As I have previously posted, the language of religion is often fundamentally different in nature to that of everyday language. Whereas everyday language encourages explicit and unambiguous expression (in fully-propositional forms), religious language is far more regularly composed using semi-propositional forms, which permit a greater multi-valency of meaning at the expense of specificity and exactness. I also suggested that this may partly explain the reason for the persistence of archaic language within religious traditions, as the incomprehensibility of the ancient language lends itself well to the fundamental ineffability of religious claims. There is, however, a second reason for the preservation of archaic language in religious expression, and it has to do with authority.
Every literate religious tradition is fundamentally conservative with respect to its attitude to scripture. While it is possible for attitudes towards such literature to change over time, and even for certain texts to fall into disuse, as a general rule the amount of new, authoritative textual material added to religious traditions dwindles to nothing given enough time. As such, the scriptures themselves slowly come to be treated with a veneration that transcends what they merely say. They are "fetishised", for want of a better word, and the style of language they employ becomes frozen in the traditions and rituals of the religions themselves, even if this same language falls into general disuse. These texts come to form the foundation of subsequent theology (this is especially true within the Abrahamic traditions) and therefore constitute a kind of theological constitution; a largely insuperable doctrine that all subsequent developments in the tradition must not find themselves to be in conflict with.1
This progressive sacralisation of the texts presents its difficulties, however. For instance, why is it specifically these texts that we should treat as normative to the exclusion of all others? By what means can we assume these texts to be "true" (in a religious context) in a sense that other, later texts are not? For the latter question, we (or rather the believer) must rely on the legitimacy of revelation and prophecy, such that one must accept the premise that certain texts are just simply derived from the mouths of gods, or else some other similarly supernatural origin. Certain "prophets" (or oracles, or soothsayers, or witchdoctors...) are imagined to be particularly inspired in this regard and it is their proclamations that come to be regarded as especially authoritative over and against the proclamations of others (if this processes seems a little arbitrary, that may well be because it is). In literate religions, the effective preservation of ancient prophecies lend them an inherent and timeless authority over a long-enough period, one that slowly excludes the legitimacy of newer prophetic claimants. As such, to answer the former question, it is the ancient texts, produced by men whose prophetic legitimacy has grown to be unquestioned, that comes to form the normative textual basis of all literate religions.2
This is possible within religious traditions because - as per my previous post - there is little in these "inspired" texts that can be treated as unambiguously true or false. That is the nature of most religious language. As such, the tests of the legitimacy of religious claims in literate religions often depend upon their longevity. The early Christians, for instance, are suspected to have preserved the Septuagint within their nascent canon as a means of establishing their link to much older traditions, in defence of the claim raised against them that they represented a new (and therefore perverse and dangerous) religious cult. The identification of ancient pedigree was apparently an important factor for the Romans in determining the legitimacy of the many religions that were practised within their empire and I'm sure they were not alone in this philosophy. If "ancientness" is truly an important prerequisite for the perceived legitimacy of a religious practice (in the absence of the possibility of any other kind of truth claim), then it is not difficult to see why the ancient languages that originally gave rise to these traditions are preserved conterminously.
(1) - In most cases the flexibility or indeterminacy of religious language makes this process quite easy. "Heaven", for instance, began as the literal kingdom of YHWH in Jewish thought, and progressed in later Christian thought to a destination for departed souls. Now, in modern Christian theology, we might expect to see it defined far more abstractly as a state of being in which one finds oneself in the presence of God. Although the "original" meaning of heaven has been lost, the inherent ambiguity of religious language allows the modern Christian reader to derive any number of equally salient (and valid) interpretations from a Biblical expression like "the Kingdom of Heaven dwells within you" that would possess little intersect with the interpretations of ancient Jewish readers.
(2) - Even if the texts were not actually written by these "prophets" originally, their names are frequently attached to them post hoc. So Moses' name was attached to the Pentateuch, the apostle John's name was attached to the otherwise anonymous gospel and so on. A similar trend can be seen in the formation of the ahadith in Islam and the sutras in Buddhism, both of which can only be assumed to be traceable back to the "original" source if one treats the texts with extreme credulity. Sometimes outright fraud is evident in this quest for authority, such as the early Christians who wrote in Paul's name (i.e. the Pastoral Epistles) or the intertestamental Jews who wrote in the names of Enoch, Tobit and so on.