Attention and Syntax Differentially Impact Word and Number-Reading

 

Are more errors made when reading longer words and multidigit numbers?

How do attention allocation and syntactic processing affect word and number reading differently?

In the previous post, we reviewed a study involving two adults with letter position dyslexia in word reading and an impairment in digit-position processing when reading multidigit numbers. These difficulties caused letter transpositions in words and digit transpositions in numbers.

Do longer words and multidigit numbers produce more transposition errors? Professor Naama Friedmann, Dr. Dror Dotan and Einav Rahamim examined this question with the same two adults. They compared the reading of four-, five-, and six-letter words with the reading of three-, four-, five-, and six-digit numbers.

They found that as the number of digits increased, the percentage of digit-transposition errors also increased. This was not the case for words. The percentage of letter-transposition errors did not depend on word length.

Whereas digit transpositions in multidigit-number reading tended to occur toward the end of the number, between the tens and units digits, (e.g., reading 36854 as 36845), letter transpositions in word reading tended to occur in the middle of the word.

What causes these differences between errors in word reading and errors in number reading? One possibility is that attention is allocated differently.

When we read words, attention is first directed to the letters at the edges of the word, the first and last letters. Only afterward is attention directed to the letters in the middle. One possible reason is that the outer letters activate entries in the orthographic lexicon that are candidates for the word being read. Attention is directed to the middle letters only later and is allocated to all of them together. These letters are therefore at greater risk of being transposed. Because attention is allocated to the middle letters simultaneously, word length does not affect the percentage of transposition errors.

Unlike in word reading, when multidigit numbers are read, attention is directed sequentially, one digit after another, from left to right. The digits on the left receive attention first, and attention binds them to their relative positions, so fewer position errors occur among them. The digits on the right receive attention last and receive the least attention. They may therefore be more vulnerable to transposition errors.

Because attention is allocated to the digits sequentially, the longer the number, the more attentional resources are depleted during reading, and the greater the probability of a transposition error.

Another possible explanation for the tendency to transpose the last two digits of a multidigit number is that although attention is allocated to the digits sequentially, it may be allocated to the final two digits simultaneously. If so, the reason may be syntactic.

The conversion of digits into number words is different when the last two digits form a teen number, as in 35416. This number is, of course, read as “thirty-five thousand, four hundred and sixteen.” It is possible that a teen number is produced as a single lexical unit, “sixteen.” If so, this may encourage simultaneous processing of the last two digits. Simultaneous processing and simultaneous allocation of attention to two digits may cause them to be transposed. The problem with this explanation is that it applies specifically to cases in which the final two digits form a teen number, between 11 and 19.

In my view, longer multidigit numbers produce more transposition errors because syntactic load increases with number length. When we read a number such as 35416, we must analyze its syntactic structure and assign each digit its appropriate value. This analysis requires increasing attentional resources as the number becomes longer.

This may explain the surprising finding that one of the two participants made many transposition errors when reading four- and five-digit numbers composed of different digits. However, when she was asked to read the same numbers in a segmented form, she performed the task almost without errors.

For example, she was asked to read 3456 as “thirty-four, fifty-six,” and 34567 as “thirty-four, five hundred and sixty-seven.” The only difference between “thirty-four thousand, five hundred and sixty-seven” and “thirty-four, five hundred and sixty-seven” is the word “thousand.” Nevertheless, reading “thirty-four thousand, five hundred and sixty-seven” is more difficult because it requires syntactic analysis of a more complex number.

It is therefore possible that even before attention is directed sequentially to each digit from left to right, attention is directed to the number as a whole so that an initial syntactic analysis can be performed.

The researchers suggest that difficulties in reading long multidigit numbers may be addressed by dividing the number into smaller units and reading its digits in pairs or groups of three. The problem with this strategy is that the child may then miss the syntactic and quantitative meaning of the number.

Friedmann, N., Dotan, D., & Rahamim, E. (2010). Is the visual analyzer orthographic-specific? Reading words and numbers in letter position dyslexia. Cortex, 46(8), 982–1004.