A brief history of syntactic theory: Early Chomsky

This is part one of a multi-part series. Part II is: A brief history of syntactic theory: Parallel-contraint based syntax.

τυφωθείς εἰς κρίμα ἐμπέσῃ τοῦ διαβόλου

What is the structure of a sentence like this? There’s a lot going on here. This clause has a pre-verbal participle, followed by a prepositional phrase that is split in half by the main verb. When we considering existing linguistic analyses, a couple options come forward. On the one hand, there are the discourse features annotated by Stephen Levinsohn and Steve Runge. On the other hand, there are the various treebanks for the New Testament that have been created over the past decade. Both types are modeling particular things about the grammar of the language

What then is the configuring
mechanism for syntactic structure?

This is sort of an ongoing challenge for people working on modeling syntax. It actually goes back several decades now. Before we can talk about possible solutions, let us examine some of the history behind syntactic representations.

Early syntactic theory

A good place to start with syntax trees is probably with Chomsky. There’s no need to get into the more complicated details of movement, transformations, and the like. For our immediate purposes, we just need the surface structure and their trees. The trees of this era in the 1960s were as much about defining grammatical relations (subjects, objects, etc.) as they were about actual phrase structure.

NP = (Det) N
VP = V (NP)

Configurational syntax

Rules like the three above were envisioned not as only explaining the basic hierarchy of English syntax, but also providing explanatory power to subjects and objects. Rather than a subject being a syntactic privative, the syntactic rules that create the tree provided derivational basis for the relations themselves. Grammatical relations, then, for Chomskyan theory only exist as locations in a tree and nothing more. It has an elegance to it that even today is quite beautiful. And you can see why movement rules become necessary here. Once you have the structure, the sense that something has moved when you encounter structures like passives or relative clauses suddenly feels like quite a natural explanation.

Basic phrase structure trees give an explanation for something in English that previously did not have one: where do grammatical relations come from?

And that is their difficulty, too. Not all languages are like English. Not all languages define grammatical relations on the basis of locations in a tree. That’s what we see with Greek. Grammatical relations can not be reduced to merely derived features of a syntactic tree. Moreover this is a problem that complicate more than just our definitions of what subjects and objects are. For the early Chomskyan work in syntax, if the tree does not predict grammatical relations in a language by position, then by implication the tree does not predict the order of the constituents at all. Constituents are no longer configured by grammatical relation. What then is the configuring mechanism for syntactic structure?

The consistent answer for the past few decades or so has been information flow.

Now, information flow is a large and intricate subject to cover, but a short summary is possible. Information flow involves the order in which a speaker/author chooses to convey the information they want to communicate. This usually defaults to the nature order from old or known information to new or unknown information. Communication between two parties relies on first establishing (or often simply assuming) an existing shared pool of information that the speaker/author can draw upon in order to most efficiently and effectively communicate their ideas.

A brief account of information flow concepts…

Common labels for helping organize information flow descriptively include words like topic and focus. The former often gets discussed in terms of aboutness or old/given information both of these convey fact that in some way there is an element in a sentence that the speaker can rely on their audience already recognizing: ‘this is what this sentence is about,’ ‘you already know this part and it is important for the piece of information I want to share with you.’ The latter label, focus likely evokes in the mind things like prominence or emphasis and there is a sense in which that is not far off, but its more important entailments in the study of information flow pertain to newness and assertion. These two words should call to mind the other side of the coin in communication: making assertions (new information) about things the audience already knows (old information).

So in the study of information flow, the bare minimum of categories necessary are:

  • Topic/topical information old, assumed, given
  • Focus/focal information new, asserted about a topic

And back to syntax…

Since the splitting of various linguists away from Chomsky over the years, accounting for information flow has varied substantially from theory to theory.

Mainstream generative theory is a carpenter with a hammer and all of language is treated like a nail.

The mainstream generative linguists have worked to squeeze information flow into their existing models by adding more layers to the trees. Now there’s a Topic Phrase, a Focus Phrase, and any number of other elements all sort of jury-rigged into the tree.

Information Structure Projections

Looking at a tree such as this one, there is likely to be a sense that some of the original elegance of the framework has been lost in the intervening years. This is not a particularly attractive solution to the challenge problem of languages that differ from English. Now, the intention of this approach is certainly clear. It operates analogically to Chomsky’s original proposal for grammatical relations. If subjects and objects are configured into the hierarchy of the syntax tree, perhaps information flow concepts are also configured in syntax similarly. Yet there’s a sense in which mainstream generative theory is a carpenter with a hammer and all of syntax has become a nail.

But this is not the only approach to account for information flow’s relationship to syntax.

Click below to follow us via e-mail or RSS or even Facebook (ugh, Facebook).

15 thoughts on “A brief history of syntactic theory: Early Chomsky

Add yours

  1. So basically the meaning lies beyond the sentence level and transformational grammar tries to connect syntax and pragmatics, right? Functional grammar approaches it from the different angle. Do you think it is possible to use both generative and functional sides for K-Greek? Thank you! 🙂

    1. Transformational Grammar doesn’t deal with semantics at all. For early generative theory, neither semantics or pragmatics are viewed as being within the domain of grammar at all. The goal for their era of linguistic theory was to explain as much as possible only using syntax. For them, meaning is a derived property of syntactic structure that only comes with the interpretation of a sentence. And the idea of talking about anything beyond the sentence just wasn’t on their horizon in the 1960’s.

      But none of this is really relevant to the discussion above. What’s important here is that early generative linguistics is the back drop against which all other approaches to syntax compare themselves for the next several decades, even to this day. We cannot understand why the current state of linguistics looks the way it does without also understanding its history.

      1. Thank you for the kind explanation. I know it is a bit off the topic but wondering if the functional view of info. structure (pragmatic, e.g., Dik and Lambrecht) and that of generative view (transformation between the surface and the deep structure) can be combined since the functional philosophy on the relationship between “form and function” is fundamentally different from that of Chomskians. (cf. forgive my a bit less precise terms, I am not aware of details of historical differences on Chomskian linguists.)

        1. Generative linguistics views the study of pragmatics as “uninteresting.” So, the short answer is no, they’re not really compatible.

          With that said, there are other non-Chomskyan generative models that are more easily related to functionalism. I’ll be writing about them next (today and hopefully next week). After that, we’ll be discussing how full-blown functional theories do syntax and finally discuss ways that it might be possible to deal with both information flow and syntactic structure at the same time.

        2. So, do you think the theories of information structure and discourse has nothing to do with the formal theory? If so, where and how do you define and categorize these, Kamp and Reyle(1993), Vallduví(1992), and É. Kiss(1987, 1994)? Furthermore, what about, concerning BH narrative, Sebastiaan J. Floor saying “Generative linguists limit the study of information structure to the clause-level”? It is true actually that only functionalists talk about info. structure and discourse until 1980s, but since then it has changed as far as I know.

          Sorry but, I don’t understand your description of Generative linguistics clearly, too. By G-linguistics you mean mainly CFG theories? I meant by G-linguistics more likely relevant to UG and TG (as a remedy of CFG). I further don’t understand why you think TG doesn’t deal with semantics at all. In 1960s, TG concerns the relationships between syntax and semantics and further implications of transformational analyses, though early 1970s Chomsky focused on making the theory so restrictive to explain the language learnability. Could u clarify it?

          Now, the fundamental characteristics of G-linguistics different from functionalists (including both conservative[non-theoretical] and moderate [FG, RRG]) I believe, is the criteria of sentence-internal and theory internal (conditions of the economy, motivation, predictiveness) as Robert D. Van Valin argues. That is, the method and range to explain certain syntactic phenomena.

          If they, functionalism (not conservative i.e.,) and formalism, are not compatible, what do you think of the works and view of Frederick Newmeyer, if you are familiar with?

        3. The point of this post and the ones that follow are designed for people who have no background in linguistics at all. You’re asking questions that I’m not trying to answer with this post. So really: Yes, to everything you’ve said here. You’ve got it.

          The primary focus of this post is early Chomsky–Extended Standard Theory, to be precise…before θ-roles, before, discourse configurationality, before lot’s of stuff. I intentionally gloss over the details of the because the important point that I wanted to explain here is the mechanism: UG’s eventual implementation of information structural cateogries into syntactic structure is predicated on the same basic mechanism that motivated grammatical relations at the very beginning: just as tree positions defined GR’s, tree positions definte information structural categories–as you stated Kiss, Vallduví, Kamp, and Reyle are all significant here. I’ll be writing about them in another post that’s forthcoming.

          But the point of this post is solely explaining configurationality within the mainstream generative paradigm, particular with Extended Standard Theory.

          The rest, TG as a corrective to CFG, the status of semantics (which isn’t important because it’s interpretive in early Chomsky; in the 1960-70s it’s a product of syntax+lexicon and thus has no status on its own…hence the Linguistic Wars).

          But you know all of this, you’re not the intended audience for this post.

          Newmeyer makes interesting proposals, but he also tends to overstate his case. The language of “Formal” and “Functional” is unhelpful here, since “Formal” can mean both “has a formal representation (i.e. is mathematically explicit)” and can also mean “Formal (syntactic) structures are genetically innate within the human brain”. Any functional framework can have an explicit formal represntation without accepting the innateness claim…a claim that then brings us back to interpretive semantics and syntactic autonomy. HPSG is formal in the mathematical sense, but it is agnostic on UG and autonomy of syntax (cf. Sag, Wasow, & Bender 2003).

          You’ve got a solid background, it sounds like. Where did you (or where are you currently) study?

        4. Thank you for your clarification. I really appreciate it. I just dabbled in a bit of linguistics in my college and mission field, now I am studying at DTS. I am working on PIA/TIA (Paronomastic or Tautological infinitive absolute) in Hebrew in light of cognitive-functional linguistics. I just needed clarification of my thoughts in connection with your posts.

  2. Maybe I’m being a bit optimistic but I see the cartographic approach with special positions in the tree for topic and focus to be a convergence of sorts to the functional view of placing items based on information structure.

      1. Granted that the tree can be made second- or third-dimensionally, but what if the generative view refuses the decisiveness of the functional principles on determining topic/focus, i.e. beyond the sentence level, context-dependant pragmatics? I mean the pre-verbal markedness does not guarantee ‘focus’ in the functional view but the broad context does.

  3. I should come here often. I have been greatly enlightened by your post especially your answers to the questions concerning the post. I’m new to the field of syntax. I think, I am your right audience.

Leave a Reply to Mike Aubrey Cancel reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Powered by WordPress.com.

Up ↑

%d bloggers like this: