Publishing Partner: Cambridge University Press CUP Extra Publisher Login

New from Cambridge University Press!


Revitalizing Endangered Languages

Edited by Justyna Olko & Julia Sallabank

Revitalizing Endangered Languages "This guidebook provides ideas and strategies, as well as some background, to help with the effective revitalization of endangered languages. It covers a broad scope of themes including effective planning, benefits, wellbeing, economic aspects, attitudes and ideologies."

New from Wiley!


We Have a New Site!

With the help of your donations we have been making good progress on designing and launching our new website! Check it out at!
***We are still in our beta stages for the new site--if you have any feedback, be sure to let us know at***

Review of  The Phonology of Welsh

Reviewer: Jean-François R. Mondon
Book Title: The Phonology of Welsh
Book Author: S. J. Hannahs
Publisher: Oxford University Press
Linguistic Field(s): Phonology
Subject Language(s): Welsh
Issue Number: 25.4632

Discuss this Review
Help on Posting
Review's Editor: Anthony Aristar


This book presents the phonology of Welsh, working through its various levels from the segmental to the prosodic, focusing throughout on productive phonological processes. As Hannahs states at the outset, the book aims for a descriptive, theory-neutral presentation of the facts but his conviction of the appropriateness of Optimality Theory (OT) as a framework for investigation compels him to adopt the model from chapter 4 on.

Chapter 1 “Introduction and Background” starts off by laying out the assumptions adopted in the book. To begin with, since Welsh has no spoken standard, the obvious question arises as to what form of Welsh is being investigated. Hannahs assumes, correctly in my opinion, that there is a shared “abstract underlying phonological system that allows us to speak of the phonology of Welsh as relatively unitary, and that dialect variation is primarily a question of the various ways in which the dialects implement the phonology” (pp. 2-3). His second assumption is that OT can not only present a nice template upon which to work but that even more importantly if offers a way of discovering hitherto unseen connections. This is particularly true with respect to the role of foot structure as Hannahs discusses in chapters 4 and 5. Contra the work of others, Hannahs focusses on a strictly synchronic analysis, stating that it offers a new understanding which a strictly diachronic analysis simply cannot. This focus does not shy Hannahs away from littering throughout the book the origin of various synchronic phenomena. The chapter concludes with a discussion of the history of the language and its modern dialectal situation.

Chapter 2 “A Survey of Welsh Phonetics” works through the various layers of the sound system of the language. It begins by mentioning those major phonetic traits which are only true of some dialects, such as the southern merger of /i/ and barred-i as well as the absence of /h/, voiceless nasals and [rh] also from southern varieties. The interesting data Hannahs highlights with respect to the stops include the aspiration of voiceless stops everywhere but word-finally, the germination of voiceless stops following a stressed vowel, and the half-voicing of voiced stops except when in a fully voiced environment. As for the other obstruents, affricates have arisen through the palatalization of [t, d] by [i] and the borrowing of English loan words such as ‘jam’. Their incorporation into the mutation system indicates that they have integrated into the sound system (this is discussed further in chapter 6). Fricatives cover the whole gamut of the vocal tract with uvular [χ] in the North to [ʒ] in borrowings. Striking is the variable deletion of [v] and [ð] word-finally as in ‘ffordd’ (road). With respect to the vowels, the inventory of monophthongs varies from dialect to dialect (Jones (1984) records 11 and Fynes-Clinton (1913) 13 for instance). Hannahs, therefore, focusses on the basic vowel inventories of the North and South. Interestingly, there exists a length difference which is contrastive only in stressed monosyllables. Unsurprisingly, the number of diphthongs differs from region to region, with the South exhibiting eight and the North exhibiting 13 since barred-i can serve as an off-glide there in addition to the normal high offglides.

Chapter 3 “Welsh Phonological Structures” begins the journey into the super-segmental structure of the language. To begin with, Welsh possesses a minimality condition mandating that a monosyllabic content word must be bimoraic. In the case of open syllables with a monophthong, lengthening occurs. Lengthening does not occur in monosyllables, however, when the coda consists of two or more consonants or of /p, t, k, m, ŋ/ alone (Williams (1989), Wood (1988)). These five consonants appear to count as moraic in coda position. This assumption is validated by the inability of a monosyllabic word which ends in one of these five consonants from having a diphthongal nucleus since a superheavy trimoraic syllable would result. Apart from two native words ‘cewc’ (glance) and ‘twym’ (warm) this appears to be true. With respect to phonotactics Welsh complex onsets and codas largely follow the sonority hierarchy. Other syllable-related impositions include the inability of schwa to occur in word-final syllables, the interchange between [i] and [u] with the glides [j] and [w] respectively, and the ability of sonorants in some dialects to be the nucleus. Finally, the morphophonology of the definite article seemingly prioritizes codas over onsets, opting for the variant [r], which is an enclitic on a preceding vowel-final word, over the variants ‘yr’ or ‘y’, yielding “o’r llyfr” (from the book) instead of “*o y llyfr” without an extra coda (cf. Hannahs & Tallerman 2006). The chapter delves next into a discussion of stress. Stress is predictably penultimate with the head of the rightmost trochaic foot bearing main stress. Certain exceptions do exist, though, primarily due to loanwords or to specific suffixes, such as ‘-(h)ad’. Strikingly, the stressed penultimate syllable bears the stress accent while an independent pitch accent falls on the final syllable. The foot and prosodic word round out the chapter, but since they serve as the centerpiece of subsequent chapters the discussion is brief.

Chapter 4 “Phonological Processes” begins with a discussion of the phonological position of schwa. Hannahs concludes that it is a phoneme with positional allophones (i.e. schwa and barred-i) and at the same time it is also a positional allophone of [u]. Next Hannahs tackles vowel mutation in which certain vocalic nuclei alternate between final-syllable form and non-final syllable form (orthographically: ‘ai, au, aw, uw, w, y’ alternate with ‘ei, eu, o, u, y, y’). Since they are not productive processes but largely diachronic artifacts, he limits the discussion to the monophthongs which are more generally regular. He lays out an OT account, revising his earlier work (Hannahs 2007) taking into account results of Green (2007). Hannahs proposes that words with non-alternating barred-i have barred-i as inputs while words with alternating barred-i have schwa as their input. He grounds these assumptions in learnability. With respect to constraints he needs only *Schwa-Final-Syllable, which militates against schwa in final syllables, and Ident-IO-vowel features. The chapter concludes with short discussions of the mainly diachronic phenomena of vowel affection and assimilation.

Chapter 5 “Foot-Based Phenomena” really shows the advantages of Hannahs’ use of OT. He begins by analyzing word-final clusters which do not follow sonority sequencing, such as ‘pobl’ (people) and ffenestr’ (window). What is of interest is how these final clusters are handled by the phonology. In some an epenthetic vowel is inserted (pobl), in others, the final consonant is deleted (ffenestr) and in still others metathesis occurs (‘ewythr’ (uncle)). Hannahs shows that the first group undergoes epenthesis in order to create a binary foot. The quality of the epenthetic vowel is determined by ‘BE-Ident-F’ adopted from Kitto and de Lacy (1999) which allows for the epenthetic vowel to copy the vowel quality of the base. The second group deletes the final consonant in order to maintain a binary foot. Finally, the small third group is limited to the cluster ‘thr’ but it too preserves a binary foot. Hannahs then moves onto the realization of /h/. It was long ago realized that /h/ surfaces only when the onset of a stressed syllable or of a word. Hannahs, argues, however that a simple constraint stating that [h] must be foot-initial fails to account for why Welsh has no other glottal sounds. For this reason he adopts the constraint *Glottal as well as a positional faithfulness constraint Max-initial-h where /h/ occurs word-initial and foot-initial. Finally, the chapter closes with a discussion of antepenultimate stress.

Chapter 6 “Initial Consonant Mutation” breaks the pattern of the immediately preceding chapters and focusses on a topic which is more morphological than anything. Hannahs rejects a purely phonetic or phonological analysis of mutations and adopts in spirit Green’s (2006, 2007) solution, in which for any given lexical item, the radical form is linked to its various mutated forms in the lexicon. Hannahs’ ‘pattern extraction’ analysis crucially differs from Green in maintaining that the link between the radical and mutated forms does not consist of full lexical items but of the initial consonants only. Additionally, since no lexical material is involved in the linking, the extracted pattern of consonantal links is not limited to the lexicon per se, especially since some phonological subregularities exist between the effects of various mutations (e.g. soft mutation changes voiceless stops into voiced stops while aspirate mutations renders them voiceless fricatives).

Chapter 7 “Remaining Issues and Further Directions” touches on two topics which would benefit from future research: provection and compounding. Provection is the ‘hardening’ of voiced stops into voiceless stops. Two types of provection occur in Welsh: one is the change of a voiced stop after a stressed syllable in some dialects (Thomas 1988) and the other is at morpheme boundaries, such as before superlative -haf. With respect to compounding, Welsh exhibits strict compounds on the one hand in which vowel mutations occur across members and only one stress surfaces, versus loose compounds on the other hand in which mutation is barred and both members retain their own stress. Which variables (be they morphological, phonological, or semantic) are involved in the division of Welsh compounds is yet to be determined.


This book does a superb job at covering the main topics which have been at the forefront of the study of Welsh phonology over the past half-century. Its ample bibliography and descriptive precision make this a suitable starting point for future research. In particular aside from chapter 7, it is littered with open questions throughout. To give but two examples, in chapter 3 (p. 37) Hannahs discusses an odd pair of facts from northern Welsh. Long vowels are licit when before the coda cluster ''llt'' (also before sb, sg, st) but when the coda consonant is ''ll'' alone, a long vowel is not permitted. How can ''ll'' be moraic when alone, but non-moraic as part of a cluster? As a second example, Hannahs concludes chapter 5 (pp. 115-119) with an analysis of antepenultimate deletion. In some trisyllabic words, the antepenultimate syllable may be variably deleted (‘[y]sgolion’ (schools)), while in others such deletion is not possible (‘hanesion’ (tales)). Hannahs proposes that for the former the antepenultimate syllable is unfooted and attached directly to the prosodic word level. In the latter, on the other hand, the antepenultimate syllable is attached to a foot to create a superfoot, which affords it protection from variable deletion. It is this difference in level of attachment which accounts for whether or not a syllable is variably deleted or not. While this solution accounts for the difference it by no means explains why the vocabulary of trisyllabic words should be bifurcated the way they are. Is this just a diachronic artifact or is some as of yet unforeseen trigger the cause? One final area for future research is with respect to the discussion of the realization of /h/ in chapter 5. Hannahs adopts the constraint Max-Initial-h where /h/ is word initial or foot initial. Such listing of contexts seems to me to be missing something. What though, if anything, is the question.

Fynes-Clinton, O. H. (1913). “The Welsh Vocabulary of the Bangor District”. London: Oxford University Press.

Green, Antony. (2006). “The independence of phonology and morphology: the Celtic mutations,” “Lingua” 116: 1946-85.

Green, Antony. (2007). “Phonology Limited”. Potsdam: Universitätsverlag.

Hannahs, S.J. (2007). “Constraining Welsh vowel mutation,” “Journal of Linguistics” 43.2: 341-63.

Hannahs, S. J. and Maggie Tallerman. (2006). “At the interface: selection of the Welsh definite article,” “Linguistics” 44.4: 781-816.

Jones, Mari C. (1998). “Language Obsolescence and Revitalization: Linguistic Change in Two Sociolinguistically Contrasting Welsh Communities”. Oxford: Oxford University Press.

Kitto, Catherine and Paul de Lacy. (1999). “Correspondence and epenthetic quality,” “Proceedings of AFLA” 4: 181-200.

Thomas, Siân Elizabeth. (1988). “A study of calediad in the upper Swansea Valley,” in “The Use of Welsh” (ed. M. Ball): 85-96. Clevedon: Multilingual Matters.

Williams, Briony. (1989). “Stress in modern Welsh”. PhD dissertation, University of Cambridge.
Jean-François Mondon is an assistant professor of Foreign Language at Minot State University. His research interests include Classical Armenian, Celtic Linguistics, and Distributed Morphology.

Format: Hardback
ISBN-13: 9780199601233
Pages: 208
Prices: U.K. £ 55