2. Historias cortas de color local
2.1 La historia corta
In this chapter, following work in Cyclic Phonology and Stratal Theories, I motivated the fact that some phonological rules (“cyclic”) apply iteratively with the addition of a morpheme while others wait until all morphemes are combined in order to apply (“non-cyclic”). Additionally, some mor- phemes seem to trigger the application of the cyclic phonology while others do not. Because there is no morphosyntactic reason to motivate a structural difference between these types of morphemes, I posited that the status of cyclic or non-cyclic is determined by diacritic marking. This diacritic status is also supported by the fact that different languages divide up the space of cyclic and non-cyclic morphemes differently, which I attribute to different strategies that learners take to make sense of a underlyingly random (= diacritic) system.
I introduced an explicit model for the calculation of the cyclic phonology (the phonocyclic buffer) and posited that non-cyclic morphemes are not integrated into a buffer, but I did not discuss how the calculation of the “non-cyclic” phonology is executed. This topic will be covered in Chapter 3.
Chapter 3
The Word Domain
3.1
Overview
In Chapter 2, I motivated a difference between cyclic phonology, phonological processes that ap- plied with the addition of each (cyclic) morpheme, and noncyclic phonology, phonological pro- cesses that wait until the entire word is built to apply. The question addressed in this chapter is: what is the domain of this noncyclic phonology?
I propose that the noncyclic phonology applies exactly within the domain of the phonological word. While I referred to the relevant block of phonological processes as “noncyclic” in the previous chapter to emphasize the contrast with cyclic phonology, in this chapter (and henceforth) I will refer to this phonological block as the word level, orř-Level phonology, to emphasize the comparison with other notions of “word”.
In Section 3.2, I discuss some of the various notions of word that have been presented in the literature. I propose that the phonological word of this framework (theř-Word) is, at least for the most part, equivalent to noncyclic and postcyclic phonologies in cyclic theories, the “word level” of stratal lexical theories, and the Prosodic Word (PWd) of Prosodic Hierarchy Theories (although the comparison with prosodic theories will continue in Chapter 4 with the discussion of affixation and cliticization).
I then turn to the problem of the structural domain of theř-Word. In the piece-based syntactic theory of morphology followed here, there is no inherent underlying notion of word as a syntactic
or semantic unit. That is, the word is not a meaningful or necessary unit for the syntactic side of derivation in this framework (see, e.g., Marantz 1997; Embick 2010a, 2014). However, it has been posited that there is a derived structure, the complex head or morphosyntactic word (M-Word), whose domain corresponds to that of theř-Word.
To motivate this correspondence, I present case studies in German and English voicing phenom- ena in Section 3.4 which show that theř-Level phonology must be calculated including all of the units in the M-Word. Furthermore, theř-Level phonology applies exactly within the bounds of the M-Word and not between M-Words.
It must be noted that the notion that there is some correspondence between some syntactic, grammatical, or lexical structure and phonological structure is not new. There have been a variety of correspondences posited, not the least of which are theories which include a word-generating module called the lexicon. For these theories, the word is of utmost importance because it the unit which is passed between the lexicon and the syntax (and semantics). It is a surprise for these theories, then, that mismatches have been identified between “grammatical words”, defined as cohesive morphosemantic units, and “phonological words”, defined as cohesive phonological units (see, e.g., Dixon and Aikhenvald 2002, pp. 27–30). That is, there appear to be cases where a single grammatical word may correspond to multiple phonological words and vice versa. For Lexical theories, this motivates a disjunction and non-interaction between the grammatical system which generates the words on the structure/semantics side and the phonological system which organizes the sounds to be pronounced.
In Sections 3.5–3.9, I present a few case studies where such mismatches have been proposed and show that we can explain these phenomena without disrupting the correspondence between M-Words and ř-Words. In one direction, the phenomena of “Aoyagi” prefixes in Japanese and interleaving word order in Vietnamese have been proposed to be cases of single grammatical words surfacing as multiple phonological words. In the other direction, compounds, such as those seen in Spanish, and long polysynthetic words, such as those seen in Plains Cree, have been proposed to be multiple grammatical words corresponding to a single phonological word. I show that, given a closer look, these do not appear to be mismatches and we can use the framework presented here to
explain the data we see. Given this resolution of mismatches, there seems to be no reason to starkly divide the grammatical (here: syntactic) system from the phonological system.
Finally, although I propose a correspondence between M-Words andř-Words, this correspon- dence seems to be one way; complex M-Words always surface asř-Words, but not allř-Words are underlyingly M-Words. There seems to be at least one other way in which ř-Words can be created. In Section 3.10, I introduce the process of Stray Terminal Grouping through which linearly adjacent morphemes which otherwise do not belong to a ř-Word may be grouped together into their ownř-Word. Although these phenomena represent a “mismatch” between M-Words and ř- Words, the grouping of stray morphemes into ř-Word perhaps has more to do with the need for every phonological item to belong to a group for the purposes of pronunciation than any deep or systematic split between the sound and structure systems.
Case studies in this chapter will more-or-less ignore the phenomenon of clitics, although it will come up, especially in the discussion of Stray Terminal Grouping. The discussion of theř-Word and the interaction with clitics will be taken up in Chapter 4.