Cultural variability and the universality of reading mechanisms

Listen to audio

Following this description of generic reading networks, largely based on Anglo-American and French literature, the lecture turned to the cultural variability of writing systems and their impact on reading mechanisms.

According to the neural recycling hypothesis, if the architecture of the human brain has not had the opportunity to evolve for reading, writing systems could, conversely, have evolved to take account of our brain's constraints. Could we then identify certain transcultural universals of writing and link them to the organization of the reader's brain? A brief examination reveals several universal aspects of writing systems :

all require the sequential acquisition of word information by means of ocular saccades ;
all feature a high density of high-contrast (typically black-on-white) strokes in the fovea of the retina ;
all are based on a small repertoire of basic shapes whose hierarchical combinations form the characters ;
all consider that the position and absolute size of characters are indifferent (presupposition of invariance).

Two remarkable invariants have recently been added to this list:

Whatever the number of characters, the number of strokes per character is approximately constant (~ 3) (Changizi & Shimojo, 2005).
The topology of traits within characters follows a reproducible distribution across cultures, the same as that observed in natural images (Changizi, Zhang, Ye, & Shimojo, 2006).

These data are compatible with the hypothesis that, in all cultures, the forms that mankind has given itself for its characters are those that are most easily encoded at the level of the inferotemporal cortex involved in visual object recognition.

In terms of brain mechanisms, brain imaging confirms the universality of the major reading networks. Visual word recognition in Hebrew, Chinese or Japanese involves the left ventral occipitotemporal region at coordinates virtually identical to those observed in French or English. This "visual word form region" specializes for the statistics of a given script, but its location remains stable from one culture to another. However, one source of variability concerns the size of the speech units coded in writing (syllables in Japanese, phonemes in French) and the regularity of this written-oral relationship. Even within alphabetic scripts, some languages have a strictly transparent orthography, where each grapheme translates univocally into a phoneme (as in Italian), while others include numerous exceptions and irregularities (as in English and, to a lesser extent, French). Orthographic transparency affects learning to read, delaying the reading performance of English and French children by several years. It also modulates the brain's representation of reading (Paulesu et al., 2000). Within a large common network, Italian readers activate more the superior temporal area, associated with phonological representation, while English readers activate more the ventral occipitotemporal and inferior frontal areas, suggesting processing of visual units of higher granularity and greater recourse to the lexical-semantic pathway.

Other recent findings also suggest that Chinese writing places greater emphasis on a left precentral region, perhaps associated with gesture memory or "addressed phonology" (Bolger, Perfetti, & Schneider, 2005). On the whole, however, an impression of the universality of the reading circuit dominates, modulated locally by small changes in position and intensity of activation according to the specific difficulties of a given script.