This paper is an attempt to apply to the phonological hierarchy in particular views expressed recently by Pike (Pike, 1972) on the nature of linguistic hierarchies in general. In a paper presented to the 11th International Congress for Linguistics, Bologna, 1972, entitled Recent developments in tagmemics, Pike spoke of the possibility of developing ‘paired levels’ or a ‘paired ladder’ in a hierarchy. His main concern in that paper, and in a similar paper presented to the German branch of the Summer Institute of Linguistics, 1971, was with the grammatical hierarchy. It is my aim in this paper to show that ‘paired levels’, or ‘double ranks’ as I shall call them, apply equally well in the phonological hierarchy, and that they offer interesting possibilities of (a) developing the notion of interpenetration of hierarchies and (b) broadening the field of general theoretical categories of language.

I. …any noticeable reiteration of the same grammatical concept becomes an effective poetic device. Any unbiased, attentive, exhaustive, total description of the selection, distribution and interrelation of diverse morphological classes and syntactic constructions in a given form surprises the examiner himself by unexpected, striking symmetries and antisymmetries, balanced structures, efficient accumulation of equivalent forms and salient contrasts,… by rigid restrictions in the repertory of morphological and syntactic constituents used in the poem, eliminations which… permit us to follow the masterly interplay of the actualized constituents. Let us insist on the strikingness of these devices (Jakobson, 1968: 602–3).

The first thesis concerning phonological representations enunciated by Anderson and Jones (henceforth A&J)(1974: 8) is ‘that phonological representations are more highly structured than the standard theory would claim’, where the level of structure with which they are particularly concerned is that of bracketing into syllables. The principles in terms of which this bracketing is assigned are given as (I) (= their (19)):

(I) (a) Medial clusters are composed of a syllable-final preceding a syllable- initial sequence. (b) ‘Precede’ includes overlap where possible.

A couple of points of elucidation are necessary here. First, it should be noted that a previous formulation of (Ia) in the same paper (4) states: ‘We claim that medial clusters are combination of clusters that can be final in monosyllables and clusters that can be initial.’ This reference to word-final and initial clusters must be borne in mind if (Ia) is not to be construed as either circular or trivial.

A Duke who has not been satisfactorily identified with any historical figure is lampooned in a traditional rhyme (believed to have been directed originally at the King of France) as follows:

The Grand Old Duke of York

He had ten thousand men

He marched them up a great high hill

And he marched them down again.

The implication is that this was incompetent and self-defeating activity on his part. Linguists very frequently seem to give evidence of a tacitly held belief that there is similarly something inept and risible about a linguistic analysis which determines that certain structures are assigned a derivation of the general form A→B→A, that is a derivation in which an underlying representation (or some nonultimate remote representation) is mapped on to an intermediate form distinct from it, and then on to a surface (or other superficial) representation which is identical with the earlier stage.

Within the standard theory of generative phonology, the phonological component consists of a set of partially ordered rules; where G1 and G2 are synchronic grammars of a given language L at chronologically earlier and chronologically later stages of L respectively, G1 and G2 may differ in at least the following ways: (a) G2 may contain new (phonological) rules ordered later than all rules in G1; (b) G2 may contain new rules ordered before some rules already present in G1 (c) G2 may contain the same rules as G1, but arranged in some different order; (d) G2 may operate upon a different underlying representation from that for G1 (and hence contain, to a greater or lesser degree, quite different rules); (e) some rules present in G1 may no longer be present in G2 (f) some rules o7f G1 may be present only in a modified form in G2.

The primary aim of this paper is to show by the analysis of an extended example that a phonological description which recognizes the functional variety of phonological description which recognizes the functional variety of phonological rules is more illuminating than one in which the data are handled merely as the output of a set of completely undifferentiated processes. In particular I hope to emphasize the value of distinguishing MOTIVATED and UNMOTIVATED processes in phonology.