1. At one point (p. 366) in hisMethods in Structural Linguistics(Chicago, 1951), Harris describes this as “the over-all purpose of work in descriptive linguistics.” Later, however, he remarks (p. 372) that “the work of analysis leads right up to the statements which enable anyone to synthesize or predict utterances in the language.” It is this set of statements that I am calling the grammar. I am not convinced that the last-quoted remark is really justified (that is, that any known procedures do provide an adequate grammar), but this is another question. At the moment I am only concerned to make the difference in aim explicit.
2. In grammars of the type studied in [SS], part of the structural description of a phone sequence is its representation on the phrase structure level by a set of abstract strings that can be presented, equivalently, as a tree with labelled nodes. See [SS], and references there, for details. The fully grammatical sentences generated by the grammar are the ones represented by trees headed by a single node labelledS, whereSis the distinguished initial symbol of the underlying constituent structure grammar. Transformational rules also play a part in assigning a phrase structure representation, in that theory, although the question exactly how they do so has not yet been satisfactorily answered. For discussion of this see myLogical Structure of Linguistic Theory(mimeographed, 1955, henceforth, [LSLT]), of which [SS] is essentially an excerpt, and “On the notion ‘rule of grammar’,”Proceedings of the Symposia on Applied Mathematics, XIII (1961).