Date: 11 Aug 2004 16:17:46 +0100 From: Jonny Butler Subject: Peripheries: Syntactic edges and their effects
EDITOR: Adger, David; De Cat, Cécile; Tsoulas, George TITLE: Peripheries SUBTITLE: Syntactic edges and their effects SERIES: Studies in Natural Language and Linguistic Theory 59 PUBLISHER: Kluwer Academic Publishers YEAR: 2004
Jonny Butler, Dept. of Language & Linguistic Science, University of York
This book contains 15 papers presented at a conference on peripheral positions at the University of York in September 2000, along with an introductory chapter.
Chapter 1, David Adger & Cécile De Cat's ''Core questions about the edge'', introduces the volume. It gives a rundown of general treatments of peripheral positions obtaining in the literature, defined in very general terms as positions at 'the edge of some syntactic domain' (p. 2), as opposed to what the authors label 'central' positions, i.e. positions 'deeply embedded within that domain' (ibid). These include pretty specific notions of the periphery, such as Rizzi's (1997) 'left' periphery, i.e. CP; generalized versions of this Rizzian idea to cover e.g. the periphery of the nominal domain (DP); more general notions still, such as the higher/edge parts of the clause, whatever syntactic category we take these to be; and of course, quite far removed from Rizzi, also the 'right' periphery of syntactic constituents. This chapter also includes brief discussion of each of the other chapters in the volume, usefully highlighting connections between them.
Chapter 2, Ronnie Cann, Ruth Kempson, Lutz Marten, Masayuki Otsuka, & David Swinburne's ''On the left and on the right'', analyses left and right clause peripheral phenomena (topicalization, relative clause formation, 'it' extraposition, right node raising, etc.) in terms of the authors Dynamic Syntax framework, wherein tree growth reflects the 'left to right' nature of production/processing, rather than say the more usual hierarchic structure building treatment of generative frameworks. They make use of two specific operations, *Adjunction and LINK, both of which are utilized to provide analyses of left vs. right constructions, typically treated as distinct, in very similar ways. The numerous constructions covered include topicalization, ''hanging'' topics, pronoun/clitic doubling, ''it'' extraposition, Right Node Raising, and relativization.
Like other analyses based in the Dynamic Syntax framework I have seen, this requires a brain-twisting rethink of how I look at syntax, but it is extremely interesting, and well worth the twist. It also features the best example sentence in the volume, ''She talks too fast, Ruth Kempson'' (p. 33).
Chapter 3, Balazs Suranyi's ''The left periphery and cyclic spellout: the case of Hungarian'', takes issue with recent 'cartographic' analyses of scope effects (e.g. Beghelli & Stowell 1997), particularly relating to Hungarian (e.g. Szabolcsi 1997), wherein QP scope is taken to be assigned via syntactic feature checking with a set of hierarchically ordered functional clausal heads. Suranyi criticizes such analyses on two main grounds: empirically, that low in the Hungarian clause scope is not reflected structurally, and this requires in Szabolcsi's treatment an apparently unconstrained iteration of the functional heads she proposes, which lessens the motivation for having a strict hierarchy in the first place; and theoretically, that various of the movements required to obtain both the correct surface relations and the correct scope relations where these don't match can be seen as violating constraints such as the ban on Improper Movement, etc. Suranyi instead proposes a treatment based on interactions of Quantifier Raising (QR) and topicalization, in a framework assuming a version of phase-based cyclic spellout (Chomsky 2000, etc.).
The analysis involves a number of assumptions which aren't without problems - e.g. overt focalization is taken to be driven by verb movement. In focus constructions, Suranyi claims V has a strong uninterpretable focus feature [uFoc]. One problem is that we don't see such a feature reflected morphologically on V. While this is not unusual in the literature generally, Suranyi levels the criticism at Beghelli & Stowell's treatment of distributivity that their head Dist doesn't show up overtly. If this is a problem, I don't see why proposing features that don't show up overtly is less of one. Leaving this aside, Suranyi claims that in order to check this feature, V raises out of the projection it heads (VP, or AspP if it has raised there already), adjoins to that projection, and reprojects as FocP. Early on, he takes the motivation for this kind of head movement to be related to phasal spellout: a head moves out of a potential phase if it has unchecked features that would cause that phase to crash at spellout. However, he doesn't take VP or AspP to constitute a phase, so the movement here seems unnecessary: there seems no reason why the [uFoc] feature of V couldn't be dealt with in situ. Leaving this too aside, the phrase to be focussed has a strong interpretable [iFoc] feature, which raises and Agrees with the [uFoc] on V, but doesn't necessarily check. Suranyi assumes all syntactic operations, including checking, are optional. If the wrong option is taken crash will ensue, but that is a chance that is taken. If the features do check, then the construction is sent to spellout. If not, V must raise and reproject again. It can either do this immediately, in which case a second expression can be focussed, and so on seemingly ad infinitum (which again seems problematic); or not, in which case QR can occur first.
QR is another optional operation, according to Suranyi, and as such it can either be overt or covert. In Hungarian, we see both these options, as opposed to say English, where it is always covert. Suranyi's explanation for this lies in the nature of the features at the base position of the movement. He claims that if strong features obtain on a base position, e.g. theta-features, then there is required to be an overt argument in argument position, essentially to support those features. English theta-features are apparently strong, so require arguments to be realized low down where optional raising takes place. Hungarian theta-features, on the other hand, are apparently weak, and so don't impose this requirement. In itself, this is somewhat stipulative; it is also problematic theoretically, since Suranyi states that in cases of featurally motivated movement, such as topicalization or A-movement, movement is necessarily overt. This conflicts with Suranyi's position that where the base-feature is strong, 'then by definition, it requires the presence of a full category' (p. 69), i.e. including its phonological features. In a case where both the base- features and the attracting features are strong, then, it is required that the moving element is realized both in the lower and the higher position, and yet clearly this isn't what we see. While Suranyi's paper raises some legitimate objections to the feature-based cartographic theory of scope, then, it also introduces some new problems of its own.
Chapter 4, Joseph Emonds' ''Unspecified categories as the key to root projections'', also argues against cartographic treatments of peripheral positions, specifically of the higher part of the clause commonly referred to as CP. In fact, Emonds pretty much does away with the category CP altogether: he argues that root clauses (essentially non- embedded finite clauses, or embedded clauses with equivalent properties) are 'Discourse Projections', i.e. they give information about the speaker's attitude towards the 'eventhood' of the proposition under consideration (realis vs. irrealis). He then treats various periphery-related transformations in terms of 'Discourse Shells', which are defined as 'Categorially unspecified projections [that] may immediately dominate (only) IPs specified as Discourse Projections' (p. 85). Any cateory of XP may then move to the specifier of one of these Discourse Shells, the interpretation that XP receives (topic, focus, etc.) then depending on independent factors. In combination with a version of his (1976) Structure Preservation, Emonds uses this notion of a-categorial Discourse Shells to provide elegant analyses of left dislocation constructions, topicalization, exclamatives, focalization, wh-movement, and verb movement. One possibly minor point remained unclear, which was that Discourse Shells may be dominated by other Discourse Shells. However, Emonds' definition of a Discourse Shell above states that they 'immediately dominate (only) IPs specified as Discourse Projections'. This seems to entail that when we have a finite IP with a Discourse Shell on top, that is equivalent to having a finite IP without, otherwise we couldn't add more Shells. This seems fine, but clarification on this point would have been useful.
Chapter 5, Kook-Hee Gill & George Tsoulas's ''Peripheral effects without peripheral syntax: the left periphery in Korean'', also also argues against a (universal) cartographic analysis of the left periphery, preferring instead an approach where left peripheral elements may be hosted in the specifier of a Topic position between C and IP, or more radically simply adjoined to IP. The left peripheral elements Gill & Tsoulas are concerned with are '-nun' marked elements, which, when peripherally dislocated, are interpreted as topics, with various very specific restrictions, such as no matter how many -nun marked elements appear in the periphery, only one gets the topic interpretation. Gill & Tsoulas argue that treating this interpretation in terms of checking of a [+TOPIC] feature, or anything similar, is not only theoretically untenable, but also empirically. Rather, they argue the interactions between movement to the relevant peripheral position and phonological stress assignment in Korean derive the possible interpretations straightforwardly. The theory they finally arrive at is impressively minimal, especially given that it derives so many effects for them - they even manage to bring short distance scrambling under the umbrella. A few questions are left open at the end, but as they are acknowledged and discussed this isn't a problem.
Chapter 6, Mamoru Saito's ''Japanese scrambling in a comparative perspective'', presents arguments that Japanese (and Korean) scrambling is a non feature driven operation. Saito first reviews evidence from his, and others', earlier work (Saito 1989; Tada 1993; Oka 1989) showing that Japanese/Korean scrambling (unlike in other scrambling languages) has the property of allowing scrambled elements to undergo radical reconstruction - that is, for scrambled elements to receive an interpretation exactly as if they hadn't moved at all. This effect seems optional with short (clause internal) scrambling and obligatory with long scrambling. Saito suggests that the best candidate for a feature that could drive this kind of movement is the P(eriphery)- feature of Chomsky (2000): a feature that can be optionally assigned to the head of a syntactic phase to trigger successive cyclic movement. Chomsky (2000) assumes that the (at least intermediate) steps of such movement are semantically vacuous. However, Saito argues that even so, they do not allow radical reconstruction. He then goes through a different formulation from Chomsky (2001) wherein the assignment of a P-feature to a head must have an effect on the interpretation. Clearly this isn't the case with the cases Saito considers, since as far as the interpretation goes it is as if the scrambled elements hadn't moved at all.
Saito concludes that if the P-feature, in either guise, is the best candidate for a feature that drives scrambling, and yet according to his analysis it doesn't, then it really looks like Japanese/Korean scrambling isn't feature-driven at all. The arguments are clear and well- presented, but we are left with the niggling question of what does drive scrambling in Japanese in that case.
Chapter 7, Enoch Aboh's ''Left or right? A view from the Kwa periphery'', shows that Gungbe, a language of the Kwa family, has a set of rigidly ordered topic, focus, and injunctive markers at the left periphery, which host topics, foci, and subjects of injunction to their immediate left, suggesting that a decomposed discourse related CP structure obtains in this language. Aboh presents a large amount of straightforward data backing up an analysis along these lines, his arguments similar to those used by Rizzi (1997) to motivate his decomposed CP structure for Italian.
More interestingly, Aboh also examines discourse markers that appear right-peripherally. On the assumption that heads universally precede their complements at the base (Kayne 1994), Aboh notes that if these discourse markers too reflect CP level discourse heads, their right peripheral nature must deriving from a leftward 'snowballing' movement of other elements to/through their specifiers. Such a snowballing analysis predicts that, for those elements that can appear in either periphery, such as topic or focus markers, their right peripheral ordering should directly mirror their left peripheral ordering, and Aboh shows that this is correct (an effect that he shows also turns up at the left and right edges of Gungbe nominals). Aboh analyses this snowballing movement as a 'disguised' version of the kind of head movement we commonly see in Romance, the distinction being whether the head X can raise alone (Romance), or whether it has to pied-pipe its XP (Gungbe). Aboh proposes an interesting scope related argument to motivate the distinction between left and right peripheral realizations of the CP heads: essentially, when, say, a Foc head is realized on the left, with a focalized XP in its specifier, this reflects the fact that that XP is the scope of the focus; when a Foc head is relaized on the right, with the rest of the clause in its specifier, this reflects the fact that the whole proposition is in the scope of focus. Interpretive facts seem to back this up. Tellingly, this also predicts that those discourse markers that obligatorily appear to the right in Gungbe, such as the interrogative marker and the clausal determiner, are those that necessarily scope over the proposition, and again this seems to be the case.
Chapter 8, Christer Platzack's ''Cross-linguistic word order variation at the left periphery: the case of object first main clauses'', again looks at the left periphery of the clause in terms of a Rizzian exploded CP layer. Data from English, Italian, Finnish, and Swedish are marshalled to demonstrate the variation in the ordering of non-object constituents found in main clauses whose initial constituent is the direct object, both cross-linguistically, and within languages depending on the status of the object as a DP topic or a wh-word.
Platzack's general claim is that fronting a wh-object over a subject should in general constitute a violation of Shortest Move (Kitahara 1994). This claim depends very much on some of Platzack's background assumptions: (i) that the canonical subject position is [Spec,FinP], i.e. an A-bar position, not [Spec,TP], an A-position; (ii) that wh- movement is attraction to Foc of a DP from a lower A-bar position by something like a generalized EPP-feature, not by any kind of [wh]- feature; (iii) that wh-objects occupy an A-bar position at the left edge of the verb phrase before movement to [Spec,FocP]. The first and third assumption get no justification other than references to Branigan (1996) and Nissenbaum (2000) respectively; the second claim gets around half a page of discussion. However, with these in place, Platzack shows how a version of Richards' (1998) Minimal Compliance offers a few ways get round the Shortest Move violation - either by moving the intervening XP, or moving the head of which it is a specifier - and demonstrates how each of the languages under consideration exemplifies one of these ways.
The paper is well-argued and elegant, though it would be nice to have seen the background assumptions (i-iii) justified in more detail.
Chapter 9, Liliane Haegeman's ''DP-periphery and clausal periphery: possessor doubling in West Flemish'', does very much as it says in the title: examines the well-known DP-clause parallelism with respect to data on possessor doubling from West Flemish (WF). Specifically, Haegeman shows how a proposal by Gavruseva (2000) to deal with the unavailability of possessor extraction - the phenomenon where a possessor wh-word is wh-moved out of a possessive DP - in Germanic fails to account for data from WF. Gavruseva proposes that the availability of possessor extraction out of DP in a language correlates with the availability of movement of the possessor to the edge of DP. Haegeman shows that in WF, this correlation doesn't hold: possessors may move to the edge of DP, but (contrary to initial appearances) they cannot extract out of DP. Haegeman solves this problem by reworking Gavruseva's analysis with a more articulated DP periphery than Gavruseva assumes, analogous to a Rizzian articulated CP, the point being essentially that in WF, the possessor moves to a low A-position in the DP periphery, from where extraction is impossible, whereas languages exhibiting extraction move their possessors to a higher A-bar position in the DP periphery, from where extraction is licit.
This is a typical Haegeman paper: the data at issue are clearly set out, the analysis is sound and well-presented, and it's about West Flemish.
Chapter 10, Cedric Boeckx & Kleanthes K. Grohmann's ''SubMove: towards a unified analysis of scrambling and D-linking'', points out five clear parallel behaviours displayed by D-linked wh-phrases and long-distance scrambled elements: neither obeys superiority; both carry discourse effects; both are, or have been argued to be, semantically vacuous in the sense that they undergo radical reconstruction; both are insensitive to islands; and both are accompanied by clitic doubling in some languages. Such parallels suggest a unified analysis, and Boeckx & Grohmann provide one, the essential ingredient being that neither type of movement serves to check phi-features, but rather they both target something like a TopP in an articulated CP.
To instantiate this notion, and capture the parallels, B&G propose an operation SubMove, in which a DP complement of a D head may extract out of the embedding DP to a higher clausal position. Somewhat along the lines of Sportiche (1996), the stranded D may then be spelt out as a resumptive element/clitic to the extracted DP. B&G suggest that both D- linked wh-movement and long distance scrambling are cases of SubMove. Assuming the embedding DP is the one that checks phi-features in the clause, the extracted DP is predicted not to check such features, and the parallels B&G note fall out: superiority won't kick in if the elements to be moved are embedded inside higher DPs, since neither will c-command the other; discourse effects follow if the element moves out of the agreement domain of the clause (IP) to CP, often viewed as the discourse-related part of the structure; radical reconstruction follows if we assume that the resumptive D is in fact the relevant one for interpretation in these cases (though without actual reconstruction); island insensitivity is a well-known feature of resumptive structures; and clitic doubling is the basis of the whole analysis. This is another excellent paper, with a novel observation very nicely analysed.
Chapter 11, Peter Svenonius's ''On the edge'', seems to catch the author on a surprisingly bad day. Like some of the other papers, it explores the DP-clause parallelism, here in terms of phase theory (Chomsky 2000 et seq). The paper seeks to explain various strategies for moving elements to/through phase edges, and explain why and how particular kinds of elements can('t) get out of phases in particular languages. Unfortunately the amount of data and theory Svenonius brings to bear are enormous and detailed, and this might have been better off as a monograph than a paper, so the points had room to be made. As it is, they don't, and this is a shame because they seem like they would be good ones if they did. A particularly bad case is the theory of structure building proposed at the end of the paper: it is highly intriguing, it certainly would have many ramifications for the way we think about various aspects of syntax, and (if it works) it is appealing. But it only takes up about a page and a half, mostly trees, and so there really isn't much we can say about it except ''oh''. As I say, this is a shame, and I hope Svenonius does expand this paper, whether to a monograph or a number of other papers, since I get the feeling I might enjoy it.
Chapter 12, Kyle Johnson's ''Clausal edges and their effects on scope'', examines the scope rigidity that is observed between the objects in double object constructions, i.e. the fact that whatever the scope relations we observe between the subject (or other scope bearing element) and either of the objects, the objects themselves invariably scope in their surface positions relative to one another. Johnson attempts to tie this down to the proposal that reconstruction into a small clause is barred, because reconstruction into a theta-marked position is barred, and small clauses only make theta-marked positions available.
Beyond this, I can say nothing since I frankly admit I was left hopelessly confused by this paper. This may be down to a lack of something on my part, but I wouldn't wish to shoulder all the responsibility, since for example, Johnson seems to switch around his labelling for the objects throughout, so that sometimes one object is referred to as the (in)direct object and sometimes the other: see for example p. 310, where we are assured in one sentence that the theory ''will allow for every permutation of scope relations ... except those in which the indirect object falls within the scope of the direct object: precisely the desired outcome''; and not only that, the very next sentence tells us the theory also succeeds in ''ensuring that the indirect object always falls within the scope of the direct object'' - presumably NOT precisely the desired outcome, but both things are presented as equally attractive. I remain puzzled.
Chapter 13, Valentina Bianchi & Roberto Zamparelli's ''Edge coordinations: focus and conjunction reduction'', marks a return to form for the volume. B&Z look at coordination structures of the type ''not only ... but also'' made famous by Peter Cook and Dudley Moore; they label these ''edge coordinations''. Data from English and Italian are given, demonstrating four possible variants for this kind of construction in each language. Edge coordinations show clear focus effects, as reflected in the (optional) ''only'' or equivalent focalizing element. B&Z assume they also involve deletion, assuming reasonably enough that they involve coordination of clauses, with ''reduction'' of parts of those clauses. They provide two possible ways for analysing each of these things: focalization either in situ, or, following Johnson (1996), as overt movement to a specifier of FocP; and reduction either as ellipsis, or as across the board remnant movement to a high CP level position [Spec,GroundP] (cf. Polletto & Pollock 1999). These distinct analyses, they show, are empirically distinguishable, and they also show very interestingly that they all play a part in edge coordination, different pairings corresponding to the four different ordering realizations for the constructions.
The paper provides a very clear, simple set of analyses for edge coordination structures, and its demonstration that they don't require a uniform analysis, and in fact require a non-uniform one, makes an important point about how much we should assume theoretically when we see pieces of data that look analogous.
Chapter 14, Theodora Alexopoulou, Edit Doron, & Caroline Heycock's ''Broad subjects and clitic left dislocation'', compares the two phenomena named in the paper's title - ''broad subjects'' (BS) (Doron & Heycock 1999), which being left peripheral elements in A-positions that are associated with a clitics, and clitic left dislocation (CLLD), a similar phenomenon except the left peripheral XP is in an A-bar position - using data from Italian, Greek, Hebrew, and Levantine Arabic. The authors give a couple of straightforward criteria to distinguish the two phenomena: the peripheral XP (PXP) in CLLD is not necessarily pronominal, whereas the PXP in BS constructions necessarily is, and necessarily bears nominative case (where case is visible); the PXP in CLLD is not focus stressed, and is standardly considered to have a topic-like discourse function, whereas the PXP in BS may be focus stressed, and can perform the discourse function either of focus or of topic.
They then proceed to go in detail through a number of other factors that demonstrate that BS should be distinguished as a discrete category from CLLD. These include: BS PXPs show standard syntactic subject behaviour, whereas CLLD PXPs do not; BS PXPs don't play a specific role in the information structure of the clause - expected if they are located in an A- position (i.e. in the IP layer) - CLLD PXPs do - also expected if they are in an A-bar position (i.e. the CP layer); the PXP in BS can be a ''bare quantifier'' (i.e. a single word quantifier with no lexical restriction, such as ''nobody'', ''everyone'', etc.), the PXP in CLLD can't; downward entailing quantifiers are licit as the PXP in BS, but not CLLD; quantificational PXPs in BS can bind pronouns, unlike CLLD; BS PXPs can be questioned (wh-moved), CLLD PXPs can't; BS PXPs don't respect island restrictions, CLLD PXPs do, as a defining characteristic; conversely, BS PXPs create islands, whereas CLLD PXPs don't; BS PXPs can't bind pronouns under quantificational reconstruction, CLLD PXPs can.
Having established the above distinctions between BS and CLLD, the authors then go on to show that both constructions can be available in the same language: while Greek and Italian have CLLD without BS, and Modern Hebrew has BS without CLLD (Doron & Heycock 1999), Levantine Arabic data split into two, some showing the characteristics of CLLD, and some the characteristics of BS. This provides yet further argument that the phenomena are distinct. This paper is a very carefully argued, solid piece of work, with the data set out clearly and the analysis following straightforwardly from it.
Chapter 15, Theodore Marinis's ''Acquiring the left periphery of the Modern greek DP'', extends the ''bottom-up'' analysis of acquisition initiated for the clause by Radford (1986) to acquisition of DP. Data come from two corpora, covering five Greek children; Marinis concentrates on the acquisition of case, nominal agreement, possessive constructions, and determiner spreading, the phenomenon where a definite determiner may be realized more than once within a DP, specifically preceding the noun and any modifying adjectives. The paper provides an analysis of the data such that first the NP, as a lexical thematic category, is acquired, then an ''FP'' layer corresponding to IP in the clause, then DP, taken to correspond to CP in the clause. Marinis ties this incremental acquisition process into Platzack's (2001) notion of ''multiple interfaces'', wherein the thematic (VP/NP), grammatical (IP/FP), and discourse (CP/DP) layers of the structure relate to distinct aspects of cognition. This provides a story for why acquisition should be incremental that is rather more plausible than just assuming one XP in some universally defined structure has to be acquired before the next one up can be.
Chapter 16, Bernadette Plunkett's ''Early Peripheries in the absence of C'', looks at null subjects and wh-questions in child French. It rejects the account of Rizzi (1993) wherein the gradual disappearance of French child null subjects is causally tied in to the maturational acquisition of the peripheral, CP, layer of the clause (cf. Crisma 1992). Data largely from two recent corpora are used to show that the picture isn't as straightforward as previous studies have assumed: Plunkett provides evidence that incremental acquisition of clause structure isn't simply a maturational, bottom to top, process, but rather an artefact of parameter settings. The idea is that children posit additional functional projections on a data-driven basis, and initially employ those projections only when a specific utterance makes it necessary. Thus until agreement parameters in the IP layer have been fully set, CP level parameters can't be fully set - e.g. I-C movement can't be acquired until V-I movement is in place.
Plunkett's analysis of the incremental acquisition of agreement features rests on a simple basic idea: settings for those features that are most fully specified in a target language are acquired before setting for those that are less specified. This basic idea makes some very interesting predictions: in particular, say a child has learned from the verbal paradigm that verbs can encode tense features (and therefore move to I), but has yet to determine whether they encode person or number features, then they will assume they don't. This will disallow feature checking between an overt pronoun, with person and number features, in [Spec,IP], and a verb without such features in I. A null pronoun must therefore be employed. As more specific agreement parameters are acquired, the number of null subjects will naturally decrease; and additionally it will become possible to set the parameters for higher functional projections, i.e. those in the CP layer. The correlation between the loss of null subjects and the rise in wh-questions is therefore not direct.
Plunkett's basic idea, as noted, is very straightforward but very interesting, and it would be still more interesting to see it applied to acquisition data from languages with different agreement paradigms, since it ought to make different, quite specific, predictions based on which features are realized, and to what relative degree, in any given language.
GENERAL EVALUATION: This is a generally excellent volume, managing to fit a bumper number of papers inside its deceptively slim covers. There are a couple of disappointments, as noted above, but overall the standard of the papers is very high. The range of phenomena discussed is impressive, as is the theoretical diversity within the papers.
I won't give an exhaustive list of typos, as occasionally appear in Linguist List reviews: it wouldn't be very interesting, I didn't make a note of them all, and you can find them yourself easily enough if you really want to. There are a couple that might be mentioned though. Two of the larger errors occur in Johnson's chapter 12: one is that he makes crucial reference to shaded text in one of his trees, when in fact no shaded text exists there. The editors have assured me, though, that the shading was there in the final proofs, and disappeared somewhere between there and publication. In fact it is clear enough from the text where the shading would have been, so this isn't too bad. The second is that the end of footnote 15 strays into the main text, causing mild confusion if you try to read either. Again, though, it is easy enough to reconstruct what should have been where. Apart from these, there are frequent bracketing problems with citations, such as they come out as e.g. Adger et al (2004) when they should be (Adger et al 2004), or vice versa. These prove momentarily distracting and could have been fixed easily enough.
REFERENCES: Beghelli, F. & T. Stowell. 1997. Distributivity and negation; The syntax of 'each' and 'every'. In A. Szabolcsi (ed.) Ways of Scope Taking. Dordrecht: Kluwer, 71-107
Branigan, P. 1996. Verb-second and the A-bar syntax of subjects. Studia Linguistica NS 50, 50-79
Chomsky, N. 2000. Minimalist inquiries: the framework. In R. Martin, D. Michaels, & J. Uriagereka (eds.) Step by Step. Cambridge, MA: MIT Press, 89-155
Chomsky, N. 2001. Derivation by phase. In M. Kenstowicz (ed.) Ken Hale: a life in language. Cambridge, MA: MIT Press, 1-52
Crisma, P. 1992. On the acquisition of wh-questions in French. Geneva Generative Papers 1, 115-122
Doron, E. & C. Heycock. 1999. Filling and licensing multiple specifiers. In D. Adger, B. Plunkett, & G. Tsoulas (eds.) Specifiers: Minimalist Approaches. Oxford: OUP, 69-89
Emonds, J. 1976. A Transformational Approach to English Syntax. New York: Academic Press
Gavruseva, E. 2000. On the syntax of possessor extraction. Lingua 110, 772-830
Johnson, K. 1996. In Search of the English Middle Field. Ms. Umass
Kayne, R. 1994. The Antisymmetry of Syntax. Cambridge, MA: MIT Press
Nissenbaum, J. 2000. Investigations of Covert Phrase Movement. Doctoral dissertation, MIT
Oka, T. 1989. On the Spec of IP. Ms, MIT
Platzack, C. 2001. Multiple Interfaces. In N.U. & E. van der Zee (eds.) Cognitive Interfaces: Constraints on Linking Cognitive Information. Oxford: OUP, 21-53
Poletto, C. & J.-Y. Pollock. 1999. On the left periphery of Romance questions. Talk presented at the Workshop on the Cartography of Functional Projections, Pontignano, November 1999
Radford, A. 1986. Small children's small clauses. Bangor Research Papers in Linguistics 1, 1-38
Richards, N. 1998. The Principle of Minimal Compliance. LI 29, 599-629
Rizzi, L. 1993. Some notes on linguistic theory and language development: the case of root infinitives. Language Acquisition 3, 371- 393
Rizzi, L. 1997. The fine structure of the left periphery. In L. Haegeman (ed.) Elements of Grammar: a handbook in generative syntax. Dordrecht: Kluwer, 281-337
Saito, M. 1989. Scrambling as semantically vacuous A-bar movement. In M.R. Baltin & A.S. Kroch (eds.) Alternative Conceptions of Phrase Structure. Chicago: University of Chicago Press, 182-200
Sportiche, D. 1996. Clitic constructions. In J. Rooryck & L. Zaring (eds.) Phrase Structure and the Lexicon. Dordrecht: Kluwer, 213-276
Szabolcsi, A. 1997. Strategies for scope taking. In A. Szabolcsi (ed.) Ways of Scope Taking. Dordrecht: Kluwer, 109-154
Tada, H. 1993. A/A-bar Partition in Derivation. Doctoral dissertation, MIT
ABOUT THE REVIEWER:
ABOUT THE REVIEWER: Jonny Butler is a member of the Dept. of Linguistics at the University of York, UK. He is freshly doctored, his thesis covering quantification, tense, aspect, modality, and phase theory. From November 2004 he will be a postdoctoral researcher at the Graduiertenkolleg "Sprachliche Repräsentationen und ihre Interpretation" ("Linguistic Representations and Their Interpretation") at the University of Stuttgart.