Grammar of An Uncodified Language: The Old Occitan Grammar of Guilhem Molinier

The literary genre of the grammar has been with us since Dionysios Thrax invented it in the second century B.C.¹ For most of its history, it was identified with the Greek and Latin languages, since grammars of most modern languages did not begin to be written until the sixteenth century. There were some notable exceptions, however, and one of these was the Old Occitan poetic koiné used by the medieval troubadours, which spawned a grammatical tradition extending from 1210 to 1356 that included no fewer than five different grammars, each more complex and complete than those previous.² The culmination of the tradition was the grammar of Guilhem Molinier forming the third and final volume of the Leys d'Amors, a treatise on literary matters intended to guide would-be troubadours of the fourteenth century.³

In 1323, a group of seven educated middle-class men of Toulouse formed an organization, the Consistori del Gai Saber, to revive the courtly troubadour tradition in a bourgeois context. This organization sponsored regular poetry competitions that have persisted in some form or another up to the present day. To aid in the composition of songs in the old poetic forms and language and to provide guidance for the judges of contests, a manual was gradually assembled and completed by 1356, primarily through the efforts of Guilhem Molinier, a long-time officer of the organization. In addition to instructions on versification and poetic genres, it was thought necessary to include as well a grammar to help clarify what constituted correct usage of the old language. Though four grammars of Old Occitan had already been written by this time, all of them were for the instruction of Italians and Catalans interested in composing songs in the language. Molinier's grammar was the first in the tradition to be written for native speakers. As had been the case with the previous grammars, it was modeled on the Latin grammars of Donatus and Priscian, the twin pillars of medieval literate education.⁴

If Molinier's grammar were just a mechanical application of the Latin grammatical apparatus to Old Occitan, it would be neither very original nor very interesting. However, it has certain features that go well beyond that. On the one hand, there are attempts to grapple with new grammatical phenomena in Romance that the Latin apparatus was not equipped to handle. On the other, there is the even more serious purpose of coping with the phenomena of linguistic variability, something the Latin grammarians never had to deal with since the literary language they described had been codified for centuries. The Occitan of the troubadours had nothing like the uniformity and consistency of the literary Latin of the Imperial period. Though the success of the song tradition it conveyed had brought it into international use, there had never been a dominant dialect nor a central authority to impose any form of conformity upon either orthography or grammar. Thus there had developed a tolerance of dialectal variation, a variation always apparent due to a phonetic orthography that had never achieved any form of conventionalization. Furthermore, the oral exchange of songs between various dialect regions had led to a randomization of variant forms, producing a dialectal mélange where the provenance of dialect variants was no longer apparent. Because the aim of recording the words of songs was oral performance, a consistent orthography was never really important, especially since the pronunciation of words was unlikely to have been so different from one region to another as to impede understanding. The situation then intensified in the latter part of the tradition when Italians and Catalans started composing songs in a language for which they could find no standard form, adding a new source of variability. Ironically, it has become traditional for scholars to marvel at the uniformity of the Occitan of the troubadours, when what is really true about it is that regional dialects cannot be identified. The variability is certainly there; it is just scrambled beyond recognition, a dialect geographer's nightmare.⁵

Molinier could have done what modern grammarians...