This often debated question is more complex, both in its meaning and its possible answers, than casual consideration might suggest. True, everyone has a concrete idea of what a living (or 'modern') language is, as well as a dead one. The former is spoken by individuals in some community in the course of their daily interactions, while the latter has no one who speaks it in his daily life and exists only in writing, if that at all. Nevertheless these two conceptions sit at opposite ends of a linguistic spectrum within which a language can manifest itself in more subtle ways. Living in our modern age of globalization, we are both hindered and helped by the distance in time and space separating us from the venerable languages of past civilizations - helped in that linguistics has emerged as a scientific discipline apart from philology and in that our great hindsight sometimes gives us amazing access to the writings and culture of a language's entire tradition, yet hindered in that such access is often fragmentary and narrowly focused, and of course, we have no native speakers of the language or members of a community that used it.

Such is how it is with Sumerian. Yet the balance of factors leans toward the side of understanding when we consider the unusually large corpus of Sumerian texts surviving for us and the cultural cohesion that held Mesopotamia together for three millennia. This provides material for supported conjectures and at least a partial answer to the question heading this section.

First though the question itself. What is a dead language, and what does it mean for a language to die? Focus in recent years has put language death in a modern socio-linguistic framework, and its approach yields information that cannot usually be gleamed from the written record. We could say that a language is living when it has some community of individuals who have enough grammatical and lexical competence to speak it with each other in order to facilitate their daily interactions. The speakers need not be fluent, nor does the language actually need to be anyone's 'mother tongue'. This distinction was often implicitly glossed over in older linguistic theories, and it will be important when we consider Sumerian. Still, 'linguistic competence' needs to great enough such that individuals can handle a wide range of daily social situations involving each other. One can go on to consider on a deeper level what linguistic competence really is (e.g. whether it should be viewed as a cerebral faculty conditioned and stimulated by one's linguistic contacts, as Chomskians might argue, or as a pragmatic tool a person uses in a community effect certain things from others). Here we will simply say a person can speak any particular language with adequate competence if he can achieve his daily needs with it in a community of its speakers.

One should remember, incidentally, that language death is not necessarily the actual end of the language in toto, since languages can become fossilized into ones that are still productive but expressed only in writing, or ones that are not productive at all but simply taught and read as part of a cultural heritage. The evolution of Sumerian fits into this description well, since one could plausibly argue that much of our understanding of the Sumerian language as a cultural phenomenon comes not only from its usage by the Sumerians themselves, but also their inheritors who dutifully preserved the language in writing.

A language begins to die when that above competence among all its users starts breaking down, and it dies when the competence is irrecoverable, practically speaking. In this intermediate phase of a language the real complications in analysis begin. Competence can be lost in a number of ways, all involving one essential, practically obvious element: the influence of another language in a speaker's environment.

One way is the prestige of the language measured against others used in the same community. Every society has concepts of 'refined', 'educated', or 'proper' language. Sometimes they are associated with high social class, or formal education, with antiquity or ceremonial occasions, or perhaps a particular ethnic identity or religious tradition. Such distinctions are particularly evident when considering sociolects and dialects within one language. The usage of Greek, Latin, French in Western civilization, and the status of 'academic English' today would be examples of these ideas. Societies also recognize what could be called 'practical' languages, that is, ones that people use as a default within a multi-lingual environment to guarantee understanding and consistency in communication. The status of English today is a clear example. Finally, societies also have the concepts of 'vulgar', 'non-standard', 'corrupt', or 'localized' language, which might be associated with low social class, low education, a stigmatized ethnic group or region, with a language's deviant novelty or its antiquated impracticability.

With this scale of prestige it is not surprising to find multi-lingual or multi-dialectical individuals switching between their usage of them based on environment. High prestige languages often have a limited scope of employment, whether that is measured in terms of the number of individuals competent in them, the actual level of that competency, or the circumstances of the languages' usage. Their scope tends to shrink with certain demographic and cultural shifts as well as pressure from other languages, and with the shrinking comes lowered competence. If the language has a written tradition it becomes more confined to text. Practical languages of course enjoy wide scope of employment and high competency and tend to influence the usage of other languages in the same speaker, most commonly through lexical transfer, but also syntactic analogy. If influential enough they can hinder the competence in other languages because of their reduced usage. Low prestige languages may have widespread or restricted employment. (Say more here?)

Demographic and cultural shifts can also affect competence in a language. At the extreme end there is the scattering of a community that isolates speakers of the language from one another, because the children of such individuals growing up in the new environment tend to have less competence than their parents. Less extreme but more prevalent are the steady immigration of individual families with children into a new linguistic environment or the passing away of the last 'pure' generations that grew up learning the old language without competition from new ones. Wide-spread changes in school curricula can effect similar changes. This is especially true for high prestige languages, the nature of whose learning and maintenance are often dependent on formalized instruction.

With such manners of change in mind one can cluster the degrees of competence in a spoken language around three points. The first, naturally, consists of fluent speakers, which would include advanced but not perfect learners of the language. Secondly there are what are often termed 'heritage speakers', that is, people who grew up in a household with fluent parents or siblings or had early, abortive instruction in the language from the community, but are otherwise largely isolated from it. Such people may have moderately developed speaking and especially listening skills but lack writing or reading skills, and their grammar can be significantly influenced by other dominant languages in the environment (Chris Woods calls these people 'semi-learners'?). Finally, there are people growing up under parents (often semi-learners) who have a weak enough competence themselves, or who use the language so rarely, that they have only a disjointed, limited vocabulary. (For them the language has become a pidgin?).

It is common for competence to downgrade along generational lines, i.e. the type of linguistic environment a child grows in. However for adult fluent speakers competency is not drastically affected (E Cook apud Michalowsky p.178). That is, fluent speakers do not lose competency when they switch to another language for the rest of their lives, or even indicate strong influence of the new language on the old. It is the creation of communities of 'semi-learners', who have little access to the language when growing up, that is the first significant blow to the language's vitality as a vernacular. Such need not happen in the case of written languages, as the special contexts of their usage imply the need for substantial formal training in them, so competency is an issue of education,and not early linguistic environments.

Finally, it should be noted that significant language interference occurs only in the context of intense cultural contact. This socio-lingual assertion is in contrast to the view that the structural characteristics of the languages themselves plays a key role. Rather if anything, significant structural influence can occur in an environment of intense cultural contact, since only there can the average speaker be heavily exposed to competing linguistic influences in his developmental stage.

As Piotr Michalowsky points out in a recently published essay (Michalowsky source), in studying the life of Sumerian it is important to remember that we carry certain biases and implicit assumptions about how languages interact and die out. This does, after all, affect the way we understand the history of Sumerian, in whose case such self-criticism is particularly important because without it the life - and death - of Sumerian will remain for us a puzzling issue.

The first is that linguistic identity equals ethnic identity, and that the understanding of ethnicity within ancient Mesopotamia was roughly the same as ours. Both modern and ancient counterexamples to the first statement abound.

The textual record indicates that Sumerian was the main vernacular in southern Mesopotamia until at least the middle of the 3rd millennium. We say main and not exclusive vernacular because by the ED IIIa period we find in texts from Abu Salabikh and Fara Semitic scribal names scattered among Sumerian ones. There are also numerous Semitic loan words attested, and even one whole literary document (a hymn to Shamash) in Semitic (Krebernik). The greater presence of Semitic influence in Abu Salabikh than in Fara indicates a gradation of language presence, with Sumerian predominating in the south and Semitic in the north - even if Semitic personal names are still attested much further south in ED IIIa (reference?).

When talking about the death of Sumerian, one should technically step back first and ask whether there were, in fact, multiple Sumerian dialects coexisting in southern Mesopotamia in the mid 3rd millennium. The 'EME.SAL' tongue has been suggested (Alster 1982, Krecher 1967a) although its restricted usage has lead most to believe it was only a form of Sumerian used in certain cultic settings. Other traces of dialectical differences can be found in some Old Sumerian Lagash texts (Enanatum inscription, cf. Bauer??). By and large, however, the written documentation has obscured variations in dialect and, with the caveat about EME.SAL, consider only one 'standard' Sumerian.

A major point of debate concerning the death of Sumerian is whether it was in rapid decline in the late 3rd millennium. The standard viewpoint, championed by scholars including Michalowski, Gelb, and Cooper, argues that influence from Semitic speakers (notable even in ED III) came to a climax in the Sargonic period, after which the 'Sumerian renaissance' of the Ur III kings was only a conscious political, administrative, and literary matter that did not involve the spoken vernacular (Michalowski, pg. 167). (The other accepts the prima facie evidence of Sumerian usage in Ur III and sees the final decline during the transition to the Old Babylonian period, with the political turmoil and demographic shifts that ensued in the south.)

This viewpoint rests on two arguments. One is the effective influence of the Sargonic conquests linguistically in the south. The other is the unconvincing correlation between personal names and language used in writing on the one hand, and language as spoken vernacular on the other.

The first argument concerns the degree to which the Sargonic kings held forceful sway over major aspects of life in the Sumerian city-states. Implicit in this, moreover, is a general theory about language contact and change. It is clear that the south always held a hostile attitude toward its northern, Semitic-speaking conquerors, as multiple rebellions broke out during the history of the empire. The first was under Rimush, the son and first successor to Sargon, right upon his succession to the throne. A coalition of cities including Ur, Adab, Lagash, and Umma led the revolt which, according to Rimush's royal inscriptions, left tens of thousands of southerners dead, enslaved, or deported (Westenholtz, pg. 41). Even that was not the end of it for Rimush, as the northern city of Kazallu rebelled later and met a similarly disastrous defeat (Westenholtz, pg. 42). Subsequently there was a larger revolt against Naramsin that included all of the southern cities as well as some northern ones like Sippar beyond the old Sumerian core. However this effort also ended in great slaughter for the rebellious cities, which did not see another major insurrection until the emergence of the Ur III dynasty.

Nevertheless, it was probably not the total number of dead southerners that mattered so much for the longevity of Sumerian as Akkadian administrative control and the policy of awarding property to the dynasty's loyal followers. The novelty of a Semitic domination of the south brought with it a novel communication structure. Correspondence with Agade was now in Akkadian. Sargon himself left most of the Sumerian city-states' bureaucracy intact, leaving local affairs up to the ensis (Westenholtz pg. 50). However after the revolts against Rimush the Akkadian kings instituted a new, mid-level bureaucracy to levy taxes and manpower (ibid.). As testament to this there are numerous administrative centers in the south within and without the cities. They have mostly Akkadian language documents in contrast to the older Sumerian administrative repositories, which have only Sumerian (Westenholz pg. 50). In addition, the Akkadian kings would grant to their followers tracts of land in the south gained through the defeat of the cities, especially properties whose previous owners died in the rebellion (Westenholtz pg. 50).

It is not clear how all this affected the language on the street. The Sargonic kings down to Sharkalisharri ruled for about 140 years (Westenholtz pg. 17), with control over the south weakening toward the end.

even if it did not operate alone at that time. First there is the dominating presence of Sumerian against Akkadian in the ED III period. the predominance of (unilingual?) Sumerian in most of documents (across all genres?). The main competitors of Sumerian would have been the Semitic languages, primarily Akkadian. The speakers of the two languages had already been in contact for a long time by the mid-3rd millennium, as attested by the Semitic names (and even one Akkadian literary text?) in administrative documents from Fara and Abu Salabikh. (Caution about alloglottography may be an issue that is, reading a text in a language different from that the script was originally designed for). The second is the attestation of Sumerian personal names in documents.

It is agreed upon that the mid-second millennium Sumerian ceased to be a vernacular, that is, a language one would use to conduct basic, mundane interactions within a community, and that the predominating languages that took its place were Semitic. How this happened, however, is under debate.

The first major signs of disruption in the use of Sumerian in the south occurred in the Old Akkadian period, when for the first time the whole of the Sumerian south was subjected to a northern empire whose rulers imposed the usage of Akkadian in government correspondence and elevated the Akkadian to the status of a prestige language as it was the tongue of the ruling elite. This change in status likely increased in the reign of Naram-Sin, who tried to quell southern recalcitrance to his rule by placing loyal, Akkadian speaking officials in charge of the cities of the south.

There are possible vestiges of the influence of Akkadian on Sumerian through this period, in the fact that in the ED period, Akkadian words loaned into Sumerian drop their inflectional ending and take on a final -a (e.g. damgara) which may be the Sumerian nominalizing morpheme, whereas in the Ur III period, we find loans with their inflectional endings (e.g. harranum). This may suggest increased bilingualism among speakers of the south.

The status of Sumerian in the Ur III period is harder to judge. Ur III is typically regarded as the time of a Neo-Sumerian renaissance, in which Sumerian was the language of state and literature. This is amply attested by the thousands of administrative and economic documents in Sumerian, as well as the collection of royal praise hymns and other literature that connect the ruling dynasty with the glories of the Sumerian past. One can also point to a high percentage of Sumerian personal names in the cities of Umma and Lagash, and especially Nippur, some of which demonstrate linguistic innovations that might indicate Sumerian was still a living language then. For instance, there is an Ur III name ga2-ke4-esz2-he2-til3 'may he live for my sake' with adverbial postposition 'akesz', which can be compared with the Pre-Sargonic version 'ga2-ka-nam-he2-til3', with the adverbial postposition 'akanam'.

One can argue that the language of the royal court may have been Akkadian, by citing the royal hymns Shulgi B and C in which Shulgi speaks of his linguistic accomplishments, boasting of his ability to speak, among other things, Amorite, Sumerian, Elamite, and the language of Meluhha. However he does not mention Akkadian, which one would expect if his native tongue were not Akkadian. There is also the recorded fact that a princess of Mari married off to the Ur III throne adopted an Akkadian name 'Taraam-Uram', thus indicating that Akkadian was the prestige language in Ur.

It is likely therefore that Sumerian continued as a living language in some of the southern cities, in particular Umma, Lagash, and Nippur, the cultural capital of Sumerian, while Akkadian dominated further north and in the rural areas. This does not mean a strict dichotomy of isoglosses, as bilingualism likely prevailed in many areas, at least at the level of commerce.

The political disruptions at the end of the Ur III empire likely dealt a severe blow to Sumerian as a vernacular (if it not already happened before) due to the decline of urban life in the southern cities like Ur, Umma, and Lagash. The agricultural and economic decline of this region saw the emmigration from these ancient Sumerian cities into other areas, where other languages dominated.

However this does not mean that Sumerian did not continue on into the Isin-Larsa and OB period as a language of the learned and of rituals. Sumerian was the language temple hymns and religious rituals. Moreover, it was the language used in the e2-dubba, or scribal school. This institution is attested in the OB, when a standard canon and teaching curriculum for Sumerian was being formed. It is likely that Sumerian was used inside the e2-dubba both orally and in writing much the way Latin was in medieval schools. In fact a group of OB texts known as the Edubba dialogues speak of students being drilled and instructed orally, as well as a distinction between the ability to speak Sumerian in the school versus merely writing it.

After this, Sumerian makes a gradual decline even as an intellectual language. From the mid second millennium on Sumerian literary texts are often given in bilingual Sumerian-Akkadian versions, and the school canon is reduced. We should keep in mind that by now Akkadian itself has ceased to be the language of the masses, Aramaic or Greek or other languages having taken its place. Thus Sumerian was two steps removed from the vernacular. It continued to be copied into the first millennium AD.