The study also highlights the constructs of current linguistic theory, arguing for distinctive features and the notion 'onset' and against some of the claims of Optimality Theory and Usage-based accounts.
The importance of Henk Zeevat's new monograph cannot be overstated. [...] I recommend it to anyone who combines interests in language, logic, and computation [...]. David Beaver, University of Texas at Austin
Date: Fri, 22 Apr 2005 11:46:43 -0700 (MST) From: Andrew Carnie <carnie@U.Arizona.EDU> Subject: Principles and Parameters in a VSO language
AUTHOR: Roberts, Ian G. TITLE: Principles and Parameters in a VSO Language SUBTITLE: A Case Study in Welsh SERIES: Oxford Studies in Comparative Syntax PUBLISHER: Oxford University Press YEAR: 2005
Andrew Carnie, Department of Linguistics, University of Arizona. [*]
I'm well aware that few people are as titillated as I am by a new book on a verb-subject-object (VSO) order language, and one might think that the audience for such a book would be narrow and very specialized. However, Ian Roberts' new book: _Principles and Parameters in a VSO Language: A Case Study in Welsh_ is surprisingly humbly titled and really deserves a wider audience than its title might suggest. A significant body of research on the motivations for and mechanisms of word order variation (e.g., case, the Extended Projection Principle (EPP) etc.) have dominated recent work in minimalist principles and parameters syntax. Roberts' book provides a very important addition to this canon of work, bringing in new evidence from a relatively understudied language. Roberts draws together a well worked out sketch of Welsh grammar with a technically sophisticated explanation for the origins of EPP-like phenomena and the nature of parameters.
In addition to this important technical and theoretical contribution, Roberts' book fills a gap in the literature on the syntax of the Celtic languages. The syntax of Welsh seems to have had one or two major works associated with each major version of the theory since the 1970s (such as Awbery (1976), Morris-Jones & Thomas (1977) (Transformational Grammar), Sadler (1988) and Hendrick (1988) (Government and Binding theory)) This book completes the paradigm with a minimalist treatment of the major phenomena in the language (VSO order, verbal nouns, direct object mutation (DOM), so- called "abnormal" sentences, particles, clitic doubling, DP-structure, and a very cursory treatment of some of the wh-phenomena in the language).
SUMMARY AND EVALUATION
The introduction to the book lays out two desiderata for parameters governing cross-linguistic variation: (1) Parameters must be learnable (that is, the parameter must be expressed or determinable from the data) and (2) parameters must be typologizable (that is, the parameter must explain cross-linguistic variation). The tradition in minimalist syntax has been to treat parameters as conditions on features of particular lexical items. For example, DP-movement is typically viewed to follow from a "strong" D-feature on the functional head that licenses the DP. Roberts' great innovation in this book is that he grounds such features in the overt morphology, developing a system where heads with interpretable features must be licensed either by merger with a particle, head movement of a lower head or movement of an XP element into their specifier. The particular set of lexical items a language has determines which of these options is chosen and by extension the parameter settings for the language. A particularly impressive result of this approach is that Roberts is able to account for Welsh word order without appealing to the fairly stipulative notion of the Extended Projection Principle.
The first chapter, "The analysis of VSO languages", has a fairly misleading title. Roberts is, as he makes clear in a footnote, only making claims about Welsh (and to a certain degree other closely related Celtic languages), and is not making any broad generalizations about VSO languages outside of Celtic. Indeed, it has become clear in recent years that different VSO languages derive the order in significantly different ways (see for example, the articles in Carnie and Guilfoyle 2000 and Carnie, Harley and Dooley (2005)). Roberts reviews the standard arguments for an underlying structure with a VP, and the arguments for a head-movement analysis. He argues against a V (to T) to C analysis of VSO order in Welsh, providing a preliminary analysis of the C-system of the language. He also presents arguments that the subject has raised out of the VP. The preliminary analysis he gives in this chapter is surprisingly conservative in that it uses AgrSP. The structure Robert's proposes is:
In the standard technology of MP, this amounts AgrS having a strong V-feature, a weak D-feature, and T having a strong D-feature. In the next chapter, Roberts grounds these feature settings in the morphological system of Welsh.
One problem, as Roberts notes for the analysis set out in chapter 1 is the lack of "EPP" effects in the Celtic languages (McCloskey 1996). A strong D-feature on T should predict the existence of expletives appearing in the specifier of TP. However, in Welsh there is fairly clear evidence that in impersonal and existential constructions the specifier is not filled by an expletive or any other DP. In order to account for this, Roberts' Chapter 2: "Case agreement and Mutation," explains the motivations for movement in a different way; deriving them partly from the richness of the morphology of Welsh. Building upon a comparison with a variety of North Italian dialects, Roberts suggests first splitting Agr into two categories (Pers(on) and Num(ber)). He claims that the higher of these two categories (Pers) in Welsh is occupied by a subject clitic (which is the morphology traditionally called "agreement" in the Welsh literature). This morpheme attracts the V. Welsh lacks agreement morphology in the traditional sense. This explains the apparent "anti- agreement" effects of the Celtic languages, where the verb fails to agree with overt non-pronominal objects. The subject DP fails to move to the specifier of PersP because Pers is already identified by the verb and the subject clitic morphology. The system proposed here makes a parametric distinction, which seems to be reflective of at least some of the possibilities. (1) A language can have no rich agreement, and no V-movement. (2) A language can have a subject clitic and no "real" agreement; this results in a language like Welsh. (3) A language can have a subject clitic and rich agreement which results in cases like some North Italian dialects where agreement is doubled by a clitic.
In Welsh, the DP raises to the specifier of NumP, not as one might expect, to get case, but rather to license Num itself. This leads us to the least pleasing part of Roberts' analysis (really the only displeasing part): his view of DP licensing. In my opinion, a striking advance of research in the Minimalist Program has been the reduction of a variety of licensing mechanisms present in GB theory to the single mechanism of feature checking. Roberts, however, divorces case from feature checking and retreats to a purely configurational view of case licensing. In fact, particular cases aren't associated with a particular position in a tree, but rather are determined by where the NP falls within a thinly disguised version of government, that is, Rizzi's (2001) "minimal configuration" (to be fair Chomsky's "Agree" relation also seems to be a thinly disguised version of government). To speak plainly, if an NP is governed by a CP-internal Agr node (Pers,Num), then it receives Nominative Case. If it falls within the minimal configuration (i.e., is governed by) of little v then it gets Accusative Case.
This is a disappointing retreat to GB theory in my opinion. The empirical motivation for Roberts' view is straightforward if you are willing to adopt his assumptions about what is "Nominative" and what is "Accusative" case marking in the language. Welsh, like all the modern Celtic languages, exhibits Initial Consonant Mutation (ICM). ICM is normally lexically triggered by a preceding word. Welsh, however, has one exception that seems to be grammatically -- rather than lexically -- determined: Most (although not all!) direct objects are marked with the "soft" mutation. This is typically referred to as Direct Object Mutation or DOM (not to be confused with Aissen's 2003 Differential Object Marking). DOM does not apply in all cases, for example it does not apply on the object DP in impersonals, nor on the objects of non- finite verb forms (verbal nouns). The usual view of DOM in the literature on Welsh is that it is triggered only when the direct object is immediately preceded by an XP (which can be null) (See for example the analyses of Harlow (1989), Borsley and Tallerman (1998) and Borsley (1999)). Roberts breaks from this tradition and identifies DOM directly with Accusative case (or more precisely, with a floating autosegment which triggers the mutation only when the element in question is head-governed by the relevant v category which impersonals, existentials and verbal noun constructions lack). In effect this means that nominative case is indicated by the lack of ICM. As such, under this particular set of assumptions, object DPs that don't exhibit DOM must bear Nominative Case. However, there is clear evidence that these NPs are not in the specifier of NumP, where subject DPs are found, so Roberts has to allow case to be configurationally assigned (i.e., under the broad domain of government) rather than being narrowly associated with one position within that configuration and reducing to feature checking. Note however, that it is possible to maintain Roberts' analysis of DOM without the concomitant assumption that DOM reflects Case. Indeed it would seem to be easier to assume that Case is simply entirely covert in Welsh and arises in the typical way (in a spec-head relation via feature checking). DOM is, as Roberts argues, a reflection of co-occurrence with a particular little v head. In fact, it could be defined in terms of simple linear adjacency to the v head bearing the triggering autosegment . If this is true, we need not appeal to government at all. This minor adjustment to the analysis would allow us to dispel with the extra mechanism of case assignment under government.
The third chapter, "Genitive Case, Word Order in DP and Objects of Non-finite Verbs" really deals with two distinct issues, which are only related on a superficial level. First, there is an interesting discussion of the internal structure of the DP and the assignment of genitive case. Building upon Duffield's (1996) work on Irish, Roberts shows that possessive constructions in Welsh are best analyzed as construct states familiar from the Semitic languages. He argues for a structure where the N head moves to a Q head in all DPs (thus accounting for the relative order of the N, adjectives, demonstratives and complements of the N), and further on to D when a possessor is present. Genitive case is assigned in a minimal configuration with a DP-internal AgrP. The second part of the chapter concerns non- finite "verbal noun" (VN) constructions of Welsh. Roberts adopts Borsley (1996)'s claim that VNs are not nouns, instead arguing they are of the category "participle" which are selected for by an aspectual head. The objects of VNs fail to get DOM because they aren't in the minimal domain of a v head (although the VN themselves are and do take the marking).
The structure of the CP is the focus of the fourth chapter: "The C- system and the Extended Projection Principle." Here, Welsh clause- initial particles are placed within the context of Rizzi's (1997) articulated CP structure. Roberts shows that some particles are best analyzed as Force markers and others as Fin(ite) markers. He draws a parallel between Welsh particles, Breton long verb movement and V2 in Germanic, all relating to the requirement that the Fin head be realized. Much of this book centers on the interaction between head- movement, various particles and functional heads. The last chapter, "Head-Movement and EPP features", is thus appropriately focused on addressing Chomsky (2001)'s claims that head-movement is not part of narrow syntax. Roberts gets around Chomsky's objections by proposing that head- movement is really a two-part operation. The first part involves movement to a specifier as found elsewhere. The second part involves a PF-sensitive formal feature "Affix", which triggers a morphological process of incorporation, where the specifier is fused with its head. On one hand, this is a very welcome addition to the theory, as it allows us to bring the empirically very well motivated phenomenon of head-movement into phase theory and overcomes Chomsky's conceptual arguments against head-movement as a syntactic operation. On the other hand, one is, of course, forced to wonder why such an empirically robust phenomenon is marginalized on conceptual grounds (alone) in the first place.
_Principles and Parameters in a VSO language_ is a surprisingly theoretically and empirically rich work for a book that comes in at just 200 pages (including references and notes). As I suggested in the introduction, this is work of importance not only to Celticists and VSO- ologists, but also to any syntactician seriously considering the nature and variation in word order derivation and constituent licensing.
[*] My thanks to Heidi Harley, Terry Langendoen, Sheila Dooley and Ian Roberts for their input on this review. Typically, errors are mine.
 There are a number of adjunct elements that could potentially intervene. However, if we adopt a Speas (1992) style analysis of adjuncts, then they do not intervene at the relevant stage of the derivation.
Aissen, Judith (2003)"Differential Object Marking: Iconicity vs. Economy." Natural Language & Linguistic Theory, 21: 435-83.
Awbery, G. M. (1976) The Syntax of Welsh. Cambridge: Cambridge University Press.
Borsley, Robert (1996) "On a nominal Analysis of Welsh Verb-Nouns" In Anders Ahlqvist et al. (eds.) Dan do Oide: Essays in Memory of Conn. R. O Cleirigh. Dublin: Institiuid Teangeolaiochta Eireann. 39-47.
Borsley, Robert (1999) "Mutation and Constituent Structure in Welsh". Lingua 109:267-300.
Borsley, Robert and Maggie Tallerman (1998) "Phrases and Soft Mutation in Welsh" Journal of Celtic Linguistics 5: 1- 33.
Carnie, Andrew and Eithne Guilfoyle (eds.) (2000) The Syntax of Verb Initial Languages. Oxford: Oxford University Press.
Carnie, Andrew, Heidi Harley and Sheila Dooley (eds.)(2005) Verb First. Philadelphia: John Benjamins.
Chomsky, Noam (2001) "Derivation by Phase." In Michael Kenstowicz (ed.) Ken Hale: A Life in Language. Cambridge: MIT Press 1-52.
Duffield, Nigel (1996) "On Structural Invariance and Lexical Diversity in VSO Languages: Arguments from Irish Noun Phrases. In Borsley and Roberts (eds.) The Syntax of the Celtic Languages. Cambridge: Cambridge University Press 314-340.
Harlow, Steve (1989) "The Syntax of Welsh Soft Mutation" Natural Language and Linguistic Theory 7: 289-317.
Hendrick, Randall (1988) Anaphora in Celtic and Universal Grammar. Kluwer, Dordrecht.
McCloskey, James (1996) "Subjects and Subject Positions in Irish." In Roberts and Borsley (eds.) The Syntax of the Celtic Languages. Cambridge University Press. 241-283.
Morris-Jones, R and A.R. Thomas (1977) "The Welsh Language: Studies in its Syntax and Semantics" Cardiff: University of Wales Press.
Rizzi, Luigi (1997) "The fine structure of the left periphery" in Liliane Haegeman (ed) Elements of Grammar. Dordrecht: Kluwer Academic. 281-337.
Rizzi, Luigi (2001) "Relativized Minimality Effects" in Baltin and Collins (eds) The Handbook of Contemporary Syntactic Theory. Oxford: Blackwell.
Sadler, Louise (1988) Welsh Syntax, A Government Binding Approach. New York: Croom Helm.
Speas, Margaret (1992) Phrase Structure in Natural Language. Dordrecht: Kluwer Academic.
Tallerman, Maggie (1998) "On the Uniform Case-licensing of Subject in Welsh." Linguistic Review 15: 69-133.
Tallerman, Maggie (1999) "Welsh Soft Mutation and Marked Word Order." In Darnell et. al (eds.) Functionalism and Formalism in Linguistics. Philadelphia: John Benjamins. 277-294.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Andrew Carnie is an Associate Professor of Linguistics at the University of Arizona. His research varies from phrase structure theory to Irish Phonology, but much of his work has centered on the analysis of VSO languages in general and Celtic in particular. In addition to writing numerous research articles, he is the co-editor of two volumes on the syntax of VSO languages (Carnie & Guilfoyle 2000, Carnie, Harley & Dooley 2005), and is the author of Blackwell's textbook _Syntax: A Generative Introduction._