My 2009 book manuscript addresses in some depth consequences of the distinction I make between PF representations and pf representations. But, in an attempt to make the presentation simpler, the CUP book does not emphasize the distinction. That has, however, resulted in difficulty in straightforwardly addressing the nature (or what one may call the ontology) of Schemata.

One of the footnotes of Chapter 3 now contains:

***
A pf representation is a non-hierarchical phonetic sequence of audible items that is directly read off a PF representation. I assume that the PF representation, unlike the pf representation, expresses hierarchical relations among the items taken from the mental lexicon and it may contain syntactic objects that have no phonetic content (so-called "empty categories"). A Schema, such as [SA1a] and [SA1b], covers, i.e., can be instantiated by, an infinite number of pf representations. It minimally specifies where the two items mentioned in the bridging hypothesis (alpha and beta of BVA(alpha, beta) in the case of BVA) occur in a phonetic sequence. The use of the square brackets in a Schema, as in [SA1] and other Schemata to be discussed below, is a reflection of the fact that the pf representation is based on a PF representation. Insofar as the hierarchical information thus made reference to by a Schema is part of the elementary constituent structure that transcends different conceptions of the CS, a Schema, and hence a schematic asymmetry, can be understood as being theory-neutral. It is in this sense that confirmed predicted schematic asymmetries are theory-neutral though the predicted schematic asymmetries in question are given rise to by our hypotheses.
***

The Glossary entry for "Schema" is as in:

***
Schema:
A schematic representation that covers, i.e., can be instantiated by, an infinite number of pf representations. It minimally specifies where the two items mentioned in the bridging hypothesis (alpha and beta of BVA(alpha, beta) in the case of BVA) occur in a phonetic sequence in such a way that (i) any pf representation instantiating Schema B is predicted to be unacceptable and (ii) some pf representations instantiating Schema A are acceptable, at least to some extent, with the dependency interpretation specified by the bridging hypothesis.
***