Minimalist Investigations in Linguistic Theory

Howard Lasnik

• Publisher: Routledge
• Year: 2005
• eBook ISBN: 9781134675326
• Topic: Sprachen & Linguistik
• Subtopic: Sprachwissenschaft

Professor Howard Lasnik is one of the world's leading theoretical linguists. He has produced influential and important work in areas such as syntactic theory, logical form, and learnability. This collection of essays draws together some of his best work from his substantial contribution to linguistic theory.

1
INTRODUCTION

The articles collected here, like the large majority of my work, are syntactic investigations in the Chomskian framework, especially Minimalism, its most recent development. I was first attracted to Chomskian generative grammar by its quest for “hidden causes.” A hallmark of Chomsky’s work (and that of many of his students, associates, and followers) is the attempt to explain the apparent chaos of language by appeal to the interaction of simple underlying principles, structures, and operations. A related impulse has been to show that languages that are superficially very different are actually far more similar just a bit below the surface.

Hidden causes at the morpho-syntactic level have been especially interesting to me. One classic analysis in such terms is presented in Chomsky (1955), and given implicit justification there, and explicit justification in Chomsky (1957). The analysis centers around Chomsky’s phenomenally successful attempt to make sense of the apparent chaos of English verbal morphology, with the very simplest kinds of basic sentences seeming to exhibit the most unpredictably idiosyncratic alternations under negation and interrogation:

(1) John left    John didn’t leave
 John should leave  John shouldn’t leave
 John has left    John hasn’t left
 John is leaving    John isn’t leaving

(2) John left    Did John leave
 John should leave  Should John leave
 John has left    Has John left
 John is leaving    Is John leaving

Chomsky’s breakthrough was the insight that the tense/agreement morpheme in English is syntactically an autonomous entity even though it is invariably realized as a bound morpheme. It is available to transformational manipulation just as much as, say, a modal auxiliary is.

The beauty of this analysis had a profound effect on me, though, interestingly, not the very first time I saw it. In early 1968, I read through Syntactic Structures and did not really “get it.” Having a degree in mathematics, I had no technical difficulty with the automata discussion or the phrase structure and transformational formalism. Rather, I failed to grasp what it was all in aid of. That was rectified in the fall of that year when I took “baby syntax,” the introductory graduate syntax course at MIT. This was the first linguistics course I ever took, and one of the very best. It was team taught by two inspirational teachers, Morris Halle and John Robert (“Haj”) Ross, who easily held the rapt attention of the class for the three two-hour meetings each week. Halle taught the first half of the course, emphasizing general issues of linguistics, especially what kind of a thing a human language is, what its special properties are (structure, infinity,…), and what a theory of language should be: an account of what speakers know and how that knowledge arises. When Halle then began to present a version of the Syntactic Structures account of the English verbal system (followed up later by Ross’s detailed presentation of virtually the entire syntax of English, as then understood), I finally “got it,” and was overwhelmed by its beauty. My reaction is still the same. In the Introduction to Lasnik (2000a), I call Chomsky’s early treatment of verbal morphology “the best set of analyses in the history of our field.” I continue to use it as a model for my own theorizing, sometimes explicitly, as in my work on verbal morphology (Lasnik 1981, 1995e) and Chapter 2 of this volume), but always at least implicitly. What particularly struck me was how seemingly different sentence types, which obviously had to be brought together somehow, were, indeed, brought together, and by a system that was simpler than any description that kept them apart.

Chomsky’s analysis crucially embodies a hidden cause, an element that has no overt realization of its own, but has striking indirect effects. Chomsky posits an S morpheme for the present tense of verbs with (third person) singular subjects, and a zero morpheme Ø for other present tense forms. Given that the latter forms are virtually always phonologically indistinguishable from the bare citation form, Chomsky observes that “An alternative we did not consider was to eliminate the zero morpheme and to state simply that no affix occurs if the subject is not third person singular.” Chomsky (1957: 64). The reason for rejecting that alternative out of hand was that it would have substantially complicated the system with no concomitant benefit. In Lasnik (1981), I present a completely parallel argument for the existence of a zero imperative morpheme in English, and in Chapter 2, a similar one for a zero habitual morpheme in African-American English.

A closely related Chomskian theme that has greatly influenced my work is that “if some phenomenon is observed overtly in certain languages, then it probably applies covertly (i.e. without overt expression at PF) in all languages in some manner; that is, the overt expression is probably a consequence of requirements of UG, which must be satisfied at S-structure and LF, even if not overtly observed at the PF level” Chomsky (1987: 68–9). Chomsky specifically mentions Case, which is richly expressed in such languages as Latin, Finnish, and Sanskrit, hence “probably … required in all languages, even in languages such as English in which there is only a very marginal residue at PF….” Case theory, as developed especially by Chomsky out of an original proposal by Jean-Roger Vergnaud in a personal letter to Chomsky and me (commenting on a draft of Chomsky and Lasnik (1977)) led the way to insightful analyses of many phenomena in numerous languages. In much of my work from the early 1980s to the present, I have relied on Vergnaud’s and Chomsky’s basic insight about the distribution of nominal expressions, and I have explored various ways of developing and extending Case theory. Examples are Lasnik (1992a, 1993, 1995a,b, 2001), Lasnik and Freidin (1981), Lasnik and Saito (1991), and Chapters 3 and 10. One early question I still worry about is exactly what class of expressions is subject to the Case Filter. Early on, Bob Freidin and I began worrying about Chomsky’s original formulation in which phonological realization was key. In Lasnik and Freidin (1981), we showed that the trace of wh-movement must have Case, even when the operator that has moved is phonetically null. This demonstration was, in part, responsible for the move to the “visibility” approach to Case suggested by Chomsky (1981), under which all and only arguments (whether phonologically realized or not) obey the Case Filter. Over two decades later, the ramifications of that move are still being worked out. This, along with a number of questions about the role of Case, is at the core of much minimalist theorizing, including my own.

Chomsky also mentions wh-movement in the context of deep similarities between languages hidden by superficial differences, suggesting that “If languages such as English exhibit overt wh-movement in interrogatives, then probably languages such as Chinese and Japanese, which do not, nevertheless have LF-movement of wh-phrases….” In the very early 1980s, when Huang (1981/82) was first laying the empirical and theoretical groundwork for a theory of covert wh-movement, Mamoru Saito, then a graduate student at MIT, and I simultaneously became intrigued by some of the technical questions that arose in Chomsky’s initial presentation of a theory based on Huang’s. Our collaboration on this topic resulted in Lasnik and Saito (1984) and, several years later, Lasnik and Saito (1992). A major concern of both works was the specification of the Empty Category Principle (ECP), a locality constraint on (the traces of ) overt and covert wh-movement. To this day, I regard Huang’s initial argument that wh in situ obeys the ECP as still the single strongest argument for covert movement. The general question of how overtness of movement is forced is the theme of Chapter 6.

At the more foundational level, Chomsky (1965, chapter 1), was a major influence on my thinking about language. Parallel to my experience with Chomsky (1957), I initially had difficulty understanding exactly what Chomsky was getting at. But when I came to understand Chomsky’s notion of “explanatory adequacy,” I made that a goal of virtually all of my work. It was the move towards increased emphasis on explanatory adequacy that made the advent of the “extended standard theory” so appealing to me, and the “principles and parameters” theory even more so. Lasnik (1990) collected nine of my articles, in the principles and parameters framework, on the explanatory themes of restrictiveness and learn-ability. I see Minimalism as a natural development of these themes.

One major Minimalist line of research with evident roots in the earliest work in transformational grammar (e.g. Chomsky (1955)) concerns levels of representation, particularly the question of exactly what levels there are. (For a historical survey see Lasnik (In press)). In the late 1960s and early 1970s, questions were raised about the existence of deep structure as a level, most directly by Postal (1972). Chomsky in his minimalist writings raises this same question (even if from a somewhat different perspective), and extends the question to surface structure. Anaphora has provided a major testing ground for the hypothesis that surface structure does not exist, since so many anaphoric connections had been thought to crucially rely on surface structure configuration. In Chapters 4 and 8–10, I discuss the issue, showing how apparent surface structure effects can be reanalyzed as dependent on LF instead (by positing more overt movement than is usually assumed), but I also continue to explore the question raised in Lasnik (1993) of why assumed covert movement does not generally create new configurations of anaphora.

Thinking about my research and my teaching, I discern a rather robust strain of what can be characterized as conservatism. I often find myself trying to resurrect old analyses or maintain current analyses that are being supplanted, feeling that the rejected accounts were rejected prematurely, or for debatable reasons. I talk about this trend specifically in my teaching in the Introduction to Lasnik (2000a):

At the beginning, I want to say a few words about why I am going to base a large part of the discussion on something that seems so old and outmoded by the standards of current syntactic theorizing. I have three reasons. First, many of the terms, concepts, and analyses in recent work are much easier to understand against a backdrop of their ancestors of a few decades ago. Second, our field is a relatively young one without a very large number of good arguments and analyses to use as models. We can’t yet afford to ignore some of the good arguments and analyses, even if we conclude that they are ultimately incorrect. Finally, and probably most importantly, I believe that many of the analyses of the 1950’s are actually correct, fundamentally and (sometimes) even in detail.

In addition to my investigations of verbal morphology, some of my other recent “conservative,” or even “reactionary” work explores Exceptional Case Marking (ECM), ellipsis, island violation “repair,” and clause-mate conditions on syntactic relations. Starting with Lasnik and Saito (1991), I have been arguing that the accusative subject of (certain) infinitive constructions raises into the higher clause, an approach to the construction that was standard in the late 1960s and early 1970s (as in Rosenbaum (1967) and Postal (1974)) but which was largely abandoned following the arguments of Chomsky (1973). Some of my most recent discussions of (an updated “minimalist” version) of this old approach appear in Chapters 3 and 10 and in Lasnik (2001). Classically, the issue of “raising to object” has been intimately connected with the formulation of the locality condition on anaphoric relations, among others. The traditional view, elaborated in great detail in Postal (1974), was that the relevant locality obtains when the two items to be related are in the same clause – clause-mates. Work since the early 1970s has taken alternative approaches, for example, in terms of the Tensed Sentence Condition of Chomsky (1973) or the Governing Category of Chomsky (1981). My work on ECM constructions implicitly argues for the classic view. I make that argument explicit in Lasnik (2002). It is an interesting question whether the classic approach is more or less in accord than its successors with minimalist ideals. My gut instinct is that, if anything, clause-mate is actually a more minimalist concept, since the notion “clause” is quite plausibly a primitive of syntactic description.

Some of my current work investigates ellipsis, one of the most persistent topics in generative grammar. In this realm, too, I have assumed, and sometimes argued for, a traditional analysis, one based on deletion, rather LF copying, the latter being the far more standard view in recent syntactic theories. Chapters 5–7 analyze some properties of ellipsis in terms of deletion. Chapter 6 argues that a certain complementary between normally obligatory movement and ellipsis receives a natural account in these terms. With respect to ellipsis, again, one might ask whether Minimalism suggests one approach over another. The answer is far from clear, but a strict interpretation of the Inclusiveness condition of Chomsky (1995a) might suggest a deletion approach, since on that approach, it is clear that no new elements are added in the course of a derivation.

As noted above, I see Minimalism, with its extreme simplification of the model, as a natural development of the drive towards explanatory adequacy. I think we still have quite a way to go to achieve explanatory adequacy, but is not unreasonable to begin to raise the possibility suggested by Chomsky (2001) that we can eventually move beyond it.

2
PATTERNS OF VERB RAISING WITH AUXILIARY “BE”

---

In this chapter, I reprise the “hybrid” analysis of verbal morphology that I presented in Lasnik (1995e), where Infl can be either a bundle of syntactic features, as in Chomsky (1993) (in which case it attracts a verb), or an affix, as in Chomsky (1955, 1957) (in which case a low-level process of Affix Hopping associates it with a verb). In terms of this theory, I suggest an account for the patterning of habitual be in African-American English (AAE). Based on the work of Green (1993), I propose an AAE habitual morpheme “Hab” and suggest that its morpho-syntactic behavior is parallel to that of the imperative morpheme (a zero affixal morpheme). In the course of the presentation, I offer a speculation about the nature of do support, suggesting that it is merely the phonetic manifestation of a finite Infl (or Imp or Hab) that has not been able to undergo Affix Hopping. I also bring new data to bear on the old puzzle that adverbs do not block Affix Hopping, a process normally requiring adjacency. The new data suggest that, contrary to initial appearance, adjacency might actually obtain in these cases.

---

One of the major breakthroughs in the history of generative transformational grammar was the discovery by Chomsky (1955, 1957) of the regularities underlying English verbal morphology. Much of the apparent chaos of this central portion of Standard American English (SAE) morpho-syntax was rendered systematic by the fundamental insight that the tense agreement inflectional morpheme is syntactically independent, even though always a bound morpheme superficially. The analysis was brilliantly successful, and paved the way for numerous refinements and extensions over the past forty years, the large majority of them sharing the same fundamental insight. Labov in a number of publications (1969, 1972) demonstrated that African-American English (AAE) falls under many of the same fundamental generalizations, the apparent striking differences often being the result of rather low-level operations. More recently, Green (1993) has explored especially the aspectual system of AAE, showing (unsurprisingly to linguists) that the same general syntactic principles are operative, though there are significant parametric differences. In this chapter, I will consider some of the revisions of Chomsky’s original proposal, ultimately presenting an analysis (Lasnik (1995e)) that constitutes something of a return to that original proposal. Finally, I will show how a modified version of Green’s analysis of the AAE habitual construction fits naturally into the approach to verbal morphology I advocate.

2.1. Earlier accounts

Many of the successive refinements of the Chomsky (1957) theory can be viewed as attempts to maintain the leading ideas of the analysis but to reconcile them with the growing concern for explanatory adequacy. For example, Lasnik (1981) was particularly concerned with the stipulated rule ordering and the arbitrary marking of particular transformations as obligatory or optional in Chomsky’s early system, and proposed that these problematic language particular formal mechanisms can be eliminated in favor of the general filter in (1):

(1) The “stranded affix” filter: A morphologically realized affix must be a syntactic dependent of a morphologically realized category, at surface structure (Lasnik (1981)).

Notice that this filter crucially assumes, along with Chomsky (1957) and many succeeding analyses, that the inflectional material on a verb is a morphological affix, even though it begins its syntactic existence as an autonomous entity. Given this assumption, and given (1) and the restrictive theory of transformations it presupposes, a typical analysis of the English verb system of the early 1980s looks something like (2):

(2)(a) S is the maximal projection of the inflectional morpheme Infl (=C of Chomsky 1957).
(b) In...

Table of contents

• Cover Page
• Title Page
• Copyright Page
• Figures
• Preface
• Acknowledgments
• 1: Introduction
• 2: Patterns of Verb Raising with Auxiliary “Be”
• 3: Last Resort and Attract F
• 4: Levels of Representation and the Elements of Anaphora
• 5: Pseudogapping Puzzles
• 6: On Feature Strength
• 7: A Gap in an Ellipsis Paradigm
• 8: On a Scope Reconstruction Paradox
• 9: Some Reconstruction Riddles
• 10: Chains of Arguments
• Notes
• References