A Construction Grammar Syllabus for Humanists

Over the past few years, a number of colleagues have said to me something like this:


“Yeah, I know the folk theories we pick up in humanities curricula aren’t plausible accounts of everyday language use and especially language creativity. But linguistics is a big field, and I don’t know what to read or where to start.”


This post – somewhere between a syllabus and an annotated bibliography – is here to help. I want to caution, emphatically, that the readings it lists do not provide a full or neutral overview of the whole field of grammatical theory at present, much less a history of the field over the last half-century and more since the Saussure or Jakobson that literary scholars are likely to have encountered in an Intro to Theory Course. It does not include work by Noam Chomsky or those working in successive Chomskyan paradigms (transformational grammar, government and binding, principles and parameters, X-bar theory, minimalism). I’ve invested quite a lot of time and effort in reading and grappling with this work, often with great pleasure, and I regard Syntactic Structures and Aspects of the Theory of Syntax as intellectual monuments of the highest order. But for various reasons (many of them controversial) I hesitate to recommend these works to literary critics, literary or intellectual historians, or cultural theorists who aren’t already independently invested in the questions they address. As I’ve suggested in earlier posts, one reason why most humanists, since about 1980, have worked with either inadequate structuralist accounts of language or even less adequate folk models (often picked up while studying a second language) is that the methods, concepts, claims, and aims of the Chomskyan paradigm are either largely irrelevant or directly inimical to the goals of humanistic inquiry.


Luckily, that paradigm is not – is no longer – the only game in town. (There are lots of games now.)  The following bibliography of work in Cognitive and Construction Grammar aims to offer humanists a point of entry to an approach or school of linguistics that is, I think, especially well suited to the kinds of work that scholars do in literary, historical, and interpretive disciplines. Taken together what this approach offers to humanists is a more rigorous, adequate, explicit, and empirically grounded account of how we routinely produce and understand utterances that we’ve never encountered before, how signs are structured, how the meaning of complex utterances are composed of the meaning of their signifying parts, how those elements come to be part of our linguistic knowledge, and how languages vary over time.


Because the constructionist approach to grammar is “sign-based” – treating our linguistic competence as a repertoire of variously abstract and complex signs – it is remarkably well suited to the intellectual traditions of structuralism and post-structuralism that are foundational to much literary and cultural theory. (It also forces us to rethink some of the central dicta of those traditions, such as the arbitrariness of the sign, but more on that anon.) Unlike the “autonomous grammar” of the Chomskyan paradigm, it offers a holistic account of grammar that is linked at every point with meaning and discursive function, even while adopting the central, paradigmatic problem of creativity posed especially well by Chomsky: how is it that we make (potentially) infinite use of finite means. Because Construction Grammar is not primarily oriented towards establishing the features of a universal grammar by isolating invariants across the grammars of natural languages, it does not need to cordon off an essential, biolinguistic grammatical “core” from the epiphenomenal cultural “periphery.” Instead it adopts, as Kay and Fillmore put it, “a commitment in principle to account for the entirety of each language,” including aspects usually deemed “peripheral.” a commitment which orients it towards the task of taking full account of cultural, linguistic, and historical specificity.


Here’s the Syllabus. I’m posting it now, but it’s far from finished, and I plan to add to and enrich it in stages. Though it bears some resemblance to an annotated bibliography, it isn’t organized alphabetically or chronologically. Its order is instead pedagogical: it aims to lead the inquisitive humanist through a fairly comprehensive theory of language.

1. The best place to start – a worthwhile read even if you read nothing else – is a short, brilliant, informal lecture by the Berkeley linguist Charles Fillmore on idioms and idiomaticity. Idioms (as the term itself suggests) have usually been treated as idiosyncratic, inessential, peripheral, or marginal epiphenomena of language. But Fillmore argues that “idioms are the irreducible units of description for the way a language works,” such that “almost every grammatical model is a theory of idioms, since the job is to isolate those principles which are in themselves irreducible, that is, which are not explained by other principles in the same language.” This is an argument of the “the stone that the builders rejected has become the cornerstone” sort, in which what have usually been regarded as marginal or inessential (idioms) are shown instead to be the “irreducible unit of description” (constructions of various complexity and abstraction).

2. Fillmore, Kay and O’Connell’s conception of constructions builds on important work on idioms happening at UC Berkeley in the 1980s. Here is the key paper by Nunberg, Sag and Wasow, which shows that idioms aren’t readily assimilable to either a rule based conception of grammar or to a list-based lexicon. Get ready for a good bit of string pulling and bucket kicking!

Even literary scholars who are uninterested in grammar can, I think, profit from some time spent thinking about idioms, which have in the past 40 years probably been the most important driver of changes in the way that linguists and philosophers of language think about the structure of the sign. We are best able to see and study the things that we are able to name.   I think it’s plausible that generic forms are essentially abstract, partially demotivated idioms, with abstract and minimally constraining slots generalized over characters and events rather than lexical items.

3. A more developed version the argument that idioms or idiom-like units are the basis of our grammatical knowledge is Fillmore, Kay, and O’Connell, “Regularity and Idiomaticity in Grammatical Constructions: The Case of Let Alone,”  which is one of the founding documents of Construction Grammar. It uses the analysis of a single construction (X let alone Y) to propose a total rethinking of the edifice of grammar and ends by suggesting that the abstract and semantically autonomous principles characteristic of Chomskyan grammar are “degenerate instance[s]” of phrasal constructions like X let alone Y, that are both combinatorial and paired with a conventional meaning.

4. It’s worth reading one further account of an unusual construction – What’s X doing Y? – by Kay and Fillmore. A payoff of this article is a new understanding of the threadbare joke:

Diner: Waiter, what’s this fly doing in my soup?
Waiter: Madam, I believe that’s the backstroke.

But the larger stakes involve learning A) how to identify individual constructions, instantiations of which have conventional meanings that are “neither given by ordinary compositional processes nor derived from a literal meaning by processes of conversation,”  B) how to determine the various but constrained ways that lexically unfilled variables in a construction like What’s X doing Y? can be filled, C) how to position a construction in a larger “inheritance hierarchy” of constructions that constitutes the grammar and lexicon of a language.

5. With a few key arguments in under your belt, it’s worth stepping back to take a more synoptic view of construction at grammar. Those who would like it in a single book should get their hands on Constructions: A Construction Grammar Approach to Argument Structure (1995) by Adele E. Goldberg. Goldberg is the current standard bearer for and an impressive champion of constructional approaches, and her first book covers issues of form, meaning, compositionality, information structure, synonymy, and more. Where Fillmore, with whom Goldberg studied for her doctorate, worked mostly on purportedly “peripheral,” idiomatic, or culturally specific constructions, Goldberg tends to focus on constructions that it would be implausible to exclude from our core linguistic capacities, like Subj V Obj1 Obj2, as in Pat threw Sam the ball, or Subj-Aux inversions, as in Am I on the hook for this? Her second book Constructions at Work: The Nature of Generalization in Language (2006) provides a similarly synoptic account of Construction Grammar while focusing especially on how children learn constructions through experience. Since the second book, her research has involved more controlled experiments and statistical evidence for mechanisms of language learning. I am eagerly awaiting her third book, Explain Me This: Creativity, Competition and the Partial Productivity of Constructions.

Goldberg’s publications are listed here, and many of them can be downloaded directly. I have yet to read one that does not repay close attention.

6. Those who would like a chapter-length rather than a book length synopsis should look to Chapter 9 of William Croft and Alan Cruise, Cognitive Linguistics (Cambridge 2004), which offers a clear and succinct introduction to the commitments of constructional approaches and how they differ from those of the Chomskyan paradigm.


So far this syllabus has included work on the nature of constructions and of their systematization in a repertoire (often called a “constructicon”) that stretches from individual morphemes and words to the most general phrase structures of a language. But while construction linguistics began in grammatical theory, it has had significant influence on other areas of linguistic research that are relevant to understanding literature, culture, and history.

7. Most obvious is historical linguistics, which study the ways that language changes over time. A compelling early essay that used a constructionist framework to understanding language change is Michael Israel’s cleverly titled “The Way-constructions Grow.”

8. More recently, Elizabeth Traugott and Graham Trousdale’s book, Constructionalization and Constructional Changes (2013),  recast the key mechanisms of language change (grammaticalization, lexicalization, etc.) in a constructionist framework. Literary and intellectual historians looking to study signs more abstract than keywords will find exemplary models in this volume.

9. Most constructionist approaches are more or less explicitly committed to the notion that grammar is an emergent phenomenon of language use rather than a pre-existing structure (biological or otherwise) that is merely fixed by experience and implemented in use. The idea of emergence was first proposed by Paul Hopper, who remains its most radical exponent, arguing that much of grammatical structure arises in the course of conversation between speakers.

10. The term “usage-based” expresses a similar idea: that we learn the resources for what can be said through our experience of what has been said. Joan Bybee’s usage-based theory is explicitly quantitative: which constructions we learn, and the character of the learned constructions, depends on frequency of use. Quantitative digital humanists are more likely than  humanists at large to take up the statistical claims of usage-based theories.

But Bybee also advances a distinct claim that is, I think, of great relevance to literary history at large. She argues for conceiving of “grammar as based on constructions and as having an exemplar representation in which specific instances of use affect representation.” The productivity of grammar, in her account, is not fully captured by abstract constructions (with open or unfilled variables that we fill in new ways). These constructions are stored with exemplary instances, which also play a role in the production of new utterances. To give an example that I’ve published on: when someone asks What would Michael Jordan do?, they are not just filling in a partially unfilled construction, What would X do? with a new name, they are also basing their utterance on a prefabricated exemplar, What would Jesus do?

Usage-based grammar, in other words, gives us an explicit model of how existing literary works form a linguistic basis – exemplary as well as abstract – for the production of new works.  Though linguists elaborate the theory at the level of the phrase or sentence, it is clearly applicable to generic form as well.  The cultural competence that allows authors to write new novels includes  exemplars (the actual novels that they have read) as well as the repertoire of forms that they possess by abstracting from and generalizing over their reading.

Galaxy Brain.jpeg

So far I have deliberately focused on construction grammar without saying much about the cognitive side of things. To study constructions is to study the nature of the linguistic sign and of the system of signs as, in Saussure’s terms, a “social product” (13), “a collective phenomenon” (19) that is “social by nature” (16). But even if signs are social and collective, they must also have a psychological existence “in the brain of one individual” (11). And to one extent or another, all of the work I’ve cited makes or at least relies on claims about the nature of human cognition – about what human brains do, how their categories are constituted and organized, how they represent knowledge, and what their capacities and limitations are.  This is true, of course, of Saussure (and structuralism more generally): his concept of langue presupposes a particular theory of mind, even if that theory is minimalist and mostly implicit.  Cognitive linguistics, as one might expect, aims to be explicit about its theory of cognition.

Claims about human cognition make humanists like me a bit uncomfortable. I’ll have to explicate the character of this discomfort another time; suffice it to say that it’s not at all obvious where an account of the sign ends and an account of human cognition begins.

11. Eleanor Rosch is a psychologist whose work on categorization has provided one of the foundations of cognitive theory and construction grammar.  She uses ethnographic study and controlled tests to argue that human categorization across cultures is not the product of historical accident but is the result of psychological principles.  Her account of categories, however, is markedly different from the “classical” accounts of philosophers like Aristotle, for whom a category specifies the necessary and sufficient conditions of its members.  Building on Wittgenstein’s idea of “family resemblance,” she shows that categories are not like empty boxes which clearly delineate what things are included in them or not.  Instead they have an internal structure.  They are organized around a “prototype” or best example (as a chair and table are best examples of the category furniture).  They are “graded,” which means that members participate in them to greater lesser degrees as they resemble the prototype (a penguin is a less good example of the category bird than a sparrow ).  They are radial, meaning that the gradations extend along different attributes (a piano is big relative to the prototype of the category furniture, a car seat is more less domestic and more mobile relative to that same prototype).  Their boundaries are frequently fuzzy (color categories fade into one another), and they frequently overlap with other categories (a piano is a peripheral member of furniture but a prototypical member of instrument).  Rosch also challenges the Aristotelian taxonomy, or system of categories, in which categories stack in a relation of genus to species.  She argues (with Brent Berlin) that in human categorization schemes, there is a “basic level” category that, in most taxonomies, is not the highest level or lowest level category, .  So, for example, children learn the category “dog” before “Mammal” or “Dachshund.”

12. George Lakoff, Women, Fire, and Dangerous Things: What Categories Reveal About the Mind (1987).  Lakoff builds Rosch’s experimentally-based prototype theory of categories along with work of others (like Paul Kay on color) into a full-blown account of cognition that is embodied, imaginative, holistic, and ecological.  This is a big and multifarious book, and I won’t summarize it all here.  (Lakoff himself is quite adept at summarizing his own claims in bullet point form.)  Literary scholars are likely to know Lakoff for his work on how metaphor is basic to thought, but so far as I can tell they have not picked up on his central role in developing construction grammar as an alternative to the Chomskyan paradigm.  Along with  Fillmore, Kay, and O’Connell’s “Let alone” article, with whom Lakoff collaborated,  the book’s third case study, on There-constructions, published at about the same time, is a founding document of construction grammar.  The study shows how grammar might be conceived of as pairings of linguistic forms and cognitive models (meaning in a capacious sense).  But it also rethinks the nature of linguistic structure using the theory of categories developed by Rosch and others.

When Saussure conceived of langue as a system of differences without positive terms, he supposed that the system – the “structure” of structuralism – was built out of two kinds of relations: syntagmatic and associative (which Jakobson later dubbed “paradigmatic“) relations. Differences along these two axes (which can also be considered as the operations combination and substitution) can be treated as binary oppositions or simplified to a set of oppositions.  What’s clear, however, is that the relations acknowledged by Saussure – while not at all spurious or superfluous – are on their own insufficient for accounting for the structure of our linguistic knowledge. Lakoff shows how constructions – partially or fully schematic signs – are arranged into radial categories, with less central members that are based on more central ones.  The open slots in these constructions have radial structures as well.  As parings of form and meaning, the relations between constructions are motivated, not arbitrary, though not always strictly predictive.  Metaphor, metonymy, and analogy are constitutive of the grammar, not supplements to it.  The system of constructions is also hierarchical, with more concrete child constructions inheriting the formal and semantic properties of their more schematic parent constructions.

13. Ronald Langacker, The Foundations of Cognitive Grammar  (1987).

Langacker was developing Cognitive Grammar at the same time as scholars in the Berkeley linguistics department were working on the first case studies of Construction Grammar.  But where Fillmore, Lakoff, and others tended to devote extensive attention to a single, representative idiom or construction (X let alone Y, What’s X doing Y) or family of constructions (There-constructions), and then draw out general consequences for our knowledge of language as a whole, Langacker set out to rebuild the system of grammar from the ground up.  Cognitive and Construction Grammar, while conflicting on occasional theoretical details, are deeply compatible enterprises, and Goldberg and other practitioners tend to draw liberally from both.



#. And, of course, humanists interested in seeing how studying abstract constructions with the help of digital tools can expand the scope of philology, and with it our understanding of literary and cultural history, should have a look at my book, Cyberformalism: Histories of Linguistic Forms in the Digital Archive.


Leave a 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