Date: Fri, 28 Mar 2003 18:40:47 -0600 From: Gunnar Olafur Hansson Subject: Laboratory Phonology 7
Gussenhoven, Carlos and Natasha Warner, eds. (2002) Laboratory Phonology 7, Mouton de Gruyter, Phonology and Phonetics 4 (1).
Gunnar Olafur Hansson, University of Chicago.
The volume under review contains a selection of papers presented at the 7th Conference on Laboratory Phonology (Nijmegen, 2000). It is also the first LabPhon volume to appear under a new publisher, and that change seems to have been beneficial in several respects. Unlike most of its predecessors, this LabPhon volume is available in paperback, which makes it far more affordable, especially to students. Secondly, the delay between the conference itself and the appearance of the proceedings -- a common source of dissatisfaction about earlier volumes (see, e.g., Marija Tabain's review of LabPhon 5 in http://linguistlist.org/issues/12/12-1991.html) -- has been reduced considerably. In fact, the current volume appeared several months ahead of the immediately preceding one in the series. A further improvement over previous LabPhon volumes is that every paper (except for the commentaries) is now accompanied by an abstract summarizing the topic under investigation and the findings reported.
The collection of papers is divided into two parts, the first dealing with phonological issues from a psycholinguistic perspective ("Part I: Phonological Processing and Encoding"), and the second addressing the phonetics-phonology relationship more generally ("Part II: In the Laboratory and in the Field: Relating Phonetics and Phonology"). As in earlier LabPhon volumes, the individual contributions are organized into blocks of 3-5 papers, where each block is followed by a commentary article.
Within Part I, the first block of articles deals with issues related to phonological encoding in production, i.e. the retrieval of phonological forms from the lexicon and the construction of surface representations. Daniel Jurafsky, Alan Bell & Cynthia Girand ("The role of the lemma in form variation") use corpus-based statistical methods to investigate whether phonetic variation can be directly sensitive to lemma distinctions--e.g., in cases like infinitival vs. prepositional "to", where two separate lemmas share the same phonological form. Niels O. Schiller, Albert Costa & Angels Colome ("Phonological encoding of single words: In search of the lost syllable") critically evaluate the masked priming paradigm as a method for testing the role of syllables as units of phonological encoding. Vincent J. van Heuven & Judith Haan ("Temporal distribution of interrogativity markers in Dutch: A perceptual study") use a gating task with synthesized Dutch intonation patterns to investigate the temporal development of cues signalling the statement vs. question contrast in Dutch. This is followed by a commentary article by Willem Levelt ("Phonological encoding in speech production") which, among other things, includes a very brief historical overview of the study of phonological encoding and the various experimental paradigms that have been developed in that area.
The next block consists of five articles addressing issues of phonological processing. In the first of these, Janet B. Pierrehumbert ("Word-specific phonetics") outlines an exemplar-based model of speech production designed, in part, to cope with word-specific phonetic detail; her proposal is discussed below. Danny R. Moates, Zinny S. Bond & Verna Stockmal ("Phoneme frequency in spoken word reconstruction") explore the effect of phoneme frequency in word reconstruction tasks--adjusting a single segment to convert a nonword into a word--to test whether frequency can account for the greater "mutability" of vowels than consonants in such tasks. Haruo Kubozono ("Temporal neutralisation in Japanese") documents the pervasive neutralization of vowel length contrasts in word-final position in Japanese, both in phonological processes and in auditory perception, and shows in a series of experiments how the same effect is also found in perception of speech by visual stimuli alone. Sharon Peperkamp & Emmanuel Dupoux ("A typological study of stress 'deafness'") demonstrate how speakers of different languages with non-contrastive stress systems differ in their ability to perceive stress contrasts, and outline an explanation in terms of a particular timeline of language acquisition. Ann R. Bradlow ("Confluent talker- and listener-oriented forces in clear speech production") investigates the extent of coarticulation in hyperarticulated speech, as well as the question of whether hyperarticulation is "global" or relativized based on such listener-oriented factors as perceptual confusability. A commentary article by Ann Cutler ("Phonological processing") concludes this block of papers.
Part II is a somewhat more heterogeneous collection of studies dealing with the relationship between phonetics and phonology broadly speaking, organized into three blocks. The first consists of four articles investigating various aspects of the phonetics-phonology interface. George N. Clements & Sylvester Osu ("Explosives, implosives, and nonexplosives: the phonological function of air pressure differences in stops") investigate the phonetics and phonology of a class of segments they call "nonobstruent stops"--including, e.g., implosives and laryngealized stops--and report on an articulatory and acoustic study of two such segments in Ikwere which appear to defy traditional phonetic classification. Maria-Josep Sole ("Assimilatory processes and aerodynamic factors") investigates the assimilation of lingual fricatives to apical trills, and explores the role of aerodynamic constraints in accounting for these and other assimilations involving fricatives. Sonia Frota ("Tonal association and target alignment in European Portuguese nuclear falls") presents acoustic evidence suggesting that nuclear falls in Portuguese declarative intonation are to be treated as bitonal pitch accents (HL) rather than as transition effects, and investigates the timing relationship between the two components of such bitonal accents. Ioana Chitoran, Louis Goldstein & Dani Byrd ("Gestural overlap and recoverability: Articulatory evidence from Georgian") use a magnetometric production study of Georgian stop-stop clusters to test (and confirm) the hypothesis that the extent of gestural overlap in such clusters is dependent on the perceptual recoverability of C1--and thus, indirectly, on such factors as position within the word and place-of-articulation sequencing. Bruce Hayes, in his commentary article ("The phonetics-phonology interface"), takes up some of the questions left unanswered by Chitoran et al., and conjectures that some of the Georgian stop-stop clusters may in fact be produced with an egressive velaric airstream mechanism.
The next block consists of papers that are either studies based on fieldwork or that address general questions of how phonetic data gathered in the field can be used to test phonological theories. In the first of these, Didier Demolin ("The search for primitives in phonology and the explanation of sound patterns: The contribution of fieldwork studies") takes as his point of departure the hypothesis that phonology is emergent and derived from phonetics (cf. Lindblom 2000), and shows several examples of how fieldwork data from underdescribed languages can be brought to bear on this hypothesis. The remaining three articles all deal with prosodic phenomena, using fieldwork data to call into question traditional categories and typological classifications in this domain. Esther Grabe & Ee Ling Low ("Acoustic correlates of rhythm class") question the rhythmic classification of languages as either stress-timed, syllable-timed or mora-timed, using measurements of durational variability in the acoustic signal that do not refer to phonological constructs such as "syllable" or "stressed syllable"; they find that many hitherto unclassified languages are intermediate between the traditional categories. Jose I. Hualde, Gorka Elordieta, Inaki Gaminde & Rajka Smiljanic ("From pitch accent to stress accent in Basque") question the categoriality of the typological distinction between pitch-accent and stress languages, showing how western dialects of Basque constitute a continuum along this dimension. Bert Remijsen ("Lexically contrastive stress accent and lexical tone in Ma'ya") uses phonological and acoustic analysis to argue that Ma'ya, an endangered Austronesian language, displays a previously unattested combination of a (three-way) lexical tone contrast and a lexical stress contrasts. The section is concluded with a commentary article by W. Leo Wetzels ("Field work and phonological theory").
The final block consists of a single article by Aditi Lahiri & Henning Reetz ("Underspecified recognition"), followed by a commentary article by Dafydd Gibbon ("Speech recognition"). In their paper, Lahiri & Reetz develop a model of the lexicon that makes crucial use of abstract, featurally underspecificied phonological representations. They support their model with evidence from historical change, language comprehension experiments and, most importantly, its performance when implemented computationally in an automatic speech recognition system.
The individual contributions that make up this volume are so numerous and so diverse that it is impossible to do justice to all of them beyond the brief descriptions given above. Here I will only mention in somewhat more detail two papers which, in my mind, stand out due to the scope of the issues they address and their implications.
The first of these is Didier Demolin's article on the contribution of empirical fieldwork to the search for primitives and first principles in phonology. This paper is quite programmatic in character; Demolin is a strong advocate of the "emergent phonology" approach (see, e.g., Ohala 1990; Lindblom 2000). This approach sees the fundamental units and processes of phonology as deductively derived from independent premises anchored in physical and physiological realities. Demolin argues that experimental research "in the field" on uncommon sound patterns can be brought to bear on this hypothesis. He surveys a wealth of fascinating phenomena from relatively underdescribed languages to demonstrate their dependence on aerodynamic principles as well as on principles relating acoustic output to vocal tract shape and to auditory representation. Sound patterns involving ejective fricatives in Amharic and bilabial trills in Mangbetu are shown to be critically shaped by aerodynamic constraints. The rich vowel inventory of Nuer is used to test principles of maximal dispersion, and it is demonstrated that several factors, physical as well as physiological, converge to account for the distribution of vowels in acoustic space. Vowelless syllables in Lendu (specifically ones where the nucleus is a sibilant), and sound changes involving clicks in Khoisan languages, confirm the important role played by perceptual and auditory factors in phonological patterning. A final question addressed by Demolin is that of the cognitive representation of sound patterns, and the psychological reality of such constructs as syllables, features, etc. Complex word game permutation patterns in Hendo (an unwritten language) are mentioned as a potential challenge to the claim that awareness of segments/phonemes is dependent on familiarity of an alphabetic writing system; however, it turns out that a quasi-alphabetic (and even feature-based!) system of color symbols is in fact used to teach the word games during initiation rites. This last part of Demolin's article is perhaps the weakest; as pointed out by Wetzels in his commentary, many of the operations typical of word games, such as infixation and the reduplication of constituents of various sizes, do exist as grammatical operations in languages, written and unwritten.
Janet Pierrehumbert's article on word-specific phonetics challenges many commonly-held views (at least among phonologists) about the nature and amount of phonetic information that goes into lexical representations. Unlike in the heyday of structuralist and classical-generative phonology, it has by now become generally accepted that a great deal of fine-grained phonetic detail is language-specific, thus necessitating at least something like a "phonetic implementation" component in the phonological systems of individual languages. What many recent studies have brought to light, however, is that the patterning of such allophonic detail is not as across-the-board as it originally appeared, but that much of it is in fact specific to individual words in the lexicon. Pierrehumbert discusses how such phenomena can be straightforwardly accounted for by an exemplar-based model of speech production. In such a model, each word is associated with its own frequency distribution over phonetic outcomes; these "clouds" of exemplars, stored in long-term memory, are continuously and incrementally updated based on experience, which allows word-specific allophonic detail to accrue over time. There are two aspects of Pierrehumbert's proposal which are particularly significant. The first is the general idea of extending exemplar theory, which was originally developed for speech perception, to the domain of speech production. The second has to do with a problem that immediately arises in a simple model where words activate exemplar clouds directly, rather than mediated by decomposition into phonological units. The long-term results of such a system would be arbitrary dispersion across phonetic hyperspace, rather than the systematicity and granularity that is so characteristic of both synchronic phonological systems and of (Neogrammarian) diachronic sound changes. Pierrehumbert's solution is to develop a hybrid model which uses an intermediate level of phonological encoding and buffering, where a phonological parse is assigned; the contents of this representation probabilistically evoke regions of the exemplar space as production goals. In this model, word-specific effects are second-order phenomena: rather than providing holistic production goals, words are able to bias productions. One of the virtues of the model is that it straightforwardly accounts for cases of "allophonic transfer" between morphologically related forms (see, e.g., Steriade 2000).
To conclude, this latest volume in the Laboratory Phonology series is a highly impressive collection of studies on a wide variety of topics, contributing to our understanding of the relationship not only between phonetics and phonology but also between each of these and the psycholinguistic domains of encoding and processing. It is worth emphasizing that laboratory phonology is not a theoretical framework in the usual sense; rather, it is "a coalition amongst groups of people, with some working in one or another of the various current frameworks, and others working in no phonological framework at all" (Pierrehumbert et al. 2000:279). As a consequence, it is not uncommon to find inherently conflicting or contradictory proposals within the laboratory phonology literature. This volume is no exception; for example, it is hard to see how Lahiri & Reetz' model, where lexical representations are heavily underspecified (and couched in discrete phonological features), can possibly be reconciled with exemplar-based models of perception and production such as the one advocated by Pierrehumbert. Similarly, there are no doubt many "laboratory phonologists" who would disagree with Clements & Osu's decision to account for the behavior of nonexplosive stops in terms of a new binary phonological feature [+/-obstruent]. What the "genre" of laboratory phonology studies does have as its common denominator, however, is the belief that careful experimental methodology, explicit mathematical modelling (typically in terms of continuous rather than discrete mathematics), and cumulation of results can and will strengthen the scientific foundations of phonology as a discipline. The respect and citation record that the LabPhon series has attained, during the sixteen years since its inception, is ample attestation to the vitality and viability of this research paradigm.
Lindblom, B. 2000. Developmental origins of adult phonology: The interplay between phonetic emergents and the evolutionary adaptations of sound patterns. Phonetica 57, 297-314.
Ohala, J. 1990. The phonetics and phonology of aspects of assimilation. In J. Kingston & M. Beckman (eds.), Papers in Laboratory Phonology I: Between the Grammar and the Physics of Speech, pp. 258-275. Cambridge: CUP.
Pierrehumbert, J., Beckman, M. & Ladd, R. 2000. Conceptual foundations of phonology as a laboratory science. In N. Burton-Roberts, P. Carr & G. Docherty (eds.), Phonological Knowledge: Conceptual and Empirical Issues, pp. 273-303. Oxford: OUP.
Steriade, D. 2000. Paradigm uniformity and the phonetics-phonology interface. In M. Broe & J. Pierrehumbert (eds.), Papers in Laboratory Phonology V: Acquisition and the Lexicon, pp. 313-335. Cambridge: CUP.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER Gunnar Olafur Hansson is Assistant Professor of Linguistics at the University of Chicago. He received his Ph.D. in 2001 from the University of California, Berkeley, with a dissertation entitled "Theoretical and typological issues in consonant harmony". The focus of his research is on theoretical phonology; specific areas of interest include harmony systems, derivationally opaque sound patterns, the phonetics-phonology and morphology-phonology interfaces, and diachronic processes of phonologization and morphologization.