A discriminative learning perspective also suggests a reconceptualization of Bloomfield’s model that resolves at least some of the “contradictions” perceived by Harris (1942:169) in ways that permit a rapprochement with a WP model. One of the reasons that Bloomfield’s model “didn’t make sense” to Descriptivists such as Hockett (1968: 20) was that it incorporated dimensions of analysis that were difficult to isolate as ‘units’ within a decompositional approach. However, these dimensions receive a particularly natural interpretation in terms of discriminative contrasts. A general view of language in terms of systems of discriminative contrasts is also compatible with a distributional operationalization of ‘meaning’ that can address the familiar objection that Bloomfield and his successors ‘excluded meaning’ from the study of language.
One of the most obscure aspects of the model set out in Bloomfield (1933) is the notion of grammatical “arrangements” of linguistic forms. This notion is explicated in the passage cited on p. 21, repeated in its entirety below:
The meaningful arrangements of forms in a language constitute its grammar. In general, there seem to be four ways of arranging linguistic forms. (1) Order is the succession in which the constituents of a complex form are spoken... (2) Modulation is the use of phonemes which do not appear in any morpheme, but only in grammatical arrangements of morphemes... (3) Phonetic modification is a change in the primary phonemes of a form... (4) Selection of forms contributes a factor of meaning because different forms in what is otherwise the same grammatical arrangement, will result in different meanings. (Bloomfield 1933:163k)
The challenge that arrangements presented to Bloomfield’s successors was that they represented abstractions over forms rather than parts of forms. Extracting individual dimensions of variation from the networks of oppositions in which they operate appeared to create “contradictions” or “make no sense”. However, from a discriminative perspective, it is the attempt to isolate these dimensions of variation and associate them with context-independent properties that is senseless, because it disrupts the system in which they function. As Bloomfield states in the passage cited on p. 22 above (and below), “features of grammatical arrangemement” or “taxemes” are individually sub-meaningful contrasts that collectively characterize the meaningful units in a system:
A simple feature of grammatical arrangement is a grammatical feature or taxeme. A taxeme is in grammar what a phoneme is in the lexicon—namely, the smallest unit of form. Like a phoneme, a taxeme, taken by itself, in the abstract, is meaningless. Just as combinations of phonemes or, less commonly, single phonemes, occur as actual lexical signals (phonetic forms), so combinations of taxemes, or, quite frequently, single taxemes, occur as conventional grammatical arrangements, tactic forms. (Bloomfield 1933:166)
Bloomfield’s successors were comfortable with the idea of sub-meaningful units of lexical form, i.e., phonemes. Yet the notion of a sub-meaningful unit of grammatical form seemed enigmatic. The central problem derives from the fact that their strategy of atomizing systems into ‘units’ of form and meaning was associative. Minimal units at each level were associated with discrete units at the other level. However, there was no recurrent ‘unit of meaning’ that could be associated with individually sub-meaningful contrasts and, hence, no way to describe the way that they collectively characterize meaningful units.
A discriminative perspective resolves this problem by cancelling the assumption that individual contrasts are individually meaningful. Instead, as Bloomfield proposes, taxemes represent the minimal discriminable contrasts that define variation within a system of forms. Combinations of taxemes discriminate forms that can be assigned functions and meanings. But the individual dimensions of variation are sub-meaningful because contrasts between meaningful forms are not mediated by associations with ‘units of meaning’.
In sum, the form classes defined by individual taxemes characterize one dimension of variation in a grammatical system. Reconceptualizing these units in discriminative terms suggests an interpretation of the Bloomfieldian treatment of form variation that is broadly compatible with a WP approach.