Historical reconstruction
In a text that was still touted as the best source on historical linguistics when I went to grad school in the 1960’s, Bloomfield’s Language (1933), the chapter on historical reconstruction is 24 pp long (297–320). Neither internal reconstruction, nor grammaticalization, nor morpho-syntactic change are mentioned; nor can they be found in the book’s index. When at the end of his discussion Bloomfield criticizes the inadequacy of the comparative method, he offers neither alternatives nor complements:
“…The comparative method, then, — our only method for the reconstruction of prehistorical language, — would work accurately for absolutely uniform speech-communities and sudden, sharp cleavages. Since these presuppositions are never fully realized, the comparative method cannot claim to picture the historical process…” (1933:318)
Almost thirty years later, in Hoenigswald’s Language Change and Linguistic Reconstruction (1960), the chapter on the comparative method (henceforth CM) is 25 pp long (119–123). A 4 pp chapter is given to the reconstruction of grammar. Only within the latter does one find a little over a page (68–69) dealing, ostensibly, with internal reconstruction (henceforth IR), with the term appearing only in the section’s title. The motivation for the comparativist to stray into IR is the standard one, following the 19th Century Grand Tradition:
“…Some fairly obvious conclusions are possible from morpheme alternants [allomorphy]. Morphophonemically related morpheme alternants normally go back to sound change…; hence their further utility in the reconstruction of sound change…” (1960:68)
The actual textbook used in my first Historical Linguistics class, Lehmann’s Historical Linguistics (1962), paints a picture that appears, at first glance, considerably more even-handed: a 17 pp chapter on CM (75–91), as against 12 pp on IR (93–104). Lehmann is, further, rather generous in recognizing the value, at least in principle, of reconstructing older forms on the basis of evidence from a single language:
“…As the comparative method is being used to determine earlier syntactic patterns on the basis of evidence from related languages, so the method of internal reconstruction permits us to propose earlier syntactic patterns on the basis of one language alone. In this way the method of internal reconstruction will lead to syntactic reconstruction of languages attested only today…” (1962:102).
While more generous to IR, Lehmann nonetheless identifies correctly the old bias of the 19th Century comparativist Grand Tradition, which resorted to IR only when CM had failed; that is, recruiting IR as an auxiliary tool where paradigmatic irregularities and variation made it impossible to posit clean, invariant correspondence sets:
“…In explaining the second group of exceptions to Grimm’s law, Grassmann made use of the comparative method. His decisive evidence, however, was furnished by observation of patterns within [only] Sanskrit and Greek. Application of the comparative method to Germanic, Sanskrit and Greek data, however, would not indicate which language better reflects the situation in the parent language. Irregularities within Sanskrit and Greek [alone], clearly show that in some environments these languages had made a departure from the Indo-European distinction between aspirates and stops…” (1962:92)
In a more recent edition of his book, Lehmann (1992) is a bit more generous to IR (13 pp, 162–174), but the goals of the reconstructive enterprise remain the same as those of the 19th Century comparativists:
“…By examining the irregularities in contrast with the regular patterns, we reconstruct earlier forms…” (1992:162)
The bias toward the Grand Tradition’s goal of reconstructing proto forms persists in the most recent, most extensive, and to my mind the most thoughtful, treatment of IR, Fox’s Linguistic Reconstruction (1995). Two whole chapters are given to IR, 72 pp in all (145–216), with a fairly extensive and surprisingly current discussion of, in particular, grammaticalization. Buried somewhere in a section titled “’Laws’ of language development” is the recognition of the most salient feature of IR: its utter dependence on a universal theory of linguistic change.
In spite of its quantum leap in sophistication, Fox’s book still winds up glossing over the profound teleological, methodological and theoretical cleavage between the two reconstruction methods.
This is so because the all-consuming agenda remains the — largely phonological — reconstruction of older forms, rather than reconstruction of the intricate and profoundly gradual natural pathways via which older forms change into newer forms. In his summary of CM vs. IR, Fox thus notes:
“…It is evident, therefore, that the two methods differ in their capacity to reconstruct splits and mergers. Internal Reconstruction favors the former. The Comparative Method the latter…” (1995:213)
The main event remains, so it seems, the same phoneme-reconstruction agenda of the 19th Century Indo-Europeanists.
Fox acknowledges, though again implicitly, the more theoretical orientation of IR as against the more empiricist, inductivist bias of CM:
“…The Comparative Method can be carried out at a fairly superficial level of language structure — sounds and phonemes — while Internal Reconstruction requires a slightly ‘deeper’ analysis — morphophonemes. Hence the former tends to be seen as more concrete, and the latter as more abstract. The comparative method often relies on lexical correspondences for its data, while Internal Reconstruction may be more grammatical…” (1995:213–214)
The conclusions, however, reveal the old 19th Century agenda, whereby IR is subservient to the reconstructive goals of CM — with the profound methodological and theoretical differences between the two largely submerged:
“…In spite of these differences, the methods are, as we have seen, in many ways parallel: both involve the recognition of correspondence sets in order to establish contrasting units; both provide abstract forms to which we may give a historical interpretation. Last but not least, in both cases the methods have weaknesses which call for due discretion in interpreting the results historically…” (1995:214).
