Research Statement (as of November 2017)

Here is the most recent statement (November 2016) of my research interests, including argument structure, anaphora, negation and African linguistics.

https://www.dropbox.com/s/dl8z5w4wdlmd2i7/Research%20Statement%20%28November%202016%29%20%28distribution%29.pdf?dl=0

Research Statement

Chris Collins

November 2016

            My current research involves three interrelated strands: the syntax of English, the syntax of African languages and foundational issues in syntactic theory. My work in each of these domains informs my work in the others. I begin by summarizing some of my early research.

1.         Argument Structure

            Much of my early work was on argument structure. In particular, I was concerned with showing how to apply Larson’s theory of VP shells to various constructions. In my thesis, I showed how VP shell structures allowed an elegant treatment of serial verb constructions in Ewe (see also Collins 1997 “Argument Sharing in Serial Verb Constructions”). In Collins and Thráinsson 1996 (“VP Internal Structure and Object Shift in Icelandic”), we proposed that the heads of the VP shells should be interpreted as null morphemes and that functional projections could intervene between the VP shells. Both of these proposals have now become standard in the syntax literature. In Collins 1997 (chapter three of “Local Economy”, see also Collins and Branigan 1997 “Quotative Inversion”), I showed how TrP (now called vP) and ApplP could be integrated into one clausal structure and form the basis of an explanation for the transitivity constraint on quotative inversion. Lastly, in Collins 2005a (“A Smuggling Approach to the Passive in English”, see also Collins 2005b “A Smuggling Approach to Raising in English”), I argued that the external argument in the passive should be projected in exactly the same way as in the active (bringing back Chomsky’s 1957 original analysis of the passive). In both cases, the external argument is in Spec vP.

            A recurring theme in this early research is the relation between the theory of locality and the theory of argument structure. Certain assumptions about argument structure (e.g., how the external argument is projected in the passive) have implications for the theory of locality (e.g., how A-movement of a DP can satisfy locality constraints such as Relativized Minimality or the Minimal Link Condition).

This early work set the stage for my later work on the linker in the Khoisan language, as described below. In particular, in my work on the linker in Khoisan I give evidence (in the form of an overt morpheme heading LKP) for the kind of functional projections argued for in Collins and Thránsson 1996.

2.         Imposters

            The monograph “Imposters” (MIT Press, 2012), co-authored with Paul Postal, discussed English expressions such as the present author, which are 3rd person singular, but refer to the speaker. We called these expressions imposters. A natural question is what kind of pronoun (third person or first person) an imposter binds. The paradigm in (1) below gives some relevant data. (1a,b) illustrate imposters binding reflexive pronouns. (1c) illustrates a camouflage DP binding a reflexive pronoun (see Collins, Moody and Postal 2008). (1d) illustrates bound variable anaphora, and (1e) illustrates a pronoun bound by a relative pronoun.

 (1)      a.         In this reply, the present authors (= the writers of the reply)

attempt to defend ourselves/themselves against the scurrilous charges which have been made.

b.         This reporter (= speaker) and his son are proud of ourselves/themselves.

            c.         Your Majesty should praise yourself/herself_.

            d.         Every one of us thinks we/they are a genius.

            e.         I am a teacher who takes care of myself/himself.

            In each example above, there is an alternation between a first/second person pronoun and a third person pronoun, with no change in truth conditions. The question in each case is what determines the person/number/gender features (phi-features) of the pronoun. On the basis of data like that in (1), Paul Postal and I claimed that the phi-features of all non-expletive pronouns are determined by syntactic agreement with a small set of sources (antecedents being the most typical case of a source). In general, we showed how such data has implications for pronominal agreement, the syntactic representation of indexicals, principle C of the binding theory, epithets, and coordinate structures, amongst other things.

            Following this work, I edited a volume “Cross-Linguistic Studies of Imposters and Pronominal Agreement” (OUP, 2014). This volume included studies of imposters by my students and colleagues on languages including Spanish, Albanian, Indonesian, Italian, French, Romanian, Mandarin and Icelandic. The work on imposters is closely related to the work on camouflage DPs (such as Your Majesty), found in Collins, Postal and Moody 2008 (“An AAE Camouflage Construction”).

The work on imposters is a good example of one of the basic premises of my research. Before our monograph, the topic of imposters had been largely untouched. In the monograph, we documented a multitude of new facts, and many new generalizations with direct implications for Universal Grammar (e.g., the principles determining pronominal phi-feature values). Work in generative syntax, even on a well-studied language like English, has just uncovered the tip of the iceberg. The vastness of English grammar results from its combinatorial nature. For example, in the case of imposters, one can ask whether a third person DP, which refers to the speaker, can bind a first or a third person pronoun. Going through the kinds of DPs that can bind pronouns with different phi-feature combinations led to the discoveries in our monograph. It is reasonable to believe that investigating other languages (e.g., Ewe, Sasi, N|uu) to the same depth that English has been investigated would also have important implications for Universal Grammar. Far from approaching the end of syntax, we are only now at the beginning of syntax. 

3.         NEG Raising

            An interesting domain which reveals the tension between syntactic and semantic explanations is NEG Raising. In Collins and Postal (2014) “Classical NEG Raising” (henceforth CP2014), we defend a syntactic view of NEG Raising against the consensus semantic/pragmatic views. Consider the sentences in (2):

(2)       a.         I don’t think this course is interesting.

            b.         I think this course is not interesting.

            One interpretation of (2a) (the NEG Raising interpretation) is equivalent to (2b). In CP2014 we claim that there is a syntactic rule of NEG Raising that derives the following structure for (2a):

(3)       I do NEG₁ think that this course is <NEG₁> interesting.

            In this structure, NEG₁ raises from the embedded clause to the matrix clause, where it appears overtly. The angled brackets indicate an occurrence that is not pronounced. So NEG₁ is pronounced in the matrix clause, but interpreted in the embedded clause.

            An alternative theory (originally due to Bartsch 1973) is that (2b) can be inferred from (2a) and a certain semantic/pragmatic assumption (the excluded middle property). If this were true, there would seemingly be no reason to invoke syntactic NEG Raising in the analysis of (2a). Modern analyses in the same spirit include Gajeweski 2007 and Romoli 2013.

            In CP2014 we give a number of arguments, rich with data, arguing for the syntactic point of view. These arguments concern strict negative polarity items, island facts, Horn clauses and parentheticals. Consider the following example of a Horn clause (named after Larry Horn, who discovered the phenomenon):

(4)       I don’t think that at any time would he steal money from a charity.

            This sentence shows that an NPI can trigger NEG Inversion in the embedded clause. Crucially, CP2014 show that not all NPIs can trigger NEG Inversion. The NPIs that do are a subset of those that CP2014 claim have an underlying unary NEG structure: [[NEG SOME] time]. As discussed in detail in CP2014, such an analysis has implications for the theory of NPIs. It means that consensus theories of NPIs based on the assumption that they are existential/indefinite DPs are untenable. Rather, some NPIs must be analyzed as underlying negative quantifier DPs since only that assumption renders them capable of triggering NEG Inversion. Furthermore, Horn clauses provide support for a syntactic NEG Raising approach (over a semantic/pragmatic approach) to sentences like (2a), since only on a syntactic NEG Raising approach is there a NEG present in the embedded clause of (4) to trigger NEG Inversion.  

This monograph has given rise to a series of five follow up papers investigating various aspects of the theory of negation laid out in CP2014. In two papers under review, we show that the distinction between unary NEG NPIs and binary NEG NPIs has cross-linguistic support from two unrelated languages, Ewe and Serbo-Croatian (see “Negative Polarity Items in Ewe” by Collins, Postal and Yevudey and “NEG Raising and Serbo-Croatian NPIs” by Collins and Postal). In another paper (accepted for publication in Glossa), we (Collins and Postal) investigate how the theory of CP2014 fares in the face of the well-known Klima tests for sentential negation (“Interclausal NEG Raising and the Scope of Negation”). In a paper for a volume edited by Larry Horn and Ken Turner, we (Collins and Postal) investigate NEG Raising phenomena in the complement of non-factive know and similar predicates (“Dispelling the Cloud of Unknowing”). Lastly, in “Not even” (2016), I show how the framework of CP2014 can be used to shed light on the semantics of even.  

Currently, I am investigating the topic of the syntax and semantics of negating quantifiers:

(5)       a.         Not everybody was there.

            b.         Not many people were there.

            Some of the questions of interest that I am investigating now are the following: In (5), does negation form a constituent with the quantificational DP or is the negation in some kind of sentence initial position? Does negation modify the quantificational DP or does it modify the quantifier directly: [not [many people]] vs. [[not many] people]? What principles determine the class of quantificational DPs that can be modified by negation? What are the scope properties of such constructions?

Because of my recent work on anaphora and negation, I am increasingly interested in the tension between syntactic and semantic explanations and the form of the syntax/semantics interface. In each of the two monographs discussed above, we grappled with the issue of whether a certain set of data should be given a syntactic or a semantic explanation. Based on the results of my recent work, I believe that rules of semantic interpretation are transparent, in the sense that they are simple and operate directly on syntactic structures, which can be very abstract and involve many empty elements that give syntactic representation to implicit elements of various sorts.

Other work that I have done in the past that bears on the question of phonologically empty elements (and ultimately on the tension between syntactic and semantic explanations) includes Collins 2005 “A Smuggling Approach the Passive in English”, Collins 2007 “Home Sweet Home” on the syntax of null prepositions, and Collins 2015 “Relative Clause Deletion”.

4.         Syntax of African Languages

My interest in African languages started when I was a Peace Corps Volunteer in Togo (1985-1987). From the experience of trying to learn to speak the Togolese language Ewe, and puzzling over its differences from English and French, I became hooked on African languages and comparative syntax. My thesis (“Topics in Ewe Syntax”, MIT 1993) focused on various issues in the Ewe pronominal system. For example, in Ewe the 3SG subject pronoun changes from é to wò if Spec CP is filled, providing strong support for successive cyclic movement. Mostly recently, I have co-authored a paper (“Negative Polarity Items in Ewe”) with Paul Postal and Elvis Yevudey. This paper argues that Ewe NPIs are unary-NEG NPIs in the sense of CP2014. Furthermore, several claims made in CP2014 are directly supported by the Ewe data (e.g., that NEG can modify an existential quantifier directly).

During graduate school, I met Jeff Gruber, who was an MIT alumnus and visiting scholar at MIT at the time. He is widely known for his work on theta-roles. What is not widely known is that he was also a pioneer of Khoisan linguistics. Through extensive discussions with Jeff, I became interested in what the Khoisan languages could tell us about the human language faculty. It became clear to me that the task of studying these languages was urgent since almost all of them are endangered. My chance encounter with Jeff Gruber led to fieldwork spanning 20 years on various Khoisan languages (yielding two grammatical sketches, a grammar, a small dictionary and numerous academic papers).

Theoretically, my work on Khoisan has focused on serial verbs, pluractionality and the linker. The linker is a grammatical particle that appears between various verb phrase internal complements, as shown in (1):

(6)       ma       ’a                     šú        Jefo     kı̀         setinkane        

1SG     PROG             give     Jeff      LK       hand-harp

“I am giving Jeff the hand-harp.”

In this example, a particle (glossed LK) appears between the indirect object Jefo and the direct object setinkane. My analysis of the linker is that it heads a vP internal functional projection LKP, providing crucial support for the minimalist assumption of vP internal functional projections.

I have written various papers documenting the cross-linguistic variation of linkers, including Collins 2003 (“The Internal Structure of vP in Ju|’hoan and ǂHoan”), Collins 2005 (“The Absence of the Linker in Double Object Constructions in N|uu”), Collins 2016 (“The Linker in the Khoisan Languages”), Collins 2016 (“Click Pronouns in N|uu”) and a paper comparing Khoisan to non-Khoisan languages (Baker and Collins 2006 “Linkers and the Internal Structure of vP”). I have recently submitted a volume of all my linker papers to OUP, where it is under review.

I did my Guggenheim research in Botswana (2015-2016), where I wrote a grammatical sketch of Kua (central Khoisan) and a small dictionary of Sasi (northern Khoisan). The dictionary has 1247 entries, each entry including both English and Setswana glosses. There are 1054 example phrases in the lexical entries. The dictionary is based on 25,346 sound files (usually four men and four women for each word). I am interested in exploring the historical implications of my dictionary. For example, it should help in reconstructing proto-Kx’a (Northern Khoisan), and in establishing historical ties between Sasi (Northern Khoisan) and Tsua/Cua (Central Khoisan).

5.         Formalizing Minimalist Syntax

Over the course of my career, I have engaged in a series of studies having the goal of formalizing minimalist syntax (Collins 1994 “Economy of Derivation and the Generalized Proper Binding Condition”, 1997 “Local Economy”, 2002 “Eliminating Labels”, 2016 “Merge(X,Y) = {X,Y}” and Collins and Stabler 2016 “A Formalization of Minimalist Syntax”). For example, in Collins 2002, I proposed to eliminate labels by defining Merge as in (7):

(7)       Merge(X,Y) = {X,Y}

            This definition for Merge has now become standard in the minimalist syntax literature. I have recently defended the proposal in (7) in Collins 2016 (“Merge(X,Y) ={X,Y}}”), arguing that as far as linearization goes there is no need for labels or labelling algorithms.

Minimalist syntax is simple enough that formalization is possible, and it is abstract enough that formalization is helpful in understanding its basic concepts. The nature of the concepts involved in minimalism, because of their simplicity and generality (e.g., the notion of copy), are just too fundamental and abstract to resolve by talking through them in an informal or semi-formal way. With formalization we can hope to state things in such a way to make clear the conceptual and the empirical properties of the various proposals, and compare and evaluate them.

Collins and Stabler 2016, in particular, should generate much interesting work since it provides definitions of many concepts that have never before been defined (e.g., occurrences, Transfer/Spell-Out and workspaces) and raises many issues which it should now be possible to treat in a precise way. Some relevant questions raised include the following:

(8)       a.         Are chains necessary? If dependencies are not expressed using chains, then how

are they expressed?

b.         How can informal ideas about spelling out occurrences and linear order be incorporated into the formal definition of Transfer/Spell-Out?

c.         How can informal ideas about the copy theory of movement and reconstruction effects be incorporated into the formal definition of Transfer?

d.         Can Agree be reduced to Merge? What would the empirical consequences of such a reduction be?

e.         How are occurrences spelled out in remnant movement?

f.          How does semantic interpretation take place in the minimalist model?

g.         Are labels needed in syntactic representations? How can the labeling algorithms of current research be formalized?

h.         What is the definition of the unit of spell-out (phase)? What are the consequences of different definitions of phases?

i.          How are spelled-out chunks of structure related to one another (the Assembly Problem)?