Long Distance Dependencies

by

Brooke Larson

• LAST REVIEWED: 27 June 2017
• LAST MODIFIED: 27 June 2017
• DOI: 10.1093/obo/9780199772810-0204

Introduction

Words bear relations and interact with other words in every sentence. These interactions can take the form of overt morphological reflexes as well as intuitions of semantic composition and constituency. But relations that hold between linearly adjacent words come, in a sense, pre-equipped with a quasi-mechanical explanation for any interactions that may arise between them: they interact in virtue of their obvious proximity. For example, in the phrase an apple, an takes the particular form that it does (as opposed to the consonant-less a) in virtue of being immediately followed by a word that begins with a vowel. Had a different word intervened between those two, the form of the determiner would hinge upon that new word: (a red apple). Some relations do not require that linear adjacency and can hold from a distance. When one speaks of Long Distance Dependencies, it is not obvious that there is a particular length of dependency beyond which a dependency must be deemed “long.” In this work I will use Long Distance Dependency to mean any dependency that need not hold between strictly linearly adjacent words or morphemes. As such, this will include obvious instances like wh-question (which apple did you buy?) formation as well as more bounded dependencies like those of anaphora (Jane saw herself). This admittedly broad definition serves to rule-in discussion of dependencies for which distance is immediately relevant if however delimited (say clause-bound anaphora). It will also however unfortunately rule in other such non-adjacent dependencies that are not immediately relevant and thus won’t be discussed (for example, thematic role assignment to indirect objects or semantic selection restrictions between verbs and nouns). In any case, long distance dependencies are of interest precisely because they do not come so pre-equipped with a quasi-mechanical explanation. Rather, something additional must be said to explain the a priori unexpected “action at a distance.” How can things apparently far apart come to interact as if they were adjacent? The answer to this question has repercussions for the study of language as a reflection of the human mind in general as well as the study of language for its own sake.

Foundational Works

As noted above, long-distance dependencies are of fundamental interest to the linguist because their explanation is non-trivial. They pose the same sort of problems that gravity did with respect to Newtonian mechanics. The answer to these questions will tell us what the nature of the world is such that things can affect others apparently indirectly. Traditional conceptions of sentence structure included the notion of constituent and could be analyzed using phrase structure grammars. However, in the 1950s, it came to be recognized that the existence of long distance dependencies present some difficulties for phrase structure grammars and that transformations from one set of derived phrase structures to another might be required. Initial concerns are found in Hockett 1952 which noted that certain morphological patterns were difficult to capture in the then-current theory that did not admit transformations. Additionally, Harris 1945 offers an early discussion of “discontinuous morphemes” that interact across a distance and Harris 1957 introduces a grammar that involves “transformations” that have the capacity to link such discontinuous morphemes. Chomsky followed up on the work on Harris with his early transformational grammar, which is explicated in a very detailed fashion in Chomsky 1955 and a much more compact fashion in Chomsky 1957.

• Chomsky, N. 1955. The logical structure of linguistic theory. PhD diss., Univ. of Pennsylvania.

  Chomsky’s PhD dissertation that motivates the use of transformations for, among other things, capturing such long distance dependencies as wh-question formation and passive formation. It provides background historical discussions of the import of long distance dependencies and their relation to the theory of language through the lens of individual psychology.

• Chomsky, N. 1957. Syntactic structures. Berlin: de Gruyter.

  A drastically shortened version, created from lecture notes, of “The Logical Structure of Linguistic Theory.”

• Harris, Z. 1945. Discontinuous morphemes. Language 21:121–127.

  DOI: 10.2307/410503

  An early attempt by Harris, Chomsky’s thesis advisor, to capture non-adjacent relations via “discontinuous morphemes.” The difficulty that such relations pose to traditional grammars is recapitulated in Chomsky’s subsequent work.

• Harris, Z. 1957. Co-occurance and transformation in linguistic structure. Language 33:283–340.

  DOI: 10.2307/411155

  Harris resents a view of transformational grammar that stands as a precursor to the version that Chomsky would create.

• Hockett, C. F. 1952. A formal statement of morphemic analysis. Studies in Linguistics 10:27–39.

  An early discussion of some of the difficulties that contemporaneous transformation-less approaches to morphology faced.