EDITOR: Carnie, Andrew TITLE: Formal Approaches to Celtic Linguistics YEAR: 2011 PUBLISHER: Cambridge Scholars Publishing
Joseph W. Windsor, Dept. of Linguistics, University of Calgary
SUMMARY ''Formal Approaches to Celtic Linguistics'' is foremost an important contribution to linguistic knowledge in that it highlights constructions often not widely available in other languages. Celtic facts treated here have consequences for theoretical linguistics across subfields and on to interfaces and information structure. Some of the articles defend specific theoretical traditions, and others challenge notions of the phonetics/phonology divide between categorical and gradient.
The book grew from the 2009 Formal Approaches to Celtic Linguistics Conference (FACL) in Arizona. As such, it reads much like a proceedings, except that it has been highly edited and revised. The contributions have thus benefited from peer feedback by a group of specialists working on the same language family. The goal of this book is to serve as both a reference work in Celtic linguistics and a launching pad for future investigation.
I now proceed with an article-by-article summary of the book's contributions.
Part 1: Papers in Syntax, Morphosyntax and Semantics.
The first contribution to the book is David Adger's ''Clefted situations: A note on expletives in Scottish Gaelic clefts.'' This paper is one of the two contributions not originally presented at the FACL conference. This article examines the difference in syntactically clefted elements, [3rd.sg.m] ''it'' and [3.sg.m] ''in,'' based on the semantics of predication. The cornerstone of the argument is that: ''the syntax of clefts in Gaelic is built on the syntax of predication, and predication is mediated by functional structure that introduces the relevant semantic relation'' (p. 14). By linking semantic predication to syntactic principles, Adger is able to make a crucial differentiation between the two clefts which, if left analyzed in parallel, would be otherwise identical and would prevent a predictive theory to capture their distribution.
''The interaction of linearization and prosody: Evidence from pronoun postposing in Irish,'' written by Emily Elfner is the second contribution. This article offers an Optimality Theoretic (OT, Prince & Smolensky 2004) explanation of the multiple grammatical locations for a weak object pronoun to appear in a construction. The article looks at both syntactic and prosodic constraints offering a predictive explanation for the possibility of <é> [3rd.sg.m] ''it'' appearing in any number of positions in a sentence such as:
1) Léigh Liam (é) ar an traein (é) aréir (é) Read Liam (it) on the train (it) last night (it) 'Liam read it on the train last night' (adapted from Elfner's (2) p. 18)
In ''The syntax and development of the Old Irish autonomous verb,'' Jenny Graver gives an LFG account of what she argues are three syntactically distinct forms of the Old Irish autonomous verb: the canonical passive, the impersonal passive, and the active subject impersonal construction. The author gives evidence that the differences between the three structures are reliant on person and number features as well as argument structure. She also motivates syntactic change through the Middle Irish period appealing to syntactic markedness.
A second paper on Old Irish is ''Old Irish pronouns: Agreement affixes vs. clitic arguments'' by Aaron Griffith. In this paper, the author distinguishes between the person and number agreement morphology on verbs, NPs and prepositions as agreement morphology on a root vs. cliticized arguments. Griffith builds on previous analyses attempting to account for the existent data we have on record from the Old Irish period to ultimately argue that while affixed pronouns were most likely clitic arguments in a proto-version of the language, by Old Irish, they no longer have referentiality, and are therefore best analyzed as agreement affixes, and not arguments.
Next, there are two papers on Breton; the first is ''Some Breton indefinites'' by Randall Hendrick. This paper looks in depth at the distribution of indefinite clitic arguments in Breton and compares the scope and distribution of the negative and affirmative indefinites finding them not to have the same restrictions. This process is briefly compared to the similar forms in Welsh to strengthen the position.
The second Breton paper is ''Post-syntactic excorporation in realizational morphology: Breton analytic tenses'' by Mélanie Joultteau. The author identifies a precise environment for excorporation of Breton verbs and inflectional morphemes. She looks at the interface and examines the predictions that would be made by both distributed morphology as well as minimalism to provide strong evidence that this phenomenon is post-syntactic, but pre-phonology.
''The shape of Irish clauses'' by James McCloskey is the second of the papers not presented at FACL. This paper endeavors to answer the broad theoretical question of whether Generative linguistics is on the right track of providing us a deeper psychological understanding of language rather than simply spinning the self-validating wheels of theory. In order to answer this question, McCloskey looks at Irish clauses representative of the construction in (2):
McCloskey uses a battery of syntactic tests as applied to the Irish clauses to argue that structure provided by Generative theory in fact not only provides a descriptively adequate account of the phenomenon, but also a predictive theory that can explain the structures without the introduction of additional machinery.
Returning to pronoun postposing in Irish; Ann E. Mulkern's article ''Left right behind: Irish pronoun postposing and information structure'' briefly touches on the work done by Elfner (same volume) which identifies pronoun postposing as an effect of linearization and phonological constraints; but argues that a full account of this phenomenon must also interface with information structure, specifically cued by giveness and newness.
''Some problems with object enclitics in Literary Welsh and Old Irish'' is Máire Noonan's article which tackles the problem of unifying the analyses of previous authors (Sadler 1988; Newton 2006, 2008) for Literary Welsh and Old Irish respectively. The phenomena under the microscope in this article are the facts in the two languages of object enclitics wherein Old Irish there is claimed to be T-to-C movement where C is not otherwise filled, and Literary Welsh makes use of an AGREE relationship. Differences between the languages are left to parameterization effects and largely accounted for by the Phase Impenetrability Condition (Chomsky 2008).
Kenji Oda's ''Preverbal particles and irregular verbs in Irish'' looks at an issue that has been receiving attention in recent years: the irregular relationship between preverbal particles which indicate negation or questions in Irish, and the eleven irregular verbs. This is investigated through a distributed morphology framework and argues that the distinction between dependent and independent verbs and their particles is the number of tense features these verbs are merged with. This accounts for what appears to be tense mismatch in the particles' paradigms.
In ''The semantics of Scottish Gaelic tense and aspect'' by Sylvia L. Reed, Reed argues for a distinction between 'simple' and 'state-relational' viewpoint aspects which would divide the aorist and imperfective into the former, and perfect, after-perfect, and prospective into the later. This article builds on several arguments for distinction raised by previous authors, but contends these previous approaches cannot account for all language data. This division makes the prediction that since there is contrasting morphosyntactic patterns, in languages that overtly mark various aspects morphologically, a division will be clearly seen. Scottish Gaelic provides this testing ground.
Part 2: Papers in Phonology, Morphophonology and Phonetics.
This section begins with a joint paper by Diana Archangeli, Jeff Berry, Sunjing Ji, Keisha Josephs, Nicole Hunt, Muriel Fisher and Andrew Carnie entitled ''ATR in Scottish Gaelic tense sonorants.'' This paper discusses the featural difference of [±ATR] in Scottish Gaelic sonorants by investigating tongue positions of a speaker of the Skye dialect using ultrasound imaging. This study adds phonetic evidence for a featural difference between tense and lax sonorants which were theoretically proposed by Carnie (2002) for Irish.
Another broadly theoretical paper which uses Celtic as a testing ground is Anna Bosch's ''Transcription: The phonetics/phonology interface.'' This paper investigates several phonetic grammars of Scottish Gaelic with particular attention to intervocalic hiatus and epenthesis to highlight in the importance of unpacking the intention behind transcription before basing a full phonological analysis on any one given record of speech. The various corpora that Bosch investigates all make mention of the phenomena under the spotlight, but each transcribes as well as describes it in a different way, highlighting the need for this consideration in future investigations of any language.
Looking at the phonology-morphology interface, Colin Gorrie's ''Adjective agreement in Gaelic: A case for morphophonological features'' argues in favor of a distinction between morphological features, morphosyntactic features, and his personal contribution: morphophonological features. Since under many assumptions both the morphosyntax and phonology separately undertake their operations based on features; and because the initial mutations of the Celtic languages seem to be a phenomenon that bridges this interface, Gorrie uses this as evidence for features that transition Spellout to Phonological Form (PF) and effect the phonology directly.
Michael Hammond takes a new look at probabilistic patterns of Welsh mutations in his contribution ''Welsh mutations and statistical phonotactics.'' This paper tackles a broad theoretical problem in the literature, namely whether there is a division between the phonetics and the phonology where the former is concerned with the gradient and the latter deals only with the categorical. By examining phonotactics that result from mutations in Welsh, Hammond argues that the gradient must also be at work in the phonology.
The final article is S. J. Hannahs' ''Unity in diversity in Welsh: The avoidance of sonority sequencing violations'' which provides an OT account of repair strategies for underlying consonant clusters in Welsh words that violate the Sonority Sequencing Principle. The key contribution of this article is in demonstrating an advantage of OT in that the diverse repair strategies demonstrated in Welsh for this high-ranking violation are actually. Also, by showing the various repair strategies as all resolving the same conflict, Hannahs demonstrates that the various phenomena he discusses are all in fact one and the same problem - something neglected in previous analyses of the individual phenomena.
EVALUATION On the whole, I am glad that this collection has been published. Regardless of the reader's theoretical background, all of the articles serve to encourage further scholarship on these languages.
Turning to the articles themselves, I unfortunately do not have the space here to comment on all of them, but some points stand out after a careful reading of this book. Looking at the two articles which look at pronoun postposing in Irish: Elfner and Mulkern, the latter claims that the former is a good indicative account of the phenomenon, but not explanatory enough. Mulkern's article relates the phenomenon of the rightward displacement of object pronouns in Irish to Information Structure and referential qualities of giveness and newness. Presumably, sentential stress comes from some sort of interface constraint that demands new discourse information be stressed prosodically. Both authors contend that the postposed pronoun will cliticize onto a phonological phrase that bears main stress. The difference between the two is that while Elfner is content to identify the adjoinment site of the pronoun as the stressed phonological phrase, Mulkern says it must be the phrase that is stressed due to a newness constraint imposed by Information Structure. Neither actually disagree with the other, and perhaps neither is more correct than the other. Rather than stating one of these two articles has better explanatory power than the other, perhaps it is first necessary to understand where prosodic stress comes from. Is prosodic stress the result of information structure as Mulkern claims, or, is there an interface constraint holding between syntactic spellout and PF that recognises a syntactic F(ocus) feature which is separately interpreted by Logical Form (LF) as Jackendoff (1972, for example) claims? I believe this is the question we need to ask. Rather than assuming a theory, we need to justify that that theory in fact makes better predictions before jumping to the conclusion that an information structure approach has better explanatory adequacy than an OT interface approach.
A second article which raises theory-oriented questions is Kenji Oda's ''Preverbal particles and irregular verbs in Irish.'' This is also a phenomenon receiving a lot of attention at the moment (Acquaviva, 2010; Windsor 2011 among others). Oda's line of attack is to assume a distributed morphology approach, like Aquaviva. However, Acquaviva and Oda do not reach the same conclusion. Oda observes that there may be two tense features, a C-tense and a T-tense, which are merged in the case of dependent verbs, where only one tense feature is merged with independent verbs. Because two tense features would be more specific than a single tense feature, where that option is available it would always be more preferable. Interestingly, both Acquaviva (2010) and Windsor (2011) analyze Irish irregular verbs as being able to be merged with both a Future tense feature and a Past tense feature as in the case of forms such as 'get/find.FUT' which is analyzed as having the -(f)aidh future suffix as well as the lenited (-h-) beginning of a past tense form. While Windsor (2011) contrasts an OT morphology (Clahsen, 2006) with realization-based lexicalism (Ackerman and Stump, 2004), the idea of two tense features remains prevalent in this analysis as well. The point of highlighting these alternative analyses is not to shed doubt on Oda's analysis, but rather to point out the common ground of all three. What is needed next is again a theoretical question - do we attempt to motivate one theory above the other, or should we be looking for more theory-external evidence of these multiple tense features, and ask the question: is there theory-neutral, or theory-external evidence for these multiple tense features, and if there is, can supporting evidence for such a phenomenon be found in other natural languages?
I will only explicitly comment on one article from the second section: ''Welsh mutations and statistical phonotactics.'' Michael Hammond undertakes a huge project, looking at the statistical probabilities of word initial consonants and consonant clusters with the express concern that mutations introduce word initial consonants, or consonant clusters that would otherwise be illegal in terms of phonotactics. This is obviously true. The apparent reason for this analysis is to counter arguments such as those made by Zsiga (2000) that phonology is that which is concerned with the categorical, while phonetics is the discrete and the gradient. Hammond wants to use this data to make a case for gradient phenomena being included in phonological analyses in an Exemplar Theory model of neighbourhood densities. I see two concerns, one building from the other. The first is the complaint raised in Watkins (2005: 307) of ''treiglo gwael'' 'poor mutating' where it is claimed that ''there has been almost complete loss of nasal mutation and aspirate mutation, and the application of soft mutation is becoming increasingly arbitrary [but] the effect on the standards of written Welsh is predictable.'' Therefore, one problem of corpora study on this type of phenomenon is that one runs the risk of analyzing non-natural, literary language. In fact, Sharp (2011) reports that correct mutations may be applied as little as 50% of the time. Additionally, the author points out that ''contexts for mutations are lexical and syntactic, not phonological'' (344). Therefore, if mutations are not phonologically governed, and one runs the risk of analyzing a literary (non-spoken) language by using corpora data, we can ask how this can be evidence for gradient effects being included in a phonological analysis?
Despite the questions above, the author has done significant service to researchers in providing the statistical analysis of initial consonants and their mutations in Welsh. The data may prove useful for future research, especially in exemplar phonetics and may beg the question, given the frequency of the alternations and the resulting phonotactically illegal initial consonants and clusters, how do Welsh speakers perceive or interpret words which are correctly or incorrectly mutated? Is there a hindrance to word recognition if either the mutated or non-mutated forms are used instead of the other? This study lays groundwork for such questions, and I suggest that this research should be continued along those lines.
Much of the scholarship presented here highlights constructions which contribute to our theoretical knowledge. Some articles make new subdivisions of semantic or syntactic classes based on evidence from Celtic. Some contribute by improving on previous analyses, and others show how previous analyses can be amalgamated into one coherent theory by using Celtic data. The common tie is that all of these articles should encourage further scholarship either building on the areas identified for further research, or by forcing authors with different theoretical opinions to come up with an alternative analysis. The book is successful in achieving that goal.
REFERENCES Ackerman, Farrell and Stump, Gregory. 2004. Paradigms and periphrastic expression: A study in realization-based lexicalism. In, Sadler, L. and Spencer, A. (eds.) Projecting Morphology. Stanford: Center for the Study of Language and Information. 111-58.
Acquaviva, Paolo. 2010. Dependent verbs and tense in Modern Irish morphosyntax. Presented at the 6th Celtic Linguistics Conference: Dublin, Ireland. September 12.
Carnie, Andrew. 2002. A note on diphthongization before tense sonorants in Irish: an articulatory explanation. Journal of Celtic Linguistics 9:13-31.
Chomsky, Noam. 2008. On phases. In Freidin, Robert et al (eds.) Foundational issues in linguistic theory: Essays in honor of Jean-Roger Vergnaud: MIT press: Cambridge, MA. 133-66.
Clahsen, Harald. 2006. Dual-mechanism morphology. In Brown, K. (ed.) Encyclopedia of language and linguistics 2nd edition. Elsevier: New York. 11-5.
Jackendoff, Ray S. 1972. Semantic interpretation in Generative grammar: Focus and presupposition. MIT press: Cambridge, MA.
Prince, Alan & Smolensky, Paul. 2004. Optimality theory: Constraint interaction in generative grammar. Blackwell Publishing: Malden, MA.
Newton, Glenda. 2006. The development and loss of the Old Irish double system of verbal inflection. PhD dissertation, Cambridge University.
Newton, Glenda. 2008. Exploring the nature of the syntax-phonology interface: A post-syntactic account of the Old Irish verbal system. Ms. University of Cambridge.
Sadler, Luisa. 1988. Welsh syntax: A government and binding approach. Croom Helm: London.
Sharp, Kathryn Morgan. 2010. The ultimate jigsaw puzzle: Acquisition of gender in Welsh. Presented at the 6th Celtic Linguistics Conference. Dublin, Ireland. September 11.
Watkins, T. Arwyn. 2005. Welsh. In Ball, Martin J. (ed.) The Celtic languages. Routledge: New York. 289-348.
Windsor, Joseph W. 2011. Irregular Irish verbal morphology: Contrasting OT with word & paradigm. Unpublished Ms. University of Calgary: Calgary.
Zsiga, Elizabeth C. 2000. Phonetic alignment constraints: Consonant overlap and palatalization in English and Russian. Journal of Phonetics 28. 69-102.
ABOUT THE REVIEWER
ABOUT THE REVIEWER:
Joseph W. Windsor is a graduate student of Linguistics at the University of
Calgary. His primary interests are in prosodic phonology and the
phonology-syntax interface. Most of his work centers around Celtic
languages, predominately on Modern Connemara Irish.