Desktop version

Home arrow Language & Literature

From segmentation to classification

From the outset, the Post-Bloomfieldians were aware of the challenges that faced the IA model. The segmentation of words into arrangements of formatives

Word and Paradigm Morphology. First edition. James P. Blevins © James P. Blevins 2016. First published 2016 by Oxford University Press sometimes produced analyses in which there appeared tobea shortfall of meaningbearing segments, and at other times produced analyses in which there seemed to be an excess of segments. In yet other cases, analyses contained segments whose status as ‘items’ seemed highly dubious. Many of the inventive solutions that the Post-Bloomfieldians developed to meet these challenges remain with us, in the form of the ‘zero’, ‘empty’ and ‘portmanteau’ morphs that still populate morphological analyses. The use of ‘replacive’ and ‘subtractive’ morphs to describe stem alternations and truncations also survives in modern accounts, in the guise of elements that trigger ‘readjustment rules’ (Halle and Marantz 1993) or other ‘morphophonemic’ devices.

The typological biases of the IA model were also evident to at least some Post-Bloomfieldians. In later reflections on the development of the IA model, Hockett (1987:8if.) remarks that, “We seemed tobe convinced that, whatever might superficially appear to be the case, every language is ‘really’ agglutinative”. As noted with particular clarity by Lounsbury (1953), the attempt to impose an agglutinative analysis on the forms of a fusional (or ‘flectional’) language often led to a type of indeterminacy that the model could not resolve:

In a fusional language, if one seeks to arrive at constant segments... conflicts arise in the placing of the cuts. One comparison of forms suggests one placement, while another comparison suggests another. Often, in fact, no constant segment can be isolated at all which corresponds to a given constant meaning. Situations of this kind often permit of more than one solution according to different manners of selecting and grouping environments. (Lounsbury 1953:172)

Significantly, cases with “more than one solution” presented as much of a challenge as those in which “no constant segment can be isolated at all”. The reason for this was that multiple solutions tended to reflect the multiple ways in which forms were organized into patterns in a language. Imposing a fixed segmentation on these cases was not only arbitrary but destructive, since isolating one pattern would disassemble a form in ways that disrupted other patterns. Given that any segmentation would privilege some patterns while disrupting others, it followed that the problem of assigning “constant segments” could not be overcome by imposing a segmentation strategy by fiat.

The issues raised by Lounsbury (1953) were not resolved or even directly addressed in subsequent developments of the IA model, which consisted mainly of technical solutions to fundamental problems of analysis. The primacy of technical elaborations in turn reflected a general reorientation of linguistics during the Post- Bloomfieldian period, as the study of the structure (and history) of languages gave way to the study of the methods and techniques employed in the description and analysis of languages (Blevins 2013). In this intellectual setting, the appearance of Two models of linguistic description marked a first, tentative, step in the shift from the exploration of narrow technical refinements to a broader assessment of the subversive effect of admitting elements like ‘subtractive’ or ‘replacive’ morphs. After acknowledging that these ‘items’ were really processes masquerading as forms, Hockett proceeded to outline a uniformly process-based alternative. Hockett termed this model the ‘item and process’ (IP) model, recognizing an intellectual debt to the process-based perspective of Sapir (1921), which he, like other Post- Bloomfieldians, had previously regarded with considerable suspicion.

The two models that Hockett articulated are still in current use, as are the terms ‘IA’ and ‘IP’, and his 1954 paper remains a landmark of morphemic analysis. But the explicit formulation of IA and IP models had brought larger issues into sharper focus.1 By 1951, when, as Hockett notes, “most of the paper was written” (p. 386), he had to come to realize that the IA and IP models shared fundamental idealizations, and that his exclusion of the ‘word and paradigm’ (WP) model reflected the somewhat parochial conception of morphological analysis that he shared with fellow Post-Bloomfieldians. After noting that this “defect in the paper” (ibid.) had initially led him to withhold publication, Hockett offers, by way of mitigation, a plea for equal treatment:

Quite apart from minor variants of IP or IA, or models that might be invented tomorrow, there is one model which is clearly distinct from either IA or IP, and which is older and more respectable than either. This is the word and paradigm (WP) model, the traditional framework for the discussion of Latin, Greek, Sanskrit, and a good many more modern familiar languages. It will not do to shrug this frame of reference off with the comment that it is obviously insufficiently general, incapable of organizing efficiently the facts of a language like Chinese. As yet we have no completely adequate model: WP deserves the same consideration here given to IP and IA. (Hockett 1954:386)

Hockett did not attempt a general reappraisal until much later, in the context of a ‘Resonance Theory’ that, as he noted, “in surprising measure harks back to Saussure” (Hockett 1987:96). By this time, Hockett’s direct influence on the field had waned and his attempts to place the Bloomfieldian programme in a broader historical context had less impact than his earlier development of morphemic analysis. But his “apologies for not having worked [a] consideration of WP into the present paper” (Hockett 1954:386) provoked a reaction that would ultimately lead to the rehabilitation of the ancient model.

Sections 1.2 and 1.3 now trace this rehabilitation, focussing on developments within the dominant intellectual lineages that fall within the WP framework. The point of departure for this morphological tradition is the classical WP model, which Hockett describes above as “the traditional framework for the discussion of Latin, Greek, Sanskrit, and a good many more modern familiar languages”. The following summary focusses primarily on two traditions. Models that formalize the classical model in terms of interpretive rule or constraint systems are collectively designated as realizational models. Models that adopt a complex system perspective and emphasize patterns of interdependency, recently formalized in terms of information theory and discriminative learning, are grouped together as implicational models.[1] [2]

  • [1] As Hockett (1968:29) remarked later “It is interesting to note that we no sooner achieved a pureitem-and-arrangement model (not yet called that) than we began to wonder whether it was really whatwe wanted”.
  • [2] Models such as Lexeme-Morpheme Base Morphology (Beard 1995) or Whole Word Morphology(Ford et al. 1997; Singh and Starosta 2003) can be viewed as variants of realizational models. Bochner(1993) is likewise a precursor of implicational approaches. Other paradigmatic models, such as Seiler
< Prev   CONTENTS   Source   Next >

Related topics