Prosodic Morphology before the 1990s

One who consults Chomsky and Halle's (1968) Subject Index for entries beginning with pros is referred exclusively to passages discussing prosodic features. This is just one manifestation of this book's explicit claim that hierarchy has no direct role in the study of sound phenomena. Even though the input to the phonological component (a separate morphological component, including a lexicon, became available only later) includes rich hierarchical structures, in that it is the output of the grammar's syntactic component, these structures are flattened out in a so-called intermediate readjustment component, resulting in completely non-hierarchical, flat strings. These strings are sequences of soundsegments and non-segments. The individual sound-segments represent the “sounds” of a given morpheme or word in an abstract sense, through binary-valued phonological features. These features include the ones traditionally called “prosodic,” such as [stress], [pitch] and [tone] (376–77), but they have no special status vis-à-vis other features, just as for instance “cavity features” ([anterior], [coronal], [back], etc.) have no special status. Non-segments are the boundary symbols and the syntactic brackets, such as those occurring between a stem and an affix, between two words, between two phrases, and so on. Roughly speaking the number of boundary symbols and/or brackets between two morphemes reflects their degree of “coherence” (or, conversely, their comparative degree of embedding in the syntactic tree), but this not a hard-and-fast rule: not only is it the task of the readjustment component to flatten out hierarchical structure, but some of the rules of this component are also allowed to manipulate (segments and) non-segments, resulting in derived representations from which the original syntactic structure cannot any longer be automatically read off.