Segmental and prosodic issues in Romance phonology contains a selection of papers presented at the second Phonetics and Phonology in Iberia conference, held in Barcelona in 2005. The focus of the volume is on Romance languages and ‘on descriptive and theoretical issues in the phonology of Romance languages’ (p. vii). Most of the contributions are concerned with experimental research, and follow a laboratory phonology approach. Laboratory phonology tests the predictions of phonological analyses through experimental investigations of phonetic behaviour (Kingston Reference Kingston1999); similarly, it uses experiments to collect concrete data that motivate phonological analysis (Face Reference Face2004: ix). Since laboratory phonology is one important theme emerging from this book, papers will be discussed with regard to their contributions to this field and to phonology in general.
Papers are organised around three content areas: ‘Segments and processes’ (Part 1), ‘Prosodic structure’ (Part 2) and ‘Acquisition of segmental contrasts and prosody’ (Part 3).
In Part 1, Noël Nguyen, Sophie Wauquier-Gravelines, Leonardo Lancia & Betty Tuller report on the results of several experiments designed to determine whether liaison consonants in French are processed and perceived differently from fixed consonants. If acoustic factors can be ruled out as a cause of any differences in processing, the experiments may offer support for one of the two theoretical approaches that have been used to explain liaison – autosegmental and lexical (exemplar-based). It was found that liaison Cs are more difficult to detect than fixed Cs and the differences in detection rate could not be accounted for by the target's acoustic characteristics. While failure to detect liaison consonants may seem attributable to higher-level factors, partial evidence was found for both autosegmental and exemplar-based models of phonology. This paper contains an excellent background section that reviews phonological accounts of French liaison and explains the motivation for the experimental studies; similarly, the results are well contextualised with regard to their significance for phonological analyses of liaison. This article is a textbook illustration of the goals and principles of laboratory phonology.
Maria-Josep Solé's paper focuses on coarticulation, feature co-occurrence restrictions and phonotactics. More specifically, the author explores the hypothesis that, since acoustic and aerodynamic factors are often at the root of within-segment feature instability, they may also serve to explain instability in adjacent segments where there is coarticulatory overlap. For instance, the feature combination of frication and nasality tends to be unstable cross-linguistically, because of the difficulty involved in maintaining the high pressure build-up required for frication with velum lowering for the nasal, which reduces the pressure needed for frication. Solé reports on the results of an experiment designed to test the hypothesis that, when frication and nasality co-occur in contiguous segments, the tendency of prenasal fricatives to weaken or disappear can be explained as the result of anticipatory velophraryngeal opening for the nasal overlapping the acoustic duration of the fricative. Her results confirm her hypothesis – velophraryngeal opening was found overlapping the acoustic duration of the fricative – and lead her to conclude that featural instability can explain a number of phonological patterns, such as sound change and morphological alternations, in addition to feature co-occurrence restrictions. An important conclusion of Solé's study is that some of the same principles may serve to explain syntagmatic and paradigmatic aspects of phonology. Another important implication for phonological theory, not mentioned by Solé, is that feature instability across segments may also provide an explanation for yet unaccounted for phonotactic restrictions, such as the ill-formedness of the sequence /ls/ across morpheme boundaries in diminutive formation in Spanish (e.g. Crowhurst Reference Crowhurst1992, Colina Reference Colina2003; but see Ohala Reference Ohala1997), and the rarity of /sr/ combinations in morpheme-internal position (e.g. Israel ‘Israel’) and their reduction to [r] across words (captured by the constraint *[strident][vibrant] proposed by Bradley Reference Bradley2006; see also Solé Reference Solé, Gussenhoven and Warner2002).
The article by Daniel Recasens, also on coarticulation, argues that the extent and direction of coarticulation (regressive or progressive, vowel anticipatory or carry-over effects) in VCV sequences can be explained by the Degree of Articulatory Constraint (DAC) model of coarticulation. The degree of coarticulation of a C or V depends on the degree of coarticulatory constraint or DAC values: highly constrained segments are more resistant to coarticulation and exert more articulatory effects on adjacent segments than less constrained ones. Acoustic data on the size of anticipatory and carry-over effects between [a] and [i] and several consonants bear out the model's prediction that the direction and extent of vowel coarticulation varies inversely with the degree of constraint of the intervocalic consonant. Sound changes observed in various Catalan dialects are also in agreement with the predictions of the model.
The contribution by Francisco Torreira provides experimental evidence for postaspiration in Andalusian voiceless stops, and suggests that the conditioning factor is the presence of a preceding /s/, usually realised as aspiration. The fact that a great deal of variation was found in the duration of aspiration and VOT values supports an Articulatory Phonology account (Browman & Goldstein Reference Browman and Louis1989) – based on vocal tract gestures – in which the timing of the glottal opening with respect to the supraglottal closure may not be accurately specified. As Torreira argues, the postaspiration facts would be more difficult to capture in a model based on features and/or segments.
Three of the papers in Part 2 deal with intonation. Intonation is one of the areas that has benefited most from the popularity of experimental methods and laboratory phonology itself, possibly because of the difficulty involved in collecting data using other methods (e.g. intuition). The papers on intonation are a good illustration of how experimental work can be used to develop and test phonological analyses.
Lluïsa Astruc-Aguilera & Francis Nolan carry out several experiments to study the prosodic characteristics of extrasentential elements; although extrasentential elements have been traditionally considered to be independent and deaccented, the results of the experiments reported in this paper show that there is a substantial amount of variation, with some trade-off between prosodic independence and tonal subordination. Extrasentential elements signal their peripheral role by being deaccented, or if accented, by compression in pitch range coupled with prosodic separation. Neither deaccentuation nor prosodic separation is obligatory. The article by Sónia Frota, Mariapaola D'Imperio, Gorka Elordieta, Pilar Prieto & Marina Vigário is a cross-linguistic study of intonational phrases across five Romance language varieties. The authors provide a typology of combinations of nuclear pitch accents and boundary tones; they also study the phonetics of boundary tones across the selected languages. Two main properties were found: (i) the five Romance varieties studied share a dominant boundary cue – the H boundary tone; (ii) they can be divided into two groups – the Catalan-Spanish group and the European Portuguese-Italian group – on the basis of pitch-accent choice and combinations of nuclear accent with continuation rise/sustained pitch configurations. Variation is found between the Catalan-Spanish and the European Portuguese-Italian groups. The phonetics of intonational boundaries offers a similar picture across the five languages. In the third paper, Marta Ortega-Llebaria & Pilar Prieto attempt to disentangle the acoustic correlates of stress from those of accent by examining duration, vowel quality, overall intensity and spectral balance in relation to stress, and the presence or absence of a pitch accent. The results of their experiments show that syllable duration, vowel-quality values and spectral tilt (rather than overall intensity) are strong acoustic correlates of stress in unaccented contexts (where pitch is flat), while they only constitute secondary cues of accent. Ortega-Llebaria & Prieto's article is an important contribution to the discussion on acoustic correlates of stress, as previous studies on the topic only examined stress in accented words, for which, not surprisingly, pitch is a strong correlate of stress.
The fourth paper on prosodic structure deals with syllable structure. Laura Colantoni & Jeffrey Steele examine voicing and manner asymmetries in stop+liquid clusters in Spanish and French by means of an experimental study. The hypothesis is that these asymmetries are due to similarity in manner and voicing, and stop+lateral clusters should thus exhibit no assimilation or dissimilation, since laterals are the more vowel-like of the liquids, and therefore more unlike the stops. In the case of stops+rhotics, the tap in Spanish is more like the stop than the fricative rhotic in French and thus dissimilation (epenthesis) is predicted. If the rhotic is less similar, as is the fricative in French, then assimilation should be possible. The results confirm these predictions. Devoicing of rhotic fricatives after voiceless stops in French is explained as the result of aerodynamic constraints against voiced fricatives. Length adjustments were also observed in the epenthetic vowel in Spanish as a compensatory effect: epenthetic vowels were shorter after the longer voiceless stops and vice versa. Length asymmetries in French occurred in the opposite direction to what was predicted. This is explained as the result of constraints on the production of the fricatives.
The implications for phonological theory of Colantoni & Steele's paper are somewhat difficult to discern, primarily because the hypotheses formulated crucially rely on rather controversial assumptions in phonology, which are treated somewhat superficially by the authors. For instance, that the laterals are more vowel-like than rhotics rests on the assumption that they are [+continuant]; although the authors admit that laterals have often been considered [−continuant] in Spanish (p. 113), they assume that they are ‘[+continuant] segments based on their acoustic characteristics’. Regardless of their acoustic characteristics, there is sufficient evidence that shows that Spanish laterals (at least in the coda) sometimes function as [+continuant] and sometimes as [−continuant], on articulatory and distributional grounds (Goldsmith Reference Goldsmith, Cressey and Napoli1981, Harris Reference Harris1984, Mascaró Reference Mascaró, Aronoff and Oehrle1984, Hualde Reference Hualde1989). A similar matter arises with respect to the tap, whose [−continuant] specification seems unlikely, as it is the trill (not the tap) that appears in the same contexts as [−continuant] obstruents (after pause, and after a homorganic obstruent) (Baković Reference Baković1997). One more controversial assumption is that voiced obstruents are underlyingly stops (p. 114), which also poses a problem for the OT formalisation presented, given that candidate evaluation in OT is based on output forms. In sum, it would be useful to give some consideration to how the experimental results should be interpreted under different phonological assumptions for laterals, taps and voiced obstruents in Spanish.
Part 3 contains papers on the acquisition of segmental and prosodic contrasts by first and second language learners. Maria João Freitas studies the acquisition of the unstressed allophones of the vowels /a e ε/ in a corpus obtained from three European-Portuguese speaking children ages 0;10–2;8 over the course of one year. This paper is another excellent illustration of how laboratory phonology serves to test the predictions of phonological theory. In particular, the research described here is relevant for the debate on the nature of early phonological representation: if the children's outputs match their targets, abstract phonological representations may not be present; however, if there are mismatches, some early phonological representation is probably under development. Furthermore, studying how children deal with allophony permits the researcher to test the prediction that the complexity of the allophonic/allomorphic alternations in the native language may enhance the rate of phonological development. The study suggests that children start acquiring vowel reduction early and that the complexity of the Portuguese systems is probably the cause of this early mastery of phonological representations. More specifically, it is shown that featural complexity has an impact on acquisition: representations involving one featural node become stable before those involving more than one node, i.e. the acquisition of /a/→[ɐ] preceded /ε e/→[ɨ]. Against the author's predictions, word-final vowels were acquired before or simultaneously with word-internal ones, probably due the presence of morphological content.
Ferran Pons & Laura Bosch focus on the perception of lexical stress patterns by Spanish and Catalan-speaking infants. Unlike the results reported in the literature on American English infants, no preference was found for a trochaic pattern. The researchers attribute this to a weaker predominance of the bisyllabic trochee in Spanish and Catalan than in English. As a result, this type of cue alone may not be as helpful in early word-segmentation in Spanish and Catalan as it is in other languages. It is hypothesised that infants may learn at quite an early age to rely on other factors for segmenting fluent speech and that one of the factors meriting further investigation is syllable weight/phonotactics. One aspect that is missing in this paper is a review of the theoretical literature on the stress system of Spanish, including the controversy surrounding whether Spanish is a truly a weight-sensitive language or, alternatively, whether the apparent weight-sensitivity effects (three-syllable window, non-branching condition; Harris Reference Harris1983) are simply the result of historical gaps (Roca Reference Roca1988, Hualde Reference Hualde2005). A review of these issues could help place the research in a context, making the results – foot form is not the only cue for segmenting early speech – more meaningful. An important implication of this paper's findings for phonological theory is that syllable phonotactics (i.e. syllable weight) may play a role in the acquisition of stress and word-segmentation cues in infants; it remains to be determined whether this is the result of true weight-sensitivity or word-frequency effects or a combination of both.
In ‘Rhythmic typology and variation in first and second languages’, Laurence White & Sven L. Mattys report on the results of three experimental studies designed to determine the discriminative performance of various speech-rhythm metrics with regard to the distinction between syllable-timed and stress-timed languages. The experiments also study the effect of L1 on L2, when L1 and L2 belong to different rhythm classes, and between different dialects of the same language (British English). Rate-normalised metrics of variation in vocalic interval duration (%V and Varco V) emerge as the most discriminatory metrics among groups. The degree of rhythmic variation found within dialects of British English also suggests that stress-timed vs. syllable-timed may be a gradient (non-categorical) distinction, thus raising the issue of accounting for gradient phenomena in phonology and how non-categorical effects may be integrated into the phonological grammar.
Finally, in an article with a somewhat different goal than the rest of this volume, Geoffrey Stewart Morrison offers a tutorial for L2 perception students and researchers who are not familiar with logistic regression analysis. He illustrates the application of logistic regression analysis to categorical response data sets from experiments on L1 Spanish vowel perception and on L1 and L2 English vowel perception in which listeners were asked to identify synthetic stimuli in terms of speech-sound categories. He also demonstrates the use of logistic regression coefficients to describe listeners' perceptual use of acoustic cues and to produce detailed graphic representations of such cues. An additional use of logistic regression coefficients shown in the tutorial consists of determining if there are significant differences in the perception of stimuli by L1 and L2 listeners. Given the frequency of use of categorical response data with synthetic stimuli in L2 speech-perception research, logistic regression analysis should be of significant use to this field of research.
Overall, Segmental and prosodic issues in Romance phonology is very successful in accomplishing the goals set up by the editors in their introduction. The volume as a whole offers a broad cross-linguistic basis to test theoretical claims, with many individual papers presenting data and analyses of several Romance languages. It also succeeds in bridging the gap ‘between traditional Romance linguistics … and laboratory phonology … subjecting the theoretical claims and data from traditional Romance linguistics to the scrutiny of experimental techniques in the laboratory’ (p. viii). As I mentioned above, some papers actually constitute excellent exemplifications of research in laboratory phonology, providing very comprehensive background sections on the phonological issues at stake and on how their experimental studies relate and contribute to differentiating between different proposals (e.g. Nguyen et al., Freitas). In fact, one is left wishing that more papers contained background sections such as these, to the extent that, if one had to point out a weakness in this volume, it would be precisely the need for more of those sections. While the emphasis on ‘laboratory’ (vs. ‘phonology’) may be a reflection of the field and a matter not entirely under the control of the editors, the need for ‘bridging the gap’ and reaching out to the more traditional side of Romance phonology would justify a call to laboratory phonologists (especially those dealing with segmental, metrical and syllabification topics) to include introductory/background sections to help Romance phonologists in general to gain a better understanding of the relevance of experimental data for phonological theory. On a similar note, more discussion of controversial phonological questions (e.g. the featural composition of the tap and the lateral in Spanish, Spanish as a weight-sensitive language, etc.) would contribute to establishing stronger ties between the fields of traditional and laboratory Romance phonology.
Regarding organisation and thematic unity, it must be said that this is a very well-organised volume, with a balanced distribution of topic areas and good thematic cohesion, something which is not easy to accomplish with conference paper selections. It is also an example of extremely careful editing, with only a handful of typographical and stylistic errors.
In summary, the volume is an excellent compilation of papers on Romance laboratory phonology, recommended for researchers and students in laboratory phonology and phonetics, as well as for traditional Romance phonologists.
