Grammar Discovery Procedures
Robert F. Longacre · 1973
THE NOTION OF GRAMMARHow is grammar related to other aspects of linguistic structure?Until recently American structural linguistics has assumed a model of language in which phonemes built into morphemes which in turn built into syntactic units.As a result, phonology, morphology, and syntax were regarded as successively higher layers of structure.Generative grammar has turned this model upside down and ordered it rule-wise with a cover symbol for sentence as the first rule and phonological rules for transcription into terminal sentences as the last section of rules.However novel may be certain aspects of generative grammar, it has not challenged the model in any essential way. 1 The present procedures are based on a more radical departure from former American structuralism than that found in generative grammar.It is here assumed that language is structured in three semiautonomous but interlocking modes, phonology, grammar, and lexicon (Pike's trimodalism).Phonology is not taken up into morphology which is in turn taken up into syntax.Conversely, 1 It may be objected that we are confusing representation with structure.Transform grammar claims, however, to provide insight into the nature of language by giving formal expression to our deepest intuitions about language.Transform grammar cannot brush off criticism with the reply that certain devices and ways of talking are merely representational symbols while claiming at the same time that these symbols give adequate expression to linguistic reality.s Compare Lees, Grammar of English Nominalizations, IJAL 26, no. 3, Part II (1960), pp.20-3.Under "Lexicon", Lees lists 47 word classes presented as follows: Aa mysterious, nice, obvious, understandable... Ad green, long, old, sad, wise.... In every case the use of suspension points indicates that the list is exemplary and suggestive.Obviously, a partial listing of this sort will not allow one to generate all possible sentences.For the latter an exhaustive cross reference dictionary would be necessary.In bulk such a work would be many PROCEDURE AND THEORYAny useful procedures must be based on an adequate theory of the ' Cf.Pike's codification (Phonetics, Ann Arbor, 1943) and Hockett's (Manual of Phonology, Baltimore, 1955)." Pike, "Dimensions of Grammatical Structure", Language 38.221-44 (1962).11 For criticism of immediate constituent analysis and presentation of systemic grammatical hierarchy, see Longacre, "String Constituent Analysis", Language 36.63-88 (1960), and Zvelebil "How to Handle the Structure of Tamil", Archiv Orientálni 30.116-42 (1962)." For an application of the two-fold structural contrast in which external distribution of a syntagmeme is considered to be relevant, see Pike."Dimensions of Grammatical Structure", p. 232.Pike believes that the dual criterion can not be applied in some cases without direct consideration of semantics.Pike, furthermore, does not insist that the dual structural contrast affect the nucleus of a syntagmeme." The number of permutations possible for a given reading is : x!-y (x = number of symbols in a reading; y = number of permutations not possible because of special rules; the x! is factorial, e.g. if χ has the value of 4, then 1 χ 2 χ 3 χ 4)." EXPONENT as used here must not be confused with superscript.Cf. the use of exponent in Halliday's "Categories of the Theory of Grammar" ( Word 17. 241-292,1961) and in Mcintosh's "Patterns and Ranges" {Language 37.325-337