EDITOR: Panagiotidis, E. Phoevos TITLE: The Complementizer Phase SUBTITLE: Subjects and Operators SERIES: Oxford Studies in Theoretical Linguistics PUBLISHER: Oxford University Press YEAR: 2010
Jason Ginsburg, Center for Language Research, University of Aizu
This book is an edited collection of nine papers that, from the perspective of the Minimalist Program, examine complementizers and syntactic structures that contain complementizers.
Chapter 1. ''Introduction: Complementizers and Their Phase'' by E. Phoevos Panagiotidis
In this chapter, Panagiotidis introduces the rest of the chapters in the book, stating that the primary focuses of the chapters are on 1) ''the role of complementizers in subject extraction phenomena'' (p. 3), and 2) ''the nature and feature make-up of complementizers, mainly as witnessed by the kind of material they can host in their specifiers'' (p. 3). Panagiotidis summarizes each chapter and explains how they relate to each other, as well as to the main themes of this volume.
Chapter 2. ''On Some Properties of Criterial Freezing'' by Luigi Rizzi
This chapter focuses on what Rizzi refers to as Criterial Freezing. The clause periphery contains criterial positions, which are ''scope-discourse positions'' (p. 18) that attract phrases to their Spec positions. According to the notion of Criterial Freezing (Rizzi 2006), a phrase in a criterial position is frozen in place. A chain can only have a single s-selectional position and only a single criterial position because these positions ''are unique'' and because ''they delimit the chain'' (p. 19). For example, in (7b) ''which candidate'' satisfies the Q criterion in the embedded clause. Thus, it undergoes Criterial Freezing and cannot undergo further movement. Rizzi argues that subjects undergo Criterial Freezing because they appear in the specifier of a functional Subj head, which is a criterial position that expresses what a clause is about.
(7b) *Which candidate does Bill wonder [ t' Q [ you voted for t ]] (p. 20)
Rizzi proposes that subextraction from an XP in a criterial position can occur. He argues that a criterial head agrees with a criterial goal - the head of an XP in a criterial position. However, subextraction of an element from this frozen XP may be possible, as long as the subextracted element is not a criterial goal. Rizzi provides analyses of several constructions in which there is subextraction from XPs, including subjects, that are in criterial positions.
Chapter 3. ''(Non-) Extraction from Subjects as an Edge Phenomenon'' by George Kotzoglou
In this chapter, Kotzoglou focuses on the Subject Condition (Ross 1967, Chomsky 1973, Huang 1982); a ban on extraction from subjects, as in (1b).
(1b). *Who did [a picture of ___] annoy our neighbours? (p. 33)
To account for the possibility of subject extraction in null subject languages such as Greek, as in (7b), Kotzoglou proposes the Restriction on Copy Reduction (RCR) (26), a modified version of ''Chain Reduction'' (Nunes 2004).
(7b) tinos ipes oti [i epimoni ___] ekseplikse tus krites? whose said.2sg that the insistence surprised the judges 'Whose insistence did you say surprised the judges?' (p. 35)
(26) Restriction on Copy Reduction (RCR) (phonological deletion of copies under identity) can apply to at most one pair of copies of an element in each phase. (p. 43)
The RCR boils down to a requirement that only one member of a movement chain can be erased within a strong phase. If there are multiple deleted copies of an element within a single phase, a derivation crashes, which Kotzoglou suggests occurs because ''PF gets conflicting instructions as to the spell-out of one copy'' (p. 42).
The RCR predicts that subextraction from subjects is banned in languages that require [Spec, TP] to be filled; i.e., in non-null subject languages such as English. If an XP is subextracted from a subject, the result is a phase that contains two unpronounced copies. In (1b), the CP phase contains two unpronounced copies of ''who'', one in [Spec, TP] and one in [Spec, vP], thus violating the RCR.
When the RCR is not violated, subextraction may occur. In the null subject language Greek, which the author assumes lacks a [Spec, TP] position that houses a subject, when there is subextraction from a subject, the subextracted phrase does not move through [Spec, TP]. Thus, the CP only contains one copy, in [Spec, v*P], of the subextracted phrase. Furthermore, Kotzoglou argues that subextraction from objects and ECM subjects in languages such as English is permitted because this type of subextraction does not violate the RCR.
Chapter 4. ''Subextraction from Phase Edges'' by Angel J. Gallego
In this chapter, Gallego examines extraction from v*P and CP edges, and argues that movement of an XP to a position in which all of its features are satisfied freezes the XP, as stated in (44).
(44) Movement to a ('maximal') feature checking position yields freezing (p. 74).
With respect to v*P edges, Gallego argues that evidence supports the Activity Condition (Chomsky 2000, 2001), which is the notion that a DP becomes frozen when it receives structural Case and it is phi-complete. Otherwise, it is not frozen and extraction is possible. Gallego compares this view with the Edge Condition, which refers to a hypothesis developed by Chomsky (2008), that ''Syntactic Objects in phase edges are internally frozen'' (p. 56). Gallego presents evidence against the Edge Condition and for the Activity Condition; for example, in Spanish and Dutch, extraction is possible from post-verbal subjects of transitive verbs, even though these subjects are in an edge position ([Spec, v*P]). On the other hand, extraction is marked in these languages when it comes from a pre-verbal subject, which is presumably in a non-edge Case position.
With respect to CP edges, Gallego argues that the evidence supports Rizzi's (2006) Criterial Freezing, which blocks movement of a phrase from a criterial position. Note that this differs from Rizzi's revised formulation of Criterial Freezing, presented in chapter 2 of this book, which allows subextraction from a frozen XP in a criterial position. Gallego argues that an XP that appears to have undergone subextraction from a complex phrase in a CP edge is actually an aboutness phrase that is base-generated outside of the complex phrase, and thus subextraction does not occur. To support this view, Gallego presents ill-formed examples in Spanish in which binding requires a ''non-aboutness pattern'' (p. 70). Gallego also discusses instances in which subextraction occurs from a complex in-situ wh-phrase in English and Spanish. This subextraction is possible because it occurs from an in-situ wh-phrase which has not had its criterial feature checked.
Chapter 5. ''Subjects on the Edge'' by Anna Roussou
In this chapter, Roussou examines the behaviors of complementizers and the subjects of embedded complement clauses. Crucially, Roussou argues that that-trace, raising, and control phenomena result from whether or not a complementizer requires a complement clause that has a pronounced (lexicalized) EPP.
Roussou takes the position that ''there is a single 'to' whose distribution varies depending on its complement'' (p. 81). The lexical item 'to' is a locative which occurs in the lower left periphery of a clause and that can take either a DP complement (the preposition 'to') or a verbal complement (the infinitival 'to'). Roussou argues that the English 'to' introduces ''a variable corresponding to the EPP'' (p. 83) in a manner akin to the Italian affix '-re'. A crucial component of this proposal is that there is no null subject in control and raising constructions. Rather, following Manzini and Roussou (2000), a DP is Merged directly in its surface position. The EPP variable associated with 'to' gets its interpretation from another DP or a generic operator in control and raising constructions.
Just as there is a single 'to', Roussou argues that there is a single nominal 'that'. The complementizer 'that' ''takes a propositional complement'' (p. 98), the demonstrative 'that' takes an NP complement, and the pronominal 'that' ''appears without a restriction'' (p. 98). When 'that' takes a propositional complement, it requires what Roussou refers to as ''a lexicalized EPP'' in its complement, as formulated in (21).
(21) 'That' requires a lexicalized EPP position that closes off the proposition. If this requirement cannot be met, 'that' has to be absent (p. 94).
(22a), which demonstrates a typical that-trace effect, is ill-formed because 'that' requires a propositional complement with a lexicalized EPP. In (22b), there is no EPP present in the embedded clause (Roussou assumes that 'who' is base generated in its surface position) and there is no 'that' present that requires a lexicalized EPP. There is a variable associated with the embedded predicate that is bound by the wh-operator of 'who'.
(22a) *Who do you think that left? (Adapted from p. 94)
(22b) Who do you think left? (Adapted from p. 94)
Roussou also extends this analysis to 'for', 'if', and 'whether'. She argues that 'if' and 'for' take propositional complements and thus, require there to be a lexicalized EPP in a complement. The distribution of 'whether' and 'if' differs in that 'if' requires a finite complement, since a finite complement has a lexicalized EPP, but 'whether', being an XP, does not require a finite complement, and thus does not require, although can occur with, a lexicalized EPP.
Chapter 6: ''On the Necessity of Phi-features: The Case of Bavarian Subject Extraction'' by Clemens Mayr
This chapter develops an account of subject-object asymmetries that relies on the idea that an extracted element (e.g., subject or object) must be along a projection line of an attracting probe, and that a subject is not along this projection line unless it forms a phi-feature agreement relation with its local head. This proposal accounts for subject-object asymmetries, exemplified by the that-trace effect in English, and the lack of a subject-object asymmetry in Bavarian.
Mayr's proposal hinges on what he refers to as a ''projection line'', which refers to the main spine of a derivational tree. A functional head F is in a projection line with an object goal O if ''all the heads from F down to O stand in selectional relations with their complements'' (p. 130). A subject, which is in a specifier position, is outside of the main projection line ''because it is not directly selected for by any of the heads within the clausal projection'' (p. 132). However, Mayr proposes that if a subject undergoes phi-feature agreement with its local head, then it becomes part of the projection line of the local head.
These proposals account for object extraction. In English and Bavarian, an object can undergo wh-movement, and in Bavarian an object can also undergo topicalization, regardless of whether or not there is a complementizer. This is expected since the object is in the same projection line as a probe (e.g. a probe in C) that attracts it, and thus no phi-feature agreement is required to put the object in the relevant projection line.
These proposals also account for the (non)-availability of subject extraction. In Bavarian, a subject can be extracted (wh-movement or topicalization) only if a local complementizer undergoes phi-feature agreement with the subject, thus putting it into the projection line of an attracting probe. In English, on the other hand, as exemplified by the that-trace effect, 'that' is unable to undergo phi-feature agreement with a local specifier. When 'that' is present, a subject wh-phrase remains outside of the projection line of a probe in a higher clause. When 'that' is absent, Mayr takes the position that there is no CP in the relevant clause. The subject moves directly from [Spec, TP], which is possible because the subject undergoes phi-feature agreement with T, thus putting it in the same projection line as the attracting probe.
Chapter 7. ''Apparent Hyper-raising in Brazilian Portuguese: Agreement with Topics Across a Finite CP'' by Ana Maria Martins and Jairo Nunes
In this chapter, Martins and Nunes develop an account of apparent hyper-raising constructions in Brazilian Portuguese (BP). In this type of construction, as in (3b), there is a resumptive pronoun in a finite embedded clause that is coreferenced with a higher raised subject in a matrix clause.
(3b) Os meninos parecem que eles viajaram ontem the boys seem-3PL that they travelled-3PL yesterday 'The boys seem to have travelled yesterday.' (p. 145)
The authors develop the following analysis. The lower pronoun appears in the embedded [Spec, TP] and the higher coreferential subject is a topic that originates in [Spec, TopP] of the embedded clause. The matrix T agrees with the embedded topic and attracts it to the matrix [Spec, TP] where it satisfies an EPP feature. Martins and Nunes argue that this is not a problem for the Phase Impenetrability Condition as formulated in Chomsky (2001), whereby the complement of an embedded CP phase head is sent to Spell-Out ''only when the next strong phase head is introduced in the derivation'' (p. 158). In a construction such as (3b), Spell-Out of the complement of the embedded C does not occur until the matrix C is Merged, since the vP of raising verbs is not a strong phase. Therefore, the topicalized phrase in the embedded clause is accessible to the matrix T. The authors go on to account for the unique properties of the raised subject, which behaves like a topic, as a result of being base generated in TopP, and which also behaves like a subject, as a result of raising to an A-position, [Spec, TP].
Chapter 8. ''The Structure and Interpretation of (Romance) Complementizers'' by M. Rita Manzini
This chapter focuses on what Manzini refers to as ''the Romance che-type'' complementizer, exemplified by the Italian 'che'. Manzini argues that this type of complementizer, which can function as either a complementizer or a wh-phrase, is an NP head. The wh-phrase version of 'che' introduces a variable that ''ranges over individuals'' (p. 170), while the complementizer 'che' introduces a variable that ''ranges over situations/possible worlds'' (p. 170). Thus, there is only a single complementizer which differs with respect to the type of variable that it introduces. In a sentence such as (1), Manzini takes the position that the matrix verb takes an NP complement that is headed by 'che', and 'che' takes the embedded sentence as its complement. The wh-phrase 'che', as in (3), is argued to be an NP that appears in an operator position above C.
(1) So che fai questo I.know that you.do this 'I know that you do this.' (p. 169)
(3) Che fai? what you.do? 'What are you doing?' (p. 169)
Manzini provides a variety of examples from different Italian dialects to elaborate on the structure of the phrase housing a che-type complementizer. In some dialects, a clausal complement of a che-type complementizer can have a V2 structure. Manzini assumes that this verb is in C, thus indicating that the complementizer must be in a higher position; the complementizer is a noun with a clausal complement that contains a verb in C. In addition, Manzini demonstrates that in some Italian dialects, a che-type complementizer can be preceded by a wh-phrase, as well as by other topicalized phrases. This is taken to be evidence that the complementizer has a left periphery of its own.
Manzini argues that her approach is preferable to a cartography-style approach, such as Rizzi (1997). Manzini's approach accounts for the same data as cartography approaches. In addition, whereas a cartography approach requires ''a hierarchy of functional categories'' (p. 184), in Manzini's approach, only a single complementizer position is required. An elaborated left periphery results from ''the recursion of predicate-argument structures'' (p. 184).
Manzini also extends her approach to account for other types of complementizers, arguing that although the che-type complementizer is nominal and is associated with a wh-element, there is no requirement of the language faculty that all complementizers be like this. For example, in Buru, a complementizer appears to be a verbal element.
Chapter 9. ''Nested Interrogatives and the Locus of Wh'' by Omer Preminger
In this chapter, Preminger examines Hebrew nested interrogatives. In (3b), one wh-phrase moves to the edge of the embedded clause and the other moves to the edge of the matrix clause, thus giving a structure with ''multiple interrogative clauses nested within one another'' (p. 202).
(3b) [ma]_2 Dina saxexa [le-mi]_1 Dan natan t_1 t_2? what Dina forgot DAT-who Dan gave '[What]_2 did Dina forget [to whom]_1 Dan gave t_2 t_1? (p. 201)
Preminger accounts for the formation of nested interrogatives as follows. There is a FocP projection which houses wh-phrases, and that occurs below a CP projection which houses a complementizer. Evidence for this FocP comes from the phenomenon of Sub-Complementizer Topicalization in which a topicalized phrase appears in a position below a complementizer. In addition, Preminger proposes that a wh-phrase contains an uninterpretable wh-operator feature and Foc contains an interpretable wh-operator feature. This interpretable wh-operator feature on Foc attracts the higher wh-phrase ('DAT-who' in (3b)) to [Spec, FocP], where the uninterpretable wh-operator feature of the wh-phrase is checked. The higher C contains an interpretable but unvalued operator feature that attracts the lower wh-phrase ('what' in (3b)) to [Spec, CP]. Crucially, since the operator feature on C is unvalued, the uninterpretable wh-operator feature on the wh-phrase remains active. Thus, the wh-phrase undergoes further movement to a higher [Spec, FocP], where a valued wh-operator feature of Foc checks the corresponding operator feature of the wh-phrase. Preminger explains how this analysis accounts for superiority effects that result when there are crossing wh-movements, and for wh-island effects in cases of long wh-movement which involve ''movement of a constituent to the periphery of a clause outside of the one where it was base-generated'' (p. 208). In addition, Preminger explains how this analysis is compatible with the Phase Impenetrability Condition (Chomsky 2000, 2001).
Chapter 10. ''Complex Wh-phrases Don't Move: On the Interaction Between the Split CP Hypothesis and the Syntax of Wh-movement'' by Jereon Van Craenenbroeck
In this chapter, Craenenbroeck proposes an analysis of wh-questions in which a complex wh-phrase is base generated in its surface position. Craenenbroeck assumes that there is a CP_1 projection that is responsible for clausal typing (akin to the ForceP of Rizzi (1997)). In addition, there is another lower projection, CP_2 (akin to the FocP of Rizzi (1997)) that hosts an operator. In a wh-question, the higher C_1 has a Question feature [+Q] and the lower C_2 has an operator feature [+Op]. Craenenbroeck argues that a simple wh-phrase undergoes movement from a base position first to [Spec, CP_1], where it checks the [+Op] feature, and then it undergoes further movement to [Spec, CP_2], where it checks a [+Q] feature. A complex wh-phrase, such as ''which boy'', on the other hand, is argued to lack an operator feature, which results in it being Merged directly in the higher [Spec, CP_2] position. Although the complex wh-phrase is Merged directly in its surface position, there is a co-indexed empty operator, base generated in theta-position, that undergoes movement to [Spec, CP_1], where it checks the [+Op] feature of C_1. Note that the general proposal is that a simple wh-phrase moves through CP_1 and lands in CP_2. However, the option of movement to only CP_2 may be available in some languages; in Frisian and dialectal Dutch, Carenenbroeck takes the position that a simple wh-phrase need only move to [Spec, CP_2], from where it can somehow check the operator feature of CP_1. Craenenbroeck demonstrates how this analysis accounts for differences in behaviors between simple and complex wh-phrases in ''doubly filled COMP phenomena in Frisian and dialect Dutch, swiping in English, wh-copying in German, free relatives in Dutch, preposition stranding in Dutch, and spading in dialect Dutch'' (p. 260).
A strength of this book is the variety of analyses that it presents for both well-known and lesser known phenomena. This book provides new analyses of familiar phenomena, such as the 'that-trace' effect, subject-object asymmetries, and asymmetries between complex and simple wh-phrases. It also provides analyses of some lesser known phenomena, such as apparent hyper-raising in Brazilian Portuguese, and nested interrogatives in Hebrew.
Notable is that this book provides differing, and not necessarily compatible, analyses of similar phenomena. This can be seen with respect to subextraction from subjects. Rizzi's chapter 2 accounts for subject subextraction phenomena with the proposal that a subject XP is frozen due to a Subject Criterion, but an element contained within the subject may undergo subextraction. Kotzoglou's chapter 3 accounts for the same phenomena in a very different manner by proposing that there is a general ban on multiple copies being pronounced within a single phase. Gallego's chapter 4, on the other hand, supports Rizzi's original Criterial Freezing position, which bans movement out of an XP that has all of its features satisfied. Subextraction can only occur out of subjects that do not have all features satisfied and some apparent cases of subextraction are really base generated aboutness phrases. Similarly, a variety of analyses are proposed for that-trace effects. Rizzi's chapter 2 suggests that-trace effects result from criterial freezing of a subject, whereas Roussou's chapter 5 takes the position that the 'that' trace effect is the result of a requirement that the complement clause of 'that' have a lexicalized EPP. In addition, Mayr's chapter 6 proposes that the that-trace effect results from the inability of 'that' to undergo phi-feature agreement with a subject. Comparison of these differing proposals regarding subextraction from subjects, that-trace effects, and other phenomena should lead to a better understanding of these and related phenomena.
The collection of papers in this book sheds light on the structures of the phrases that contain complementizers, as well as on the behaviors of complementizers, and the analyses raise a number of interesting questions. This book should be of interest to syntacticians, especially those working in the Minimalist Program, who are interested in complementizers and their projections.
Chomsky, Noam. 1973. Conditions on transformations. In S. R. Anderson and P. Kiparsky (eds.), A Festschrift for Morris Halle. New York: Holt, Rinehart & Winston, 232-286.
Chomsky, Noam. 2000. Minimalist inquiries: The framework. In R. Martin, D. Michaels, and J. Uriagereka (eds.), Step by step: Essays on Minimalist syntax in honor of Howard Lasnik. Cambridge, MA: MIT Press, 89-156.
Chomsky, Noam. 2001. Derivation by phase. In M. Kenstowicz (ed.), Ken Hale: A life in language. Cambridge, MA: MIT Press, 1-52.
Chomsky, Noam. 2008. On phases. In R. Freidin, C. Otero, and M. -L. Zubizarreta (eds.). Foundational issues in linguistic theory: Essays in honour of Jean-Roger Vergnaud. Cambridge, MA: MIT Press, 133-166.
Huang, James. 1982. Logical relations in Chinese and the theory of grammar. Ph.D. dissertation, Massachusetts Institute of Technology.
Manzini, M. Rita and Anna Roussou. 2000. A minimalist theory of A-movement and control. Lingua 110: 409-447.
Nunes, Jairo. 2004. Linearization of chains and sideward movement. Cambridge, MA: MIT Press.
Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of grammar: A handbook of Generative Syntax. Dordrecht: Kluwer, 281-337.
Rizzi, Luigi. 2006. On the form of chains: Criterial positions and ECP effects. In Lisa Cheng and Norbert Corver (eds.), Wh-Movement: Moving on. Cambridge, MA: MIT Press, 97-134.
Ross, John R. 1967. Constraints on variables in syntax. Ph.D. dissertation, Massachusetts Institute of Technology.
ABOUT THE REVIEWER
ABOUT THE REVIEWER:
Jason Ginsburg is an Assistant Professor in the Center for Language
Research at the University of Aizu in Japan. He has a PhD in linguistics
and an MS in Human Language Technology, both from the University of
Arizona, and an MA in TESOL from American University. His research
interests are in syntactic theory (in the framework of Generative Grammar),
computational modeling of syntactic theory, and applications of syntactic
theory and natural language processing for teaching languages.