Workshop on Representations in Intonational Phonology (RIP) 2020

In segmental phonology, the notion of a direct or one-to-one relationship between abstract symbols (features, segments) and the elements of the acoustic signal that realize them has all but been abandoned. We no longer believe that every aspect of the signal that is under the speaker’s independent control must be “specified” by a distinct feature in the phonology (à la Chomsky and Halle 1968), any more than we believe that phonological features can be read directly off invariant properties of the acoustic signal. Instead, abstract phonological features are typically held to represent sets of contrasts among classes of segments, which are implemented by clusters of language-specific, contextually variable and dynamically interacting cues in the phonetics.

In intonational phonology, by contrast, typical analyses remain more tightly bound to the acoustic signal, reflecting a pervasive sense that F0 contours can be inspected for purposeful events (i.e. targets), and that our abstract representations can be read more or less directly off of these. An F0 maximum in the signal seems to imply, necessarily, a High tonal autosegment of some kind, somewhere, in the abstract phonological representation. Phonology, in other words, still seems mostly there to provide detailed instructions to the phonetics, and can be read, still, relatively transparently, off of it.

Part of the reason for this may be that the phonological processes and natural class behaviors that so often motivate representational decisions in segmental phonology are notably less abundant when it comes to intonation. Indeed, Pierrehumbert (1980: 11) observes, somewhat offhandedly: “In other languages, rules which alter tonal values or delete tones can apply to such a representation. English appears to lack such rules, with the result that the underlying and derived phonological representations of intonation are identical. The rules of interest are thus the rules which assign phonetic values to tones and construct the F0 contour between one tone and the next.”

But is this in fact true, either of English, or of intonational phonology more generally, and if so, why? Given this, furthermore, how are we to arrive at decisions regarding the nature of our phonological representations, beyond simple inspection of the phonetic record? In segmental phonology, as soon as we untether phonological representations from the signal to the extent that the facts seem to warrant, a host of representational decisions become substantially murkier. Is a given contrast really about [voice], or is it instead [spread glottis]? Is a given vowel contrast about [ATR], or is it instead breathy vs. modal voice, when both vowel quality and voice quality cues are present, and dynamically interacting, in its realization? In tone systems, when a contrast seems to use both timing and scaling cues in its implementation, in a trading relationship that seems also to vary in its details from speaker to speaker (as in Gothenburg Swedish [Segerup & Nolan 2006, Barnes, et al. 2015], and potentially Shilluk [Remijsen & Ayoker 2014, Barnes, et al. 2019]), how do we know whether this is encoded in the phonology as timing, or as scaling? Perhaps it is both, or again perhaps neither? How would speakers know, and does it matter? If features are emergent, rather than universal, as many now suggest, the problem becomes all the thornier. If, in despair, we resign ourselves to contrasts designated by phonetically contentless abstract categories, Segmental Class A vs. Segmental Class B, for example, how much, exactly, do we stand to lose? And is anything comparable gained?

For this workshop we invite papers addressing questions related to any of the issues raised above. Sample question areas contributors may wish to address include:

  • Is intonational phonology different from other phonological systems in important ways? How (dis)similar are intonational phonology and (lexical/grammatical) tonal phonology? How (dis)similar are intonational phonology and segmental phonology?
  • Does every acoustic event in a pitch contour (or in prominence marking, or in phrasing) map onto a phonological element? (I.e. how direct is the relationship between acoustic-phonetic events and elements of the symbolic-phonological representation?)
  • To what extent do we expect phonological elements in intonation always to map onto a (constant) set of phonetic exponents?
  • Are there phonological processes or patterns in intonation systems that are best analyzed in terms of the insertion, deletion, or rearrangement of abstract phonological symbols, rather than at the level of phonetic interpretation thereof? Are there any ‘null’ elements in intonational phonology? (Like the now-abandoned downstep-inducing trailing L of the English H*+ L pitch accent in Pierrehumbert 1980?)
  • What is the nature of phonological features in speech prosody? (Are there, for example, natural class behaviors that require phonological representations to look one way (and not another) symbolically? To what degree are phonological features defined by the acoustic characteristics that manifest them?)
  • What kinds of data and/or arguments are relevant for resolving these questions? What counts as an argument in favor of / against a particular stance related to these questions?

Submissions should follow the INTERSPEECH 2019 guidelines. The number of pages is up to 4 for text with an additional page only for references.

The INTERSPEECH 2019 kit for papers (LaTeX and MsWord) can be found here:

Submissions via SPro2020’s EasyChair page no later than December 20th, 2019,
23:59 Tokyo time:

Select [WS3] in the ”Workshop topics” section.