like German and Spanish, every letter has one and only one sound, and thus all words can be sounded out using grapheme to phoneme correspondences. These languages therefore have regular spelling systems and requires a small grain strategy.

The idea that children learning to read in a regular language read on a letter by letter basis while children who read in irregular languages read more at the whole word level (look-and-say) is referred to as Grain Size Theory. This theory accounts well for possible differences in strategies for learning to read between European languages, but is less useful when considering learning to read in Japanese where there are two spelling systems - Kana and Kanji. Kana is a regular spelling system in which each letter (or character) has a single sound (a syllable) and so children learning to read in Kana should be able to use a grapheme-to-syllable, small unit strategy similar to that used in regular European languages like Spanish. However, in the other spelling system (Kanji), one or two characters are used to form a whole word. Here, there is no direct mapping between letter (character) and sound so it is very irregular. In contrast to English, words are very short, and so they can still be read on a character by character basis. Thus, in Kanji, there is a conflict between the regularity of the language (irregular) and the size of the unit used when learning to read (small).

A second theory of reading development (the hypothesis of granularity and transparency) was created to account for differences in the spelling systems used in languages like Japanese. This considers both the regularity of the spelling and, separately, the size of the unit of sound that each letter/character represents. In this theory, it is easier to read when spelling is regular, or when the unit of sound represented by a character is large (syllable rather than phoneme).

We wanted to test which kind of strategy would be used by children learning to read in Japanese, and to compare this to the strategy used by children learning to read in English. To do this, we used a method that can disrupt the overall appearance of a word (large unit appearance) without disrupting processing of individual letters to the same degree (small unit appearance). To do this, we randomly rotated individual letters in the word by different amounts to both left and right of centre. This makes it more difficult to recognise the word since the overall form changes, but each letter can still be read.

In this study, we used rotated text to investigate orthographic processing in English, Kana and Kanji. We used these languages as it enabled us to make different predictions about how rotation would affect text based on two the theories of orthographic processing - the grain size theory, and the hypothesis of granularity and transparency. Our predictions are outlined below.

Grain Size Theory

In this theory, if children are using a small grain strategy, rotation should have less effect on their reading accuracy and speed than if children are using a large grain strategy. Small grain strategies are thought to be used for languages with regular spelling while large grain strategies are used for languages with irregular spelling. We predicted that if this theory was correct therefore, then the regularity of each language would predict the degree to which each was affected by rotation. Since Kanji is most irregular, and Kana is most regular, we predicted that rotation would result in slower and less accurate reading for Kanji, with Kana least affected (Kana>>English>>Kanji).

Hypothesis of Granularity and Transparency: Transparency

Transparency (or regularity) of the language in this theory is similar to grain size, and so if differences in transparency affect the strategy that children use to learn to read, we would expect the same order of disruption by rotation as in the Grain Size Theory (Kana>>English>>Kanji).

Hypothesis of Granularity and Transparency: Granularity

Here the size of the unit of sound that is represented by single characters is thought to be important, with languages in which characters represent a larger sound unit (syllable) being easier to read than languages in which characters represent small sound units (phonemes). Thus, since each Kanji character codes a syllable, it should be less affected by rotation, whereas English where multiple letters are sometimes required to represent a single phoneme (e.g. -ough), thus Kanji words would be identified most accurately and fastest, while English words would be identified least accurately and slowest when rotated (Kanji >Kana >English).

Hypothesis of Granularity and Transparency: Both

If both the regularity of the spelling system and the size of the sound unit represented by a single character are protective against disruption of words by rotation, then English should be much more affected that either of the Japanese spelling systems since it has both irregular spelling and a small grain structure. By contrast, Kanji should be less affected because of its large grain structure and Kana should be less affected because of its regular spelling. Thus, the prediction for this theory is that English will be read less accurately and slower than either of the Japanese spelling systems when rotated (Kanji>= Kana >>English).


Adults and children were shown individual words and non-words (strings of characters that could be pronounced but had no meaning) and were asked to say whether each character string was a word or a non-word by pressing different keys on a computer keyboard. The computer measured the accuracy of the response and the time taken to respond (speed of reading).


Our findings, however, did not match any of these predictions. Instead we found the order of the effect of rotation on the accuracy of identifying real words amongst non-words was that Kanji was read most accurately and that English was read least accurately Kanji > English > Kana for adults and English > Kanji > Kana for children. When we looked at the effects on the speed of identifying words, we found that the order of performance was the same for adults (Kanji > English > Kana) but that for children English was read fastest and Kana slowest (English > Kanji > Kana).

While the hypothesis of granularity had predicted that reading in Kanji would be the least affected by rotation (and therefore more accurate and faster than English or Kana) and so gained some support, none of our predictions suggested that Kana would be the least accurate or the slowest when rotated. We therefore had to create a new theory to explain this. One possible explanation is that the experience of reading in one spelling system (Kanji) has an effect on reading in another (Kana). While not all spelling systems can be read at small grain size (due to irregularities in spelling), they can all be read using a large grain strategy. Additionally, experienced readers, even in very regular languages do not sound out the letters of familiar words, but identify these as whole words. There is therefore some advantage to learning to read using a large grain strategy since this is the strategy used by experienced readers. It is possible that children learn this more quickly if their spelling systems require a large grain strategy (English and Kanji) than if the spelling is completely regular and so can be learnt using a small grain strategy (Kana and European Languages). In addition to this, in Japanese, experience of reading in Kanji might increasingly promote a large grain strategy for both spelling systems. This prediction could be tested by looking at the changes in grain size strategy in children of different ages in different reading systems with different exposure to regular and irregular spelling systems (e.g. English, Japanese and European languages).
Exploitation Route The findings of this study will be of interest to researchers in the field of reading development, educationalists and parents of children with reading difficulties. By determining the typical development of sensitivity to groups of letters in words we will be able to inform both teaching of reading cross-culturally, and interventions for children who struggle to learn to read. This has implications for the UK economy since it is estimated that failure to learn to costs the economy ~ £1 billion per year. The findings from this research extend our current knowledge about learning to read in languages with different spelling systems, and suggest that new theories are required to account for the results. We are currently preparing 3 papers for publication:

1. A cross-cultural investigation of the effects of granularity and transparency on Lexical Decision in adults reading Kana, Kanji and English.

2. A cross-cultural investigation of the effects of granularity and transparency on Lexical Decision in children reading Kana, Kanji and English.

3. Developmental changes in the effects of granularity and transparency on Lexical Decision in children reading Kana, Kanji and English

We are continuing our work on this topic by looking at the relation between reading with rotated text and visual attention span in an MSc project supervised by the PI. This will allow us to determine whether the number of letters that can be identified simultaneously is related to our ability to read words in unrotated and rotated conditions in English adults. We would like to extend this work further by addressing this in cross-cultural research within different spelling systems and in both adults and children. This will be the topic of a grant application.
Sectors Communities and Social Services/Policy

Description Comparison of statistical learning in spoken and written language in typically and atypically developing readers 
Organisation University of Nottingham
Department School of Psychology Nottingham
Country United Kingdom 
Sector Academic/University 
PI Contribution Grant application to ESRC (not funded)
Start Year 2011
Description Reading twisted text : an exploration of the relationship between visual attention span and orthographic processing 
Organisation University of Winchester
Country United Kingdom 
Sector Academic/University 
PI Contribution MSc project at the University of Reading co-supervised with Dr Rachel Pye (PDRA on grant), Winchester University
Start Year 2012
Description The effects of letter rotation on lexical decision in English and Japanese adults 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Primary Audience
Results and Impact Presented at 2nd UK Orthography Group (UKOG) workshop
Year(s) Of Engagement Activity