Date: Sun, 7 Nov 2004 07:52:16 -0800 (PST) From: Monika Basic Subject: Syntax and Semantics of the Left Periphery
EDITORS: Lohnstein, Horst; Trissler, Susanne TITLE: Syntax and Semantics of the Left Periphery SERIES: Interface Explorations 9 PUBLISHER: Mouton de Gruyter YEAR: 2004
Monika Basic, unaffiliated scholar.
This book contains 13 papers presented at the annual meeting of the German Linguistic Society in Mannheim in 2002, together with an introductory chapter.
The introduction by Horst Lohnstein and Susanne Trissler entitled ''Theoretical developments of the left periphery'', opens up the discussion by giving a very concise historical overview of the theoretical developments regarding the clausal left periphery. Establishing functional categories as core elements of sentential structure paved the way for fruitful research regarding properties of clauses, and ultimately a more fine-grained decomposition of different functional categories. The left periphery came to be seen as determining the illocutionary force of the sentence (declarative vs. non-declaratives) and hosted projections linked to the discourse, such as force, topic and focus. On the semantic side, researchers either focused primarily on the semantic representation of different sentence types, or tried to develop a sentence mood theory that would take into account both their syntactic and semantic aspects.
This chapter also includes brief summaries of each of the other papers in the volume, giving us a taste of what is to come. The various contributions by and large deal with German, but occasionally cross- linguistic data is addressed where it bears on the issue of universal characterization of the left periphery.
Katrin Axel's paper ''The syntactic integration of preposed adverbial clauses on the German left periphery: a diachronic perspective'' looks at the properties of preposed adverbial clauses (preAC) in the history of German. In Old, Middle and Early High German documents, adverbial clauses can precede the initial constituent of the declarative sentence, thus violating the verb-second constraint (V2). These verb-third constructions disappear at the end of the Early New High German (ENHG)period. Axel argues that the verb-third effects are due to the impossibility of generating preAC in sentence internal position. She assumes that preAC in OHG, unlike their modern counterparts are not syntactically embedded, but rather adjoined at the left periphery. Regarding the position of the adjunction, she argues that it is the leftmost edge of the C-domain, based on the observation that preAC precede certain illocutionary particles. While this could suggest that they are extrasentential, it proves to be very difficult to find unambiguous evidence for or against such a claim. The order 'preAC-V finite' which is also attested in OHG, could indicate that the preAC is either in the position within the left periphery or adjoined to it, and thus does not provide strong evidence for sentence- internal placement of preAC. It is somewhat surprising then that Axel takes the major rise in the use of precisely this pattern in ENHG as ''strongly suggesting that adverbial clauses became syntactically embedded'' (p41), making it possible to base- generate them in the specifier of CP. There seems to be some correlation between 'preAC-V finite' pattern and the availability of V-first in declaratives, although this point has not be particularly stressed. Axel informs us that main clauses of 'preAC- V finite' constructions could appear as V-first declaratives in independent contexts (which is predicted if preAC are extrasentential in this case). It would be interesting to see what the status of V- first declaratives in ENHG was (in MHG both 'preAC-V finite' and V-first patterns are absent).
Searching for the reasons why adverbial clauses came to be syntactically integrated in the period of ENHG, she evaluates and rejects two previous proposals. Unfortunately, a more promising alternative is not worked out.
Josef Bayer's ''Decomposing the left periphery - Dialectal and cross- linguistic evidence'' brings into play cross- linguistic data in order to show that the CP domain is more articulate than it is apparent from the surface form. The basic idea is that there is a uniform set of features associated with the structure of certain clause types, but languages differ in the feature content of the available lexical items, which in turn leads to a different phonetic spell-out of the same feature structure. The main evidence put forward comes from substandard varieties of German and Dutch with doubly filled COMPs (i.e. both C and SpecCP contain lexical material), and the form of the CP-recursion in substandard Dutch (i.e. constructions that seems to have doubly filled C position). In addition comparison is drawn between Western languages and certain East- Asian languages. Bayer argues that the full feature specification of the interrogative clause includes the feature of a wh- expression [wh], a subordination feature [C], and a feature for disjunctivity [disj]. The latter feature is not simply stipulated, but Bayer attempts to derive it from the semantics of questions. As a result, the left periphery expands into DisjP, CP and IP projections. In languages like English which obey the Doubly-Filled-Comp Filter (DFCF), the wh-expression is specified for [wh,disj,C], and thus is able to identify C and preclude the insertion of a complementizer. In languages which do not respect the DFCF, a wh-item is unspecified for the categorial feature C and a separate lexical item is merged, giving rise to a situation where both a wh-expression and a complementizer are present.
Things get a bit more complicated when we turn to cases of CP recursion, where a wh-phrase, and complementizers 'of' and 'dat' can all occur at the same time. Here we would expect 'of' to be specified for [disj], 'dat' for [C], and a wh-phrase to be unspecified for both. However, in order to ensure that the wh-phrase moves to the edge of the clause, Bayer assumes that 'of' carries an optional EPP feature which needs to be checked, and in order to make sure that it is checked by a wh-phrase he assumes that the wh-elements carry [disj] feature after all, which is deleted after being checked by 'of'. The assumption that these wh-phrases have [disj] feature is supported by the observation that, unlike their German counterparts, Dutch wh-phrases can't be used as indefinites. I must admit I was left confused here as up to this point I thought that German wh- phrases did carry this feature, except possibly in case of Bavarian 'wos' ('what') which is argued to be radically underspecified. Several other points could have also been made clearer. For instance, why is it necessary to merge 'of' at all if Dutch wh-phrases have a [disj] feature? Since 'of' does not have [C] feature in cases of CP-recursion, what checks [C] in cases where 'dat' does not occur, such as example 16 (p 65): Hij weet hoe of je dat moet doen. he knows how if you this must do
In the second part of his paper Bayer turns to representation of illocutionary force in root clauses. After pointing out several problems with the assumption that I-to-C movement is a device for establishing force, Bayer concludes that although verb movement enables finite clause types to activate force features, it is not sufficient for determining force.
Ellen Brandner also explores V2 as a Force marking mechanism in her contribution ''Head-movement in minimalism, and V2 as Force-marking.'' According to Brandner, certain languages (e.g. Korean) have the option of lexically encoding force values through the insertion of appropriate particles, while other languages have to resort to specific syntactic configurations. Brandner argues that this is the case in Germanic V2 languages where both verbal and phrasal movement are required for clausal typing. It is important to note that there is no feature checking of the usual type in Brandner's system; instead a verb acquires a feature through the mechanism of dynamic spec-head agreement when a functionally marked phrase moves to its specifier. A further distinction is made between languages that explicitly mark all Force values (e.g. particles in Korean, V2 in Germanic); and languages where only the deviation from the declarative is specified (e.g.Japanese and English).
The rest of the paper is dedicated to the analysis of verb- initial structures in German. Here the construction is left underspecified with respect to clause typing, which is as Brandner points out, a desired outcome as it opens up the possibility for wide range of interpretations. As a result, verb-initial structures can be ambiguous between interrogative and declarative reading. Thus she manages to tackle some arguments against V2 as a clause typing strategy that Bayer raises towards the end of his paper. The solution however cannot be applied in Bayer's system, as he assumes crucially different feature checking mechanism.
The paper by Franz-Josef D'Avis, ''In front of the prefield- inside or outside the clause?'', brings us back to the issues regarding the level of integration of adverbial clauses in German, this time from the synchronic perspective. D'Avis focuses on a particular class of adverbial clauses, namely conditionals of irrelevance (ICs),which are characterized semantically as being related to the proposition of the main clause but irrelevant for the truth of this proposition. Interestingly, these clauses can precede the initial constituent of V2 clauses. This position is normally unavailable for standard cases of adverbial clauses. Moreover, unlike standard adverbial clauses, ICs are excluded from filling the prefield position (i.e. they cannot immediately precede the verb in V2 constructions). D'Avis thus concludes that ICs are not syntactically integrated into the main clause and should be treated as a kind of parentheticals. On the other hand, he suggests that there must be a kind of semantic/pragmatic integration, since standard ICs remain related to the proposition of the main clause.
In order to help him classify ICs, D'Avis considers the corresponding data from Swedish. There are significant differences between Swedish and German ICs, which bring to light certain problems with the parenthetical hypothesis. In Swedish, unlike in German, all verb-final ICs are allowed in the prefield, while V1 and V2 ICs are grammatical only in front of the prefield. He takes this to indicate that verb-final ICs in Swedish are in fact subordinated. The situation is rather more complicated in German where a certain type of ICs ((auch)wenn-ICs) behave like their Swedish counterparts in that they are possible both in the prefield and in front of it. He concludes that 'wenn...auch' clauses are special because they need a concessive adverbial in the middlefield of the main clause. However, the reasons for this remain obscure, as well as the reasons for why Swedish classifies ICs differently from German. The semantics also does not fall into place; ICs seem to share semantic interpretation with other adverbial clauses, which is problematic under the assumption that they are not syntactically integrated.
When all is summed up, the data are interesting and clearly presented, a possible solution is suggested (in the theory of parentheticals, although the theory itself is unfortunately not sketched out), the challenges are outlined, and the answers are left for future research.
Jan Eden's paper, ''Uniformity and Variation: On the relation of wh-phrases and sentence moods in German'' argues for a unified treatment of wh-phrases in different sentence types. In feature-based theory, the class of wh- phrases is divided into subclasses with distinct feature contents (in addition to [+wh], [+relative] and [+exclamative]is needed) in order to account for their varying behaviour and interpretation. Eden argues that this redundancy could be avoided by assuming a unified representation for all wh- phrases and deriving the differences in the behaviour of the constituent questions, relatives and exclamatives from factors other than the feature makeup of a wh-phrase. Three main factors are assumed to determine sentence mood:
(i) verbal mood , (ii) the occupation of a specifier of a mood phrase, and (iii) the occupation of a head of a mood phrase by a finite verb or a complementizer.
Constituent questions constitute the basic mood of all wh-constructions, which can be modified as a semantic object by means of embedding, lexical items and intonation. The difference between an embedded question and a wh- relative is argued to depend on the immediately dominating category: embedded questions are dominated by a VP, while wh-relatives are dominated by a lexically empty DP. The concept of extended partition lies at the core of all wh-constructions, i.e. they all represent a range of possibilities. The feature [+wh] is thus retained in order to trigger obligatory wh-movement to the left periphery in German, but it becomes a syntactic reflex of specific semantic representation of wh-phrases.
The paper by Werner Frey, ''Notes on the syntax and the pragmatics of German Left Dislocation'', offers an analysis of German Left Dislocation constructions, investigating both their syntactic and information- structural properties. Frey first shows that one of the distinctive properties that distinguishes German Left Dislocation (GLD) both from English Left Dislocation constructions and from the German Hanging Topic Left Dislocation (HTLD) is its behaviour with respect to the binding phenomena: GLD shows binding effects, HTLD does not. Once this property has been established as a criterion for unambiguously differentiating the two constructions, Frey goes on to argue that GLD and HTLD differ in their information- structural properties as well. Although standardly both constructions were considered to be topic marking , Frey claims that this is true for GLD only, after investigating the position that the resumptive pronoun occupies in the two constructions. Regarding its discourse properties, Bayer shows that GLDed phrase picks out a referent which is already available in the discourse and it signals the shift of sentence topic. GLD crucially differs in these properties from the HTLD and verb- second clauses.
From the point of view of syntax, it is demonstrated that GLD shows sensitivity to islands. After a brief critical discussion of the two recent movement analysis, Frey argues for an approach where GLDed phrase is base-generated in the position adjoined to the CP projection, from which it dominates the base position of the RP. The adjunction approach is called upon in order to avoid certain island violations. It took me quite some time to work through the actual analysis though. The following assumptions appear to be required:
(i) no islands can intervene between surface and base positions of the RP and the dislocated phrase, (ii) adjunction voids islandhood, and (iii) GLD is basically a root phenomena and can only be adjoined to complement clauses of certain class of verbs (bridge verbs).
Curiously, the last assumption which seems to be crucial in accounting for virtually all examples in this section, is first brought up and explained in a footnote (footnote 14), which we are referred to later. In my opinion, the section would benefit a lot if this point is stressed early on, and having a couple of labelled bracketing to illustrate the analysis and the position of the constituents would also help. It would further be nice to see how the binding effects are captured in more detail (they are given one paragraph only), particularly when we have in mind that binding phenomena constitute the main criteria for differentiating GLD from other relevant constructions.
In their paper ''Inflectional morphology and sentence mood in German'', Horst Lohnstein and Ursula Bredel argue for a theory in which the elements of the verbal inflectional system and syntactic processes of phrasal and head movement to the left periphery conspire to yield different sentence moods in German. Looking at the inflection morphemes ''schwa'' and ''-t'' they assign them an invariant core meaning intended to cover their various uses. This meaning is then relativized with respect to various dimensions of deictic interpretation; namely person, time and world. While the person dimension is linked to the predication structure, time and world dimensions (which determine tense and mood) are linked to the propositional structure. Thus, in indicative present and if related to the predication, ''schwa'' marks the speaker and ''-t'' some other individual. If related to the proposition, the two markers are in complementary distribution, discriminating indicative preterite (marked by ''-t'') and subjunctive present (marked by ''schwa'') In subjunctive preterite the two markers appear together. The assigned meanings are then used to derive some general properties of different sentence moods. The syntactic processes of head and phrasal movement to the functional projection M(ood) P contribute to the determination of the sentence mood, together with the information provided by the inflectional morphology. If the verbal mood is ind pres/pret or sub pret, the occupation of the SpecMP by +/- wh-phrase determines whether the resulting sentence mood will be a wh-question or a declarative. We end up with a theory which derives sentence moods in a strict compositional fashion.
The analysis is sound and well-presented, especially considering the amount of theory they manage to squeeze into this article.
Uli Lutz's paper ''ET, parasitic gaps, and German clause structure'', discusses properties of the construction known as emphatic topicalization (ET),only found in southern German and its consequences for the left periphery. The basic form of ET such as ''Den Hans, wann i derwisch!'' ('the Hans, if I catch!') can also be integrated into the left periphery of a V2 main clause. Furthermore, it is possible to have a gap in both the adverbial and the matrix clause, which is commonly analyzed as a parasitic gap construction.
The paper starts with a descriptive overview of the properties of ET constructions and their parasitic gap variants, illustrating that the two constructions show a number of similarities, including the choice of the complementizer and the categorial options regarding the dislocated constituent. However, there are also important differences, such as
(i) only ET construction allows a variety of constituents to appear between the adverbial clause and the verb in the second position, (ii) the quantified DP takes scope over the matrix clause only in the parasitic gap variant, and (iii) apparently only ET construction is compatible with wh-clauses.
In the rest of the paper, Lutz discusses the consequences that the analysis of these constructions has for the organization of the left periphery in German verb-final clauses in terms of traditional topological approaches, the standard generative analyses, as well as a Rizzi-style view of the left periphery. He concludes that ET and its parasitic gap variant require modifications of the usual assumptions regarding German clause structure. For the ET construction, he assumes that an optional discourse related feature of C ([et]-feature) needs to be checked by an appropriate phrase in the specifier position, thus giving rise to structure with multiple specifiers. On the other hand, the parasitic gap variant is best analyzed with the real gap in the matrix clause, and the parasitic one in the adverbial clause, a conclusion he reaches after evaluating several alternative analyses. The resulting structure calls for a re-evaluation of the standard assumption that only one constituent can precede the verb in German V2 clauses.
The analysis is not without problems, and Lutz points to some of them himself. However, since the aim of the paper was not to give a detailed analysis of these constructions, but rather to discuss their properties, and to ''evaluate some of its consequences for the structural organization of the left periphery'' (p 265), we can disregard its shortcomings, and admit that the paper does achieve its aim.
In his contribution, ''Verb position, verbal mood and the anchoring (potential) of sentences'', Andre Meinunger takes a look at subordinate clauses with verb-second in German. Meinunger starts by identifying classes of predicates that allow for V2 in subordinate clauses. Drawing a parallel between embedded V2 in German and the possibility of root transformations in subordinate clauses in English, Meinunger claims that the decisive factor is assertivity, i.e. root transformations (and thus V2) are possible where the embedded sentence conveys an assertion. Consequently verbs that easily introduce an assertion will allow for embedded V2. A further correspondence is drawn between German and Romance where Meinunger makes the claim that exactly those predicates and grammatical phenomena that block V2 in German embedded clauses trigger subjunctive mood in Romance. Assuming that clauses marked with subjunctive mood introduce possible worlds and are thus not evaluated with respect to the real world, they do not make any assertions about it either. The same is true of verbs that ban embedded V2 in German, such as factive predicates which are presuppositional, and thus cannot be asserted. Therefore, the assertive character of the embedded proposition is taken to be the key factor in allowing for V2 in subordinate clauses.
Meuninger further claims that embedded V2 clauses are interpreted as complement clauses if they stay in their base position, or as an assertion made by a speaker if they attach very higher in the structure, in which case the interpretation of variables depends on the context. Semantically, he postulates the existence of an ASSERT operator which syntactically occupies the highest position in the sentence structure. An embedded V2 clause undergoes movement from its base position to the position where it is in the immediate scope of the illocutionary operator, anchoring the proposition to the actual world and the speaker. This gives rise to a double-access reading where V2 clauses are interpreted first in their base position, and secondly in the derived position where they license their assertive illocutionary force. The paper is clearly written and presented, and particularly interesting is the observed parallel between Germanic and Romance.
In the next two papers we take a break from Germanic languages to look at the left periphery of Romance and Slavic. The article by Hans-Georg Obenauer, ''Non-standard wh-questions and alternative checkers in Pagotto'', explores structural and interpretative properties of non-standard questions in the North-Eastern Italian dialect Pagotto. The claim is that non-standard questions should not be treated as non-canonical uses of standard questions; rather UG makes available structural means for distinguishing the two question types. In Pagotto standard questions, non- bare wh- phrases move to sentence initial position, while bare wh- phrases stay in-situ. The interesting property of non- standard questions then is that their bare wh-phrases appear sentence-initially. Adopting a Rizzi- style highly articulated structure of the left periphery,
Int(errog.)ForceP > G(round)P > Op(erator)P > Top(ic)P > IP
Obenauer argues that this happens because they target designated functional layers on top of the InterForceP. Identifying three types of non-standard questions, namely:
(i) surprise/disapproval questions, (ii) rhetorical questions, and (iii) ''I-can't-find-the-value-of-x'' questions,
Obenauer claims that in these cases wh-phrases move beyond IntForceP to specifiers of SurprP, RhetP, and cfvP respectively. The specific interpretative properties are associated with each functional projection, and movement of a wh-phrase in non- standard questions has specific semantic effects.
Obenauer also investigates the role of so-called ''alternative checkers''. These semantically bleached, grammaticalized lexical elements can bear the relevant features and thus take over the checking function of wh- movement. Obenauer argues that such elements can be identified for each of the non-standard interrogatives. The paper is carefully argued with the data clearly set out.
Kerstin Schwabe focuses on Slavic clitic 'li' as a marker of interrogativity and focusation in her contribution ''The particle li and the left periphery of Slavic yes/no interrogatives''. Schwabe first embarks on a difficult task of systematizing the various distributional properties of the clitic 'li' in different Slavic languages. Two version of 'li' are distinguished:
(i) V-li is associated with the verb, (ii) XP-li is associated with the focused marked XP.
Within the framework of Rizzi's (1997) split C-domain, Schwabe argues that in Russian and Serbian/Croatian 'li' is always generated in the head projection of the ForceP and acts as a kind of complementizer. On the other hand, in Bulgarian and Macedonian 'li' seems to act as a marker of a clause type and is independent of ForceP. It is either adjoined to V, or in case of XP-li, it is adjoined to the Foc-marked constituent and moves up together with it to FocP. Different syntactic position for V-li in Serbian/Croatian on the one hand, and Bulgarian and Macedonian on the other, is motivated by the behaviour of non-finite verbs in polarity questions; namely in Serbian/Croatian non-finite verbs cannot act as hosts for the clitic. The syntactic analysis of polarity questions is followed by semantic representation of interrogative and focus features.
Although I found it interesting that the difference in the ability of non- finite verbs to host clitics was attributed to different syntactic placement of 'li' rather than the ability of non-finite verbs to move high enough in Serbian/Croatian (as for example, in the analysis of Boskovic (2001) who gives evidence that non-finite verbs cannot raise to C-domain in these languages), the actual implementation of this idea left me somewhat confused. In Schwabe's analysis, the verb in fact moves as high as possible, taking all the clitics along. Most of the work is then done in the phonology where a number of operations and constraints interact to yield a licit derivation. Two key processes are cliticization and phonological inversion (PI), which places the clitic after the first phonological boundary. Putting aside the problems that arise with PI which are well known in the literature (see Boskovic 2001, Progovac 1996), it is unclear to me how these two processes are ordered with respect to each other in a way that would yield a desired output. In case of a finite verb acting as a host, PI of 'li' precedes cliticization. When the verb is non- finite, it is crucial that the verb first forms a clitic phrase with the auxiliary clitic, after which the PI of 'li' cannot cannot cross thus created two phonological boundaries and the derivation is ruled out. The problems do not end here because 'li' can intervene between the full form of the auxiliary or complementizer and the clitic cluster. The conclusion is that cliticization must precede PI if there is an auxiliary clitic and the potential host is a participle. ''This may be determined by morphological means'' (p 405). I remain perplexed.
The paper by Arnim von Stechow, ''Binding by verbs: Tense, Person and Mood under Attitudes'' analyzes certain temporal and modal phenomena in constructions with verbs of attitude. Person, mood and tense are seen as features of the verb which are checked by the corresponding arguments of the verb, namely an individual, a world and a time variable. At Logical Form (LF), the features of the variables are interpreted and non- interpretable features are deleted. The central idea of the theory is that features of semantically bound variables are deleted and therefore not interpreted. For instance, in the following example, ''Only I did my homework'', 'I' is interpreted as a variable with the interpretable feature 1st person. This feature is projected to the only-DP which being a generalized quantifier binds the relevant variables, namely the trace of 'I' and the variable corresponding to the possessive pronoun. As a result of binding, these features are not interpreted which enables the variables to range over the entire domain of individuals, a desired outcome. The idea then is that verbs of attitude are quantifiers that bind a person, a world and a time variable simultaneously, and in that way the features of the bound variables get deleted. Temporal adverbs are taken to be binders as well. They semantically bind the time argument of the verb they modify.
The theory is then applied to a broad range of empirical phenomena, such as the derivation of the sequence of tenses rules in English and Russian, the behaviour of temporal adverbs, modals, as well as subjunctive in German. The analysis is carefully elaborated, and the number of phenomena it is applied to is really impressive.
The final paper, ''Complementizer selection and the properties of complement clauses in German'' by Angelika Wollstein addresses the difference between finite and non- finite complement clauses in German with respect to the presence of lexical complementizers. The lack of complementizer with non-finite clauses, Wollstein derives from the referentiality of a clause. Verbal categories are argued to have referential properties by default. Assuming that only referential categories extend to functional projections, the verb extends to a single projection, labelled FP. The principle of visibility ensures that the head of the highest functional projection is made visible by moving a finite verb to F-head or inserting a lexical complementizer. Non-finite complement clauses do not have overt material in the left periphery, which leads to the conclusion that they do not project the relevant functional structure at all. Analysing the three types of infinitival forms in German, namely the bare infinitive, the 'zu' infinitive, and the past participle, Wollstein argues that the infinitival marker 'zu' is able to block the referential properties of verbs. As a result, no functional structure is projected, and consequently no position that would host a lexical complementizer. Other infinitival forms can be referential and thus can display certain root properties. The discussion is extended to include the expression of sentence mood. Assuming that sentence mood is specified in the left periphery, it is predicted that non- finite complement clauses which cannot be functionally extended (i.e. zu- infinitival clauses) are unspecified with respect to sentence mood.
Overall, this is an excellent volume. It reveals the challenges put before us when trying to provide syntactic encoding and semantic representation of certain discourse related phenomena. The various contributions do a great job of bringing together the syntactic, semantic and pragmatic aspects of the constructions in question. The theories and solutions offered are at times very intricate, but they are certainly worth our time and attention.
Boskovic, Z. 2001. On the nature of the syntax-phonology interface: Cliticization and related phenomena. Amsterdam: Elsavier Science.
Progovac, Lj. 1996. Clitics in Serbian/Croatian: Comp as the second position. In A. Halpern and A. Zwicky (ed.) Approaching second: Second position clitics and related phenomena. Stanford, California: CSLI, 411- 428.
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
ABOUT THE REVIEWER:
ABOUT THE REVIEWER Monika Basic has recently received an MA degree in English Linguistics from the University of Tromsoe, Norway. Her thesis focused on issues regarding CP/DP structure and certain constructions related to focus assignment. Her research interests concern particularly syntactic theory, movement, locality, as well as clausal and nominal functional architecture.