So it seems that Sumerian, like many other languages all over the world, had a special dialect for women’s speech. What marks out Sumerian is that this had gained a special, explicit, status, recorded in the grammar books: this could be taken as further evidence of the high status of women in Sumerian literature.

      Returning to the question of Sumerian–Akkadian bilingualism, specialists agree that the balance of language spoken in Sumer shifted over the period 2400–1600 BC from total Sumerian to total Akkadian. Sumer began this period as a collection of independent city-states, suffered Akkadian domination in the twenty-third century, Amorite and (briefly) Elamite domination in the nineteenth, and the Babylonian rule of Hammurabi in the eighteenth. It ended with a restored independence, or rather anarchy, after the breakdown of this first Babylonian empire, but the language on the streets and in the homes was now Akkadian.

      It was an interesting example of unstable bilingualism, since in many ways the situation is reminiscent of the relation between Greek and Latin in the Roman empire, one dominating cultural and the other political life. In that case, despite the political instability, and the generally shifty reputation of the Greeks, contrasting with the towering political prestige and steadiness of the Romans, Greek nowhere lost ground to Latin. Yet here in Mesopotamia, where the various Semitic peoples, for all their political dominance, were sources of disruption, where there was apparently no major movement of the Sumerian population, and where Sumerian culture’s prestige was unchallenged, Sumerian steadily lost ground.

      In some cases even Semitic rulers attempted to fight a rearguard action on behalf of Sumerian culture. In the kingdom of Isin, which held the three most important Sumerian cities of Nippur, Uruk and Eridu in the twentieth to nineteenth centuries BC, the ruling dynasty stemmed from Mari, in the Akkadian-speaking north of Mesopotamia; yet its king termed himself ‘King of Ur, King of Sumer and Akkad’, all its official inscriptions were in Sumerian, and there was flourishing production of new editions of the classics of Sumerian literature.

      One factor working against the survival of Sumerian as a living language alongside Akkadian may have been the fact that the influential newcomers already spoke a Semitic language, and so found it easier just to get by in Akkadian. Only the Akkadians had lived in close proximity with Sumerian from time immemorial, and perhaps become bilingual. Others would be more impatient of the cultural complications they found down south. It is easy to imagine the average Amorite on the move saying: ‘After all, they all speak Akkadian, don’t they?’ The whole Fertile Crescent was familiar with some Semitic language or other, and by their nature they were all very similar, and to some extent mutually comprehensible. For all their cultural prestige (which clearly never diminished), the Sumerians found themselves having to compromise on language in their daily and business lives.

      In a sense, though, Akkadian had already taken up the burden that the speakers of Sumerian were laying down. It remained unthinkable, as it always had been, to learn to write Akkadian in any way but as an extension of Sumerian, and this despite the fact that Sumerian and Akkadian were poles apart as languages, with all their basic vocabulary totally unrelated, and quite different sound systems. The system never provided the means to distinguish consistently between b and p, among d, t and , or among g, k and q in Akkadian. Akkadian appears to be rather lacking in many of the phonetic subtleties that are characteristic of many of its Semitic sisters, having only one h sound where they may have up to three, three s sounds where they have up to four. It is difficult to tell whether it is just the poverty of Sumerian spelling which causes this appearance.

      The only innovations that Akkadian scribes appear to have permitted themselves were a new sign for the glottal stop, ’, and considerable licence with the Sumerian word symbols or logograms: they had always been available as punning devices, able to symbolise the same sound as the word they signified in Sumerian; now they could do the same trick for Akkadian as well. So, for example, the Sumerian sign , meaning ‘hand’, could now be read as idu, ‘hand’, in Akkadian; it could also represent the syllables id, it, i, ed, et and e.

      Sumerian word symbols, and Sumerian literature, remained the basis of written Akkadian, even as the language swept all round the Fertile Crescent, and well beyond the domains of the Semitic-speaking peoples, as a lingua franca for international communication. The same educational system, based on the edubba ‘tablet-house’ schools, was maintained for at least two millennia, since sign lists to teach the symbols, in the same order, have been found in the Sumerian city of Uruk dating from the third millennium and Ashurbanipal’s library in the Assyrian capital Nineveh from the mid-seventh century BC. Mastery of the classics of Sumerian literature, a canon of texts which was not extended after the mid-second millennium, remained the pinnacle of scholarly achievement, and the focus of later years spent at school. Even in mathematics, most of the terminology was in Sumerian, though the textbooks were written in Akkadian. It appears that Sumerian went on being spoken in the classroom: this has made the remaining exercises and textbooks less explicit on pronunciation than we should have liked.

      It is the Akkadian culture’s enthusiasm for all