Schedule
25 January: Introduction
Introduction and background on head movement in linguistic theory. Focus on history and origins on the current controversy surrounding the status of head movement as a narrow-syntactic operation.
1 February: Verb Second
Verb-second order, or V2, is one of the central phenomena around which the discussion of head movement has been based. Its reflexes are familiar in the form of English subject–auxiliary inversion, but it exists in many forms throughout Germanic, and several languages display outwardly similar phenomena. Here we look at various arguments about how V2 is derived. Is it always verb movement to C⁰? Or is it sometimes movement just to T⁰? Is there a way of explaining it as phrasal movement?
- Handouts:
- Required reading:
8 February: Verb Initial
Verb-initial orders are plausibly derived in two ways. Either a verb-initial verb phrase is fronted, or a verb is moved to a functional position toward the left edge of a clause. While VOS orders are most easily explained with VP fronting, they can also be derived by verb movement and object scrambling. Similarly, VSO orders can be derived by either moving a verb over a subject, or moving an object out of a verb and the fronting a remnant VP. Here, we look at what distinguishes these approaches, and the arguments for each of them.
- Required Reading
- Other Readings:
Things to bear in mind here are the similiarities to the arguments for V2: Do the same arguments for head movement in verb-second orders hold for verb-initial orders? How do VP movement derivations of verb initial orders compare to Müller's contention that verb-second can be derived with phrasal movement? How far do things move?
15 February: Verb movement and argument positions
Head movement interacts with the placement of arguments in a clause. Scandinavian object shift is perhaps the most well known case of this, but outwardly similar cases are known elsewhere. For instance, in Irish, it appears that a subject must be immediately right-adjacent to the inflected verb, similar to what
den Besten notes for certain pronominal subjects in Germanic languages.
These sorts of facts are often discussed from the point of view of Stowell's (1981) claim that case assignment must happen under adjacency, but it is not immediately clear how to implement this in an Agree-based theory of case assignment.
Vikner (1995) seems to suggest that the subject adjacency condition in Germanic may have to do with the need for C⁰ to be overt; verb movement to C⁰ thus occurs when C⁰ is otherwise phonologically empty. Again, if we assume a modern view with post-syntactic lexical insertion, it is unclear that the need for C⁰ to be overt could drive head movement in the syntax. Looking forward to coming discussions, if head movement is not a narrow syntactic operation, how do we account for these clusters of facts?
22 February: Rich agreement
Behind a lot of the discussion of head movement is the idea that the phenomenon plays a substantial role morphology and in word-building, and this in turn deals directly with the underlying motivation for head movement.
Pollock (1989) proposes that the difference between the position of verbs in French and English is related to the amount of inflectional verbal morphology in the language: If a language has ‘rich agreement’, then a language will have V-to-Infl movement. This is the so called
Rich Agreement Hypothesis.
- Required Reading
- Other Readings:
There have typically been two overlapping debates in this literature. The first question is how much morphology counts as ‘rich’ and how ‘rich’ should be defined. The second is focused on how strong the condition is. While it is fairly well accepted that rich agreement requires verb movement, it remains contentious whether movement can only happen if agreement is rich. We will look at several perspectives on this discussion.
1 March: Strong Agreement
A distinct proposal regarding the link between agreement morphology and verb movement is the
Strong Agreement Hypothesis, which proposes that certain kinds of agreement morphology is pronominal and can, therefore, check the EPP on I⁰. Alexiadou and Anagnostopoulou spell out this proposal. This is a difficult, fascinating paper with many intriguing proposals about the relationship between verb movement and clausal structure. One may wonder, however, whether the proposal is viable in 2017. Is their conception of the EPP as a strong [D] feature still plausable? What is the nature of the parameter that allows verb movement to check the EPP in some languages if all variation is relegated to the lexicon?
8 March: Movement in DP
So far we have looked almost exclusively at cases of head movement in the clausal domain, but presumably it exists in the nominal domain as well. Here we'll look at a case study: Definiteness marking in Scandinavian. In some (but not all) of the North Germanic languages, the definite article alternates with a definite suffix, as shown in the Danish examples here:
- den gamle hest
the old horse
- hesten
horse.the
In the generative tradition, a very influential work has been
Lars-Olof Delsing's (1993) dissertation, which proposes that nouns have the definite suffix, as in (2), when they move to D⁰. movement to D⁰, however, may be blocked by some intervening element like an adjective, as in (1). (In fact,
Santelmann (1993) argues that this is something like the equivalent of V⁰-to-T⁰ movement). However, it is not fully clear how to get this to work: Adjuncts must count as blockers for this to work, but not all adjuncts seem to block movement. Moreover, many of the languages permit both the prenominal article and the suffix to appear at the same time, complicating the hypothesis that movement is actually occurring. Hankamer and Mikkelsen (2005) argue against the verb movement view in Danish (a view I try to extend to Swedish in the paper listed below).
Julien (2005), on the other hand, proposes a hybrid analysis, combining roll-up head movement of material in DP along with movement to SpecDP.
- Handout
- Required reading
- Suggested reading
22 March: Verb-stranding – VPE
Head movement interacts with other syntactic phenomena, including phrasal movement and ellipsis. One phenomena in which head movement plays a roll is verb-stranding verb phrase ellipsis. Beginning with work in the late 80s and early 90s by
Doron (1990),
McCloskey (1991), and Raposo (1986), it started to become clear that languages other than English had a verb phrase ellipsis operation, but that this operation was obscured by the fact that verbs appear in inflectional positions outside VP and by the fact that some of these languages also exhibit object drop. Recent work has attempted to show that verb stranding VPE is a natural consequence of our theories of ellipsis and head movement. Lotus Goldberg’s dissertation remains the touchstone work and is required reading for anybody working on VPE. Gribanova’s NLLT paper is a particularly lucid argument for verb stranding verb phrase ellipsis in Russian.
- Required reading
- Suggested references
Much of this work focuses on how one can tell verb movement out of a deleted VP apart from a full VP with null or topic-dropped objects. A central conclusion is that languages
tend to require that a verb (root) extracted out of an ellipsis site match the verb in the antecedent; no such requirement seems to hold of object/topic-drop (though see Anikó Lipták's
fascinating and worrying handout from the Stanford Head Movement Workshop). This matching requirement, known as the Verbal Identity Requirement, is thought to be a semantic requirement on heads extracted from ellipsis sites. Explaining this semantic requirement has played a role in arguments about the theoeretical status of head movement. Goldberg proposes that moved verbs have to reconstruct into their base positions, but through the footnotes of her dissertation she proposes an alternate idea: If heads never move to begin with (as proposed by Chomsky 2001), then no reconstruction is necessary.
29 March: Verb-stranding – VP movement
Another place where we see head movement and phrasal movement behave differently is in cases of verb phrase fronting. VP fronting in many (but not all) languages with with V⁰-to-T⁰ movement require the verb to be pronounced not only in T⁰ but also in the fronted verb phrase. Phrases moved out of topicalized verb phrases can only be pronounced once. The general view is that this is related, somehow, to the distinction between heads and phrases. Head movement, as we've discussed, seems to have something to do with building morphologically complex words, and so the need to pronounce moved heads in more than one place may be due to some morphological requirement with which phrasal movement does not interact. However, as we'll discuss next week, it may well be the case that this inidcates some deeper difference between the two movement processes (which might explain, in part, why head movement and phrasal movement are subject to different morphological restrictions).
- Required reading
- Suggested reading
5 April: Evidence for head movement as a ‘PF’ phenomenon
We are now at the point where we can begin to ask the question whether head movement should be construed as a narrow-syntactic operation (in the sense of
Chomsky 2001) or some sort of non-movement operation (usually infered to be a PF operation). Today we will look at constructions which people have used to argue that head movement must be post-syntactic. Two will be familiar to us: Schoorlemmer and Temmerman discuss verb-stranding verb phrase ellipsis, while I discuss verb-stranding verb phrase topicalization in LaCara 2016. A third phenomenon, which we have not yet seen, is pseudo-gapping, a separate VPE-based phenomenon discussed by Boeckx and Stjepanović (though what they discuss may simply be verb-stranding VPE).
- Required reading
- Suggested reading
We should go through these arguments carefully, since in many cases they are parsimony arguments. Two of the papers (Schoorlemmer and Temmerman, and Boeckx and Stjepanović) come just short of implying orderings of post-syntactic operations. These two papers also look at phenomena related to VPE; are these arguments telling us something about head movement in general, or are we just running across something specific to ellipsis? Part of the original intent of LaCara 2016 (cut from this version of the paper for space reasons) is to address this by looking at a non-elliptical phenomenon. But here the narrow focus is on whether a system like the one in Nunes 2004 can handle the facts. Could another, better linearization system handle the data better?
12 April: Evidence for head movement as a syntactic phenomenon
There have been several arguments that head movement must be a narrow syntactic movement operation. Much of this argumentation rests on the idea that head movement displays interpretive or LF effects. Lechner argues that, given certain syntactic assumptions, head movement can be seen to have LF scope effects. Hartman argues that the traces of head movement are included in calculating parallelism domains for ellipsis identity. From a slightly different direction, Matushansky (98–105) provides several arguments that head movement as a PF phenomenon is incompatible with several Minimalist theoretical assumptions.
- Required reading
- Suggested reading
The questions to ask here are a bit different from the ones above. Are Lechner's assumptions about reconstruction well founded (see McCloskey's criticisms here)? Given that a number of arguments about head movement are made on the basis of verb-stranding verb phrase ellipsis, are these arguments compatible with Hartman's arguments from ellipsis? Should Matushansky's arguments be seen as theory-internal quibbles, or do they represent true challenges to removing head movement from the narrow
syntax?
19 April: Non-movement implementations of Head Movement
If head movement is not movement, then what is it? Head movement has a number of properties that make it look like syntactic displacement. For instance, it affects X⁰ constituents, which are syntactic elements, and the Head Movement Constraint looks outwardly as though it is constrained by the syntax. As Matushanksy (2006) seems to point out, if head movement is a PF operation, PF will need to have some amount of access to the syntactic structure in order for these sorts of constraints on head movement to hold.
With the recent exception of work by
Gribanova and Harizanov, most non-movement implementations are actually properly syntactic. Harley's
Conflation approach, as we've seen, inolves passing phonological features up the tree to morphologically deficient heads. Platzack presents a technically different view: A series of heads relate their features via Agree, and an EPP feature forces pronunciation of those features on only one of those heads, (allegedly) explaining why verb movement shows no intepretive effects. Finally, Brody's
Mirror Theory builds apparent head movement into the syntax; however, the trade-off is an unfamiliar approach to Syntax.
- Required reading
- Suggested reading
The questions we should ask here are whether these implementations avoid the theoretical challenges head movement presents to syntactic theory, and whether they are able to account for the range of empirical data we've seen. Can we explain (apparent) phenomena like object shift or rich agreement under these approaches? By introducing new theoretical machinery to deal with head movement, are we introducing new complications to our theory that would be rendered unnecessary by modifying Merge/Move or Agree?
26 April: Novel syntactic implementations
If head movement is movement, then how do we make it play nice with Minimalist assumptions about how movement works and how structure is built? Matushansky proposes that head movement is actually the result of two separate operations. A head first undergoes A′-movement to a specifier position, after which it undergoes a separate morphological operation that combines it with an adjacent head. This sidesteps issues with the extension condition and allows a head to c-command its trace (at least, before m-merger applies).
