This is the first volume that is specifically dedicated to the study of geminates, looking closely at their phonetic manifestations and phonological representation. There have been smaller-scale Ph.D. dissertations that were published on this topic in the past (e.g. Aoyama Reference Aoyama2001, Ham Reference Ham2001, Isei-Jaakkola Reference Isei-Jaakkola2004), but this book provides a more comprehensive coverage of this topic and an interesting interdisciplinary approach. There is cross-linguistic coverage of geminate consonants but a heavy focus on Japanese for understandable reasons, given the editor’s own work in this area. The book is suitable for intermediate-to-advanced students in linguistics, and researchers interested in geminates and the relationship between phonetics and phonology more generally.
The book consists of fourteen chapters divided into two equal parts, the first one covering studies on the production and perception of geminate consonants and the second covering the phonological representation of geminates. In the first chapter, Shigeto Kahawara & Melanie Pangilinan investigate perceptual reasons behind languages favouring obstruent over sonorant geminates, and demonstrate a graded perceptual difficulty within sonorants. Obstruents have more clearly distinguished spectral properties from their surrounding vowels, while sonorants share more of these properties, particularly glides. This leads to more blurred boundaries between them, making it more difficult to perceive duration changes. In a series of identification and discrimination experiments using non-speech stimuli mimicking obstruents and sonorants, and manipulated for duration, the authors highlight a further element which contributes to a greater difficulty in perceiving duration in glides than in nasals, that of reduced amplitude changes between glides and surrounding vowels. This is thought to lead to the hierarchy in preference for obstruent over sonorants cross-linguistically, and for nasals over glides with the sonorant category. As the listeners were native-English speakers, further evidence from speakers of languages with gemination and from real speech would be welcome.
Olga Dmitrieva picks up on the importance of sharp edges between geminates and their neighbouring sounds in Chapter 2 and reflects on how this may have shaped where geminates are found within the prosodic word. The author investigates the latter point in a production study of Russian geminates looking at how segmental and prosodic context influences geminate duration. Two production experiments – one focusing on geminates in real speech and the other on singleton and geminate consonants in nonsense words – show systematic differences in the duration of geminates in initial, intervocalic, final and pre-consonantal position, in pre- and post-stressed position; intervocalic stressed geminates are the longest, and notably the most common cross-linguistically. The ratio of geminate-to-singleton consonants is shown to be important too, as a high ratio can potentially explain why sonorants and voiced obstruents retain their salience. Interesting language-specific differences emerged as well, such as longer final than initial geminates despite the higher instances of gemination in the former.
Rachid Ridouane & Pierre Halle focus more closely on initial geminates in Chapter 3, exploring how voiceless obstruents in particular manage to retain their singleton–geminate contrast when closure duration is typically the main cue for the geminate contrast in most languages. Using production data from Tashlhiyt Berber, they show that acoustic cues for voiceless stops are indeed reduced in initial position, and that non-durational cues such as relative intensity of the release serve as important alternative cues which are used by listeners, especially in prosodically focused contexts. Listeners still perform worse on the perception of initial voiceless geminates than voiced ones or fricatives, and the authors reflect on how weak acoustic cues may lead to a change in the phonological representation of a contrast, leading to [tense] as the main feature.
Chapter 4 focuses on a very interesting language-specific feature of Japanese whereby closed syllables lead to vowel lengthening rather than the typical Closed Syllable Vowel Shortening (CSVS) effect observed in many other languages (e.g. Maddieson Reference Maddieson1985), in order to preserve the isochrony between syllables. Hajime Takeyasu & Mikio Giriko show how this pattern plays out in geminate environments, which exhibit varying influences of phonetic and phonological length in the way Japanese speakers produce and perceive CV(V)C(C)V. In a perception experiment manipulating the duration of the medial consonant (C2) and the preceding vowel (V1), the authors asked Japanese listeners to decide on the phonological length of each of V1 and medial C; they found that V1 was more often judged as long when C2 duration was long, and C2 as geminate when V2 duration was long. Conversely, C2 was less often judged as geminate when V2 was perceived as long. A production experiment replicated those results and demonstrated that Japanese exhibits a CSVL (Closed Syllable Vowel Lengthening as opposed to CSVS) which stands even in geminate environments, contrary to results reported for other languages with phonological vowel and consonant length (e.g. Khattab & Al-Tamimi Reference Khattab and Al-Tamimi2014); on the other hand, the avoidance of superheavy syllables is exhibited in the asymmetrical results for C2, which is less often judged as a geminate if V1 is perceived as long. The study shows a great interplay between phonological structure and phonetic implementation in the way speakers and listeners process speech.
In Chapter 5, Anders Lofqvist examines interarticulator timing differences between Japanese and Italian with the view to test for the moraic timing that is claimed for the former, but finds little evidence for that. In a kinematic study looking at lip and tongue movement timing in CVN(N)C words in both languages, the author finds substantial variability within and across speakers and no reliable results for language-specific timing of vowel–consonant sequences between the two languages as was suggested by Smith (Reference Smith, Connell and Arvaniti1995), who claimed that mutual co-ordination of VC in Japanese as opposed to consonants was superimposed on vowels in Italian.
Chapter 6 is one of two chapters in this book to deal with acquisition. Elinor Payne, Brechtje Post, Nina Garmann & Hanne Gram Simonsen look at Norwegian children’s gradual adult-like acquisition of absolute and relative durations of VːC and VCː sequences in mono- and multi-syllabic words. The study employs a wide age range (from two-and-a-half years to six years) and shows that children distinguish VːC and VCː rhymes from an early age and perform particularly well at the proportional duration of the vowel in the V(ː)C(ː) rhyme. Producing adult-like durations for each of (Vː) and C(ː) takes long to perform and shows an interesting U-turn for the six-year-olds, which the authors put down to children becoming more fluent over time and better at intergestural co-ordination at the expense of more local segmental sequences that they had produced accurately at an earlier age. U-turns are common in phonological development and have been reported in the acquisition of gemination (e.g. Khattab & Al-Tamimi Reference Khattab, Al-Tamimi, Vihman and Keren-Portnoy2013). U-turns show the outcome of children’s focus on different skills at a time and/or of their phonological reorganisation of the input.
Yukari Hirata shows some parallels between these results and the acquisition of gemination by second language (L2) learner adults in Chapter 7 (e.g. large variation within and across individuals), but with a marked difference in the way children and adults approach this task. Adult intermediate L2 learners of Japanese managed to produce a reliable singleton-to-geminate contrast in a production task when C duration was looked at, but their proportional durations (C/V1, C/word) were less target-like, showing more-segmental than prosodic learning compared with children, and expected interference from their L1 patterns. These two studies provide and interesting insight into how children and adults differ in the size of the prosodic units that they process when learning a language.
The second part of this volume is entitled ‘Phonological representation of geminates’ and starts with a study by Sandra Kotsor, Allison Wetterlin & Aditi Lahiri using behavioural and neurolinguistic evidence to find out if Bengali singleton and geminate consonants are better represented using Skeletal or Moraic Theory (e.g. Ringen & Vago Reference Ringen, Vago, Cairns and Raimy2011 vs. Hayes Reference Hayes1989). Singleton consonants under an X-slot representation have one timing unit while geminates have two; in Moraic Theory, however, phonological length is privative and geminates have two timing slots while singleton consonants are unspecified. The investigators shortened geminates and lengthened singletons before using them as primes in a lexical decision task or as stimuli in an ERP study. The asymmetry with which participants accepted lengthened singletons but had difficulty processing shortened geminates provided evidence for a Moraic representation of Bengali geminates. Lara Ehrenhofer, Adam Roberts, Sandra Kotzor, Allison Wetterlin & Aditi Lahiri follow that up with a similar study on Swiss German in Chapter 9, replicating the Kotsor et al. results and providing further evidence for the asymmetrical representation of timing. As Swiss German allows geminates in all word position (as opposed to only word-medially in Bengali), the authors discuss implications for a two-step processing of timing in languages with geminates, one that accesses a CV template and another that accesses timing.
Stuart Davis provides further evidence for a moraic account of gemination in Chuukese in Chapter 10. Drawing on a collection of alternations such as the deletion of word-final short vowels, the shortening of word-final long vowels, compensatory lengthening and geminate throwback, the author uses an Optimality Theory (OT) analysis to demonstrate how these alternations in Chuukese provide convincing evidence for the underlying moraicity of geminates. In chapter 11, Nina Topinzi & Stuart Davis present data on edge geminates (initial and final) from a large number of languages and compare these with edge clusters, showing that (i) edge geminates in languages with edge clusters may be non-moraic (but they argue that the majority still are); (ii) where edge clusters are moraic, geminates are, too; and (iii) where geminates are weightless, clusters are, too. The authors consider whether there are two types of geminates, a moraic and a non-moraic type, potentially solving the long-standing debate in phonology about whether geminate representation is weight-based or length-based, but warn that their data are preliminary and more typological data are required.
In Chapter 12, Junko Ito, Haruo Kubuzono & Armin Mester tackle loanword adaptation and carry out a thorough OT analysis of English loanwords in Japanese. Using markedness constraints that govern word structure in Japanese and Faithfulness constraints that preserve certain aspects of final consonant production in English, the authors demonstrate how both types of constraints are needed to explain the patterns of adaptation found in English loanwords. Similarly, Shin’ichi Tanaka looks at the loanword adaptation of Italian geminates in Chapter 13, providing an interesting context for investigating adaptation from a language that also does geminates. Given that perception of gemination in itself is not the expected source of the adaptation, the author shows that perception and production of Italian geminates by Japanese participants are driven both by phonological rules of the native language (L1; e.g. manner of articulation effect) and perception constraints of Italian geminates that are driven by the prosody of the language. This demonstrates the combined role of phonological representation and acoustic patterns of perception in processing geminates in another language. In Chapter 14, on the other hand, Hyunsoon Kim presents perception data which suggest that Seoul Korean learners of Japanese process their L2 through their L1 system and are influenced by their L1 perception, lexicon and phonology. Participants listened to Japanese voiceless geminate obstruents and perceived them as either a sequence of a coda and an onset or just an onset consonant due to constraints on Korean syllable structure.
Altogether, the papers in this volume offer a multifaceted view of geminates using insights from perception, production and learning, and state-of-the-art methodologies. The cross-linguistic coverage enhances an appreciation of the diverse manifestation of geminates and their varying roles in the phonologies of different languages. While the book is presented in two parts, one entitled ‘Perception and production of geminates’ and the other ‘The phonological representation of geminates’, there is crossover between the two. Another advantage of this volume is that it presents experimental work and theoretical debates around geminate representation side-by-side throughout. The laboratory phonology approach adopted in the majority of the chapters is ideal for the treatment of this class of consonants which are a rich source for the investigation of the relationship between phonetics and phonology.
