Abstract EN:

Ch. I sets the epistemological (rationalism vs empirism) background. Ch. II zeroes down on "constituent grammars" as a formal tool for investigation. Ch. III introduces a looser class of trees, the "arboriferences" (with optional crossing of branches, etc. ). A proper sub-class (the "canonical trees") is shown to be definable in 4 equivalent ways (by conditions on rules, etc. ), and 4-way isomorphic with, yet distinct from a class of "equivalence classes of derivations". A consideration of the open sets of distinct though equivalent representations for objects of the latter two classes is then argued to vindicate non-parcimonious use of both of the conplementary resources of "algebraic" and "geometry-like" representation, and to lead to a non-formalist view of formalisation (with a distinction between "symbol", "grapheme" and "inscription-token"). Ch IV reviews some fundamental results. Ch. V shows (via assessment of the notion of "adequation" between "formal(ized)" and "nai͏̈ve" theories) how a vital need could be felt for a "neo"-Bloomfieldian inductivist refundation (with semantic-free "contrast" and checking of representations for "bi-univocity" and "linearity"). It argues (via proof of the "heritage principle") that though constituent grammars are as "adequate" as can be, the Chomskyan invalidation claims (as involving considerations of empirical relevance) must fall short of theorem status. The conclusion argues that though the solutions ever offered are variations on an invariant formal scheme, the object of Chomskyan theory has known one radical (internalist) shift; that any theory claiming to be explicit and to account for accessibility of meaning or reference by mere use of formal representation or of sets of primitive "individuals" as values is vacuous; and that Chomskyan theory lacks the form it would need for the claimed ontological parity with physics. Throughout, "hyperbolic explicitation" is used as an expository and heuristic method.