For centuries, the spoken word in Arabic reigned supreme; scholars, poets, professional reciters, tribal historians and genealogists depended on their prodigious memories for declamation. As the Arabic script developed and written texts steadily came into prominence, a mistrust of the written word appeared. As Beatrice Gruendler shows in her superb history of the creation of the Arabic book in the ninth century, the old ways long prevailed: the notion of ‘book learning’ would have been incomprehensible – or contemptible – to traditional scholars. Even the Koran was known and recited from memory until late in the seventh century. There were good reasons for this. A written text in Arabic is subject to inherent ambiguities due to the very nature of the script. That script is consonantal; the short vowels have no place in the Arabic alphabet (though the two long ‘semi-vowels’, W and Y, do figure there). Diacritical marks for the three short vowels would be devised eventually, but texts written in Classical Arabic, whether in manuscript or in print, do not usually include them, even today. Of course, certain texts, such as the Koran and the Hadith, are always printed fully vocalised to avoid misreadings. To make matters more problematic, though, many early manuscripts omit the dots that distinguish certain letters from one another; the letters for B, T, Th, F and Q are identical without those distinguishing dots.

This tension between the written and the spoken is reflected in the pedagogic methods used today in the teaching of Classical Arabic. When I was a graduate student, struggling to learn the classical language, and later, as a teacher of it myself, I became accustomed to a classroom practice that