Acceso usuarios
Joseba Andoni Lakarra Andrinua 
In this paper, we show the direct relation between reconstruction and etymology and between linguistic theory and philology, as well as the need to overcome atomistic approaches which are based on anecdote or which are not backed up by any theoretical model of linguistic change. At the same time, we also try to extend a new etymological paradigm which runs parallel to the reconstruction model developed from the analysis of the canonical form and evolution of Basque morphemes. This paradigm sets out to cover periods and explain aspects of the language never before dealt with or even contemplated by previous paradigms (Lakarra 1995a and later work).
After a review (§1) of different definitions of the notion of «etymology» and its fundamental role in reconstruction, we provide an analysis (§2) of the etymological research centered upon Latin-Romance borrowings which was carried out by the main linguists who worked up to the 1960s on the reconstruction of the prehistory of Basque and proto-Basque: Schuchardt (§2.1), Gavel (§2.2), Martinet (§2.3) and Mitxelena (§2.4); in §2.5 we deal with the situation of etymology and reconstruction in the wake of Mitxelena�s work.
In §3 we analyze two alleged alternatives (§3.1 glottochronology and §3.2 massive comparison) to the traditional comparative method and evaluate their impact (§3.3) �unconstructive, with no tangible progress� on the studies of Basque diachronic linguistics.
Sections §4 and §5 constitute the core of our proposal: after an overview of some of the implications of the fact that Mitxelena did not use the canonical form (CF) in his etymological work, we contend that future etymological work must include internal reconstruction based on the analysis of the CF, associated with the diachronic holistic typology and the philology «of precision» set out by Meillet (§4.2). In (§4.3) we review the antecedents of the theory of the Basque root (mainly Uhlenbeck 1942) and the data that required the adoption of a reconstructive and etymological model which was more constrained and explanatorily more powerful than the previous one: namely, irregularities and gaps in morphophonological paradigms which could not be accounted for by the old model (§§4.4-4.5). In §4.6 we show that the view of the alleged Basque-Caucasian kinship as rather implausible and barely productive is clearly reinforced by the use of a fundamental and lasting feature such as the canonical form of morphemes. In §§4.7-4.8 we examine alleged exceptions to phonetic laws or to restrictions on the co-occurrence of certain phonemes in Basque words as formulated by Mitxelena. We provide evidence that it is absolutely necessary to make the best possible use of philology in both etymological analysis and reconstruction, by availing of advances in linguistic theory and, particularly, the analysis of the CF of morphemes. In §4.9, our use of the reconstruction of lexical families and features of the old grammar such as reduplication allows us to overturn the claim that items such as adar �horn, antler, feeler� are borrowings, in opposition to the atomist analysis. Since these items fit into old reduplicative paradigms, their status as borrowings is rendered rather improbable, thus leading further into the reconstruction of proto-Basque grammar by ruling out false leads. In §4.10 we show that Vennemann and other comparativists sacrifice the principled explanation of specific terms such as handi «big» or ahuntz «goat» (both analyzable within an older morphology) and of real diachronic problems of the Basque language in favor of comparisons or alleged borrowings that neither shed light on their past nor further their reconstruction. In §4.11 we analyze zemai «threat, menace», not previously identified as a borrowing despite the / m / and the diphthong in final position; this word can be explained along with abagadaune «occasion», a borrowing (Mitxelena 1964) for which no specific origin has been provided in previous work. We elaborate new criteria for the identification of borrowings such as the impossibility of coexistence of strictly identical allomorphs in the same dialectal areas and we point out the importance of metathesis in etymological work. The example of andere «lady, woman» (§4.12) is used to illustrate certain problems posed (and types of help offered) by the analysis based on the CF when deciding whether a specific term is patrimonial or a borrowing.
In § 4.13 we summarize a study on the evolution of dentals and conclude that azal �skin�, ahal �can, be able�, ohol �log�, ahan-tz �forget�, ihin-tz �dew�, etc., constitute cases of V1CV1C (old reduplications of roots with a dental in initial position). In §4.14 such a dental drop is shown to explain a great number of diphthongs which appear in verbal roots, a situation that leads us to reconstruct primitive roots as CVC �and old prefixes (*da-) previously unattested in uninflected forms� in opposition to the generalized polimorphism accepted by Lafon (1943) and other authors.
In §5 we contend that among the most relevant consequences of the monosyllabic root theory and etymology based on the CF of morphemes is the possibility �a crucial one given the exiguous Basque corpus� of developing a formal etymology and elaborating a sketch of the prehistory of the Basque patrimonial lexicon. The criteria that may help to distinguish between old and new root types in a principled manner include: (a) the relationship between attested as opposed to potencial roots in each model; (b) words with and without a known etymology as opposed to the set of possible words in each root model; (c) the presence of borrowings, compounds and derived words or phonosymbolisms present in each model; (d) the phonotactic and geographic conditions that hold or fail to hold for each model.
In §6 we summarize the main conclusions and in §7 we review the most relevant bibliography. In §8 we include two appendixes: §8.1 is an excursus about false illusions in literary language and some of their consequences in previous etymological work and §8.2 contains several tables with analyses of some root models according to several of the criteria and issues presented in §5. In work in progress (cf. Lakarra in progress-1) we provide a list of around 450 new etymologies, the result of our etymological work based on the analysis of the canonical form of morphemes and on the monosyllabic root theory.
