During the first decades of the twentieth century many epistemological movements favored turning away from the "modern" (i.e. the Enlightenment's) starting point of the privacy of one's own conscious experiences -"ideas" or "concepts" or "judgments" or "beliefs" in one's own mind- to something that could be analyzed in a more public objective manner: "statements" or "propositions" in a clear and precise, unambiguous language. This move is popularly known as "the linguistic turn."
This move was undoubtedly motivated by a feeling that classical modern epistemology had gotten trapped in the tar pit of subjectivity by its Cartesian cogito subjective starting point, an error which all of these movements sought to avoid. Furthermore, it went nicely with the development of the powerful tools of symbolic logic which demonstrated exactly how the sort of analysis of complex propositions could clarify their meanings by reducing them to logical truth functions of simple propositions; and this was indeed an impressive contribution to the advancement of philosophical technique.
Unfortunately, a language can be thought of as embodying a certain conceptual framework or scheme of concepts with which it organizes what the user of the language can say about the world. This was familiar already from Kant, but whereas Kant's concepts were (allegedly) deduced from the nature of Reason itself, when one considers different languages (as well as the languages of earlier epochs) one cannot help but to be struck by the fact that different "languages" employ different "conceptual schemes" for organizing experience in a way to make linguistic utterances about it possible. Typically the scientist reporting the observational evidence for a theory will express it in a language comprehensible only to other professionals who employ (at least roughly) the same language. This thus represents an arbitrary or conventional element in all language. Thus the linguistic turn introduces conventionalism into epistemology: the conceptual framework with which we speak about the world we observe is a product of "conventions" adopted for the sake of successful communication for certain goals or purposes. If there is then an uneliminable element of conventions in our language, and if knowledge is composed of statements (the linguistic turn); it seems there must be an uneliminable conventional element in any system of acceptable beliefs. In short there are different possible "foundations" none of which has bed-rock necessity, all of which are designed to serve certain goals or purposes. Those purposes may include the satisfaction of certain expectations, thus epistemology gets "pragmatized" from the linguistic turn.