Can linguistics be reunifed? How the “general vs. theoretical” distinction might help

If you are reading this text, you are most probably a p-linguist, at least most of the time. It’s not that I enjoy dividing people into categories, but when there’s conceptual confusion, I think that making up new terms for existing concepts often helps. So we can distinguish between general linguistics and particular linguistics (abbreviated “g-linguistics” and “p-linguistics”), and I have the hope that this distinction might help in reunifying the field of linguistics. So the purpose is not division, but unification, and with this (perhaps overambitious) goal in mind, I wrote the paper “General linguistics must be based on universals” which was just officially published (in Theoretical Linguistics). This blogpost tries to explain my original motivation for writing this paper.

The “g-“ vs. “p-“ distinction may be easy enough to understand (and I reflected on it in earlier blogposts) – a p-linguist is someone who studies a particular language, and since the great majority of linguists study particular languages, the linguist readers of this post are probably p-linguists. They can also be g-linguists at the same time, of course, but how exactly general and particular linguistics relate to each other is the topic of the new paper – I argue that they are related via universals.

Now where does the “general vs. theoretical” distinction come in that I put in the subtitle of this blogpost? What I say in the paper (which is behind a paywall, but is available as a preprint on Lingbuzz and Zenodo) is that this is an orthogonal distinction – theoretical linguistics is what we all do when we do not pursue practical or applied goals. The contrast between theory and practice goes back all the way to G.W. Leibniz (the picture shows the logo of Germany’s Akademie-Verlag, which emphasized its roots in Leibniz’s Prussian Academy).

So it really makes no sense to say that analyses (or descriptions) of particular languages are somehow opposed to “theory”. When we analyze a language (for non-applied purposes), we are engaged in a theoretical activity of analysis. There is no such thing as “atheoretical description”, as has often been emphasized (e.g. in Matthew Dryer’s brilliant 2006 paper on different types of linguistic theories). According to Chomsky (1957), a grammar of a language is a theory of that language. Linguists sometimes oppose “description” to “theory”, as in the title of a recent edited volume on African languages, but does this have any clear scientific meaning? Or is it merely related to a split in communal identities – with some linguists identifying as working on “language documentation and description”, and others identifying as “theorists”? I have been mulling over these questions for quite some time, also in relation to the term “formal linguistics”, which is used by some people and whose scientific (non-communal) meaning is hard to grasp.

Communal identities are a ubiquitous feature of human societies, but they have no legitimate role in science (understood as the quest for universal truths). So to the extent that actually existing scientists can be observed to operate in such ideological communities, I think we must ask whether these opposed communities can be unified. In the case of linguistics, there were no (strong) ideological divisions until the 1960s, so what we should aim for is a “reunification” of the “camps”, to restore the pre-1960 situation.

A concrete example of the unnecessary division of linguistics is the twofold treatment of “Current issues in African morphosyntax” by Denis Creissels and Jeff Good in the 2018 handbook edited by Tom Güldemann. While Creissels deals with “morphosyntactic typology”, Good discusses “formal linguistic frameworks”. The kinds of phenomena that they cover are closely related, but they report on work from two communities. I was struck particularly by one sentence in Good’s paper, where he mentions papers such as Taraldsen’s  (2010) treatment of Nguni gender classes and markers:

“The general focus of these works is to show how well-known patterns can be given a a formal analysis within a particular framework, and they do not generally provide major new descriptive insights.” (Good 2018: 824)

So what is the purpose of the “formal analysis” if it does not provide “descriptive insights”? Here it seems to me that the distinction between p-linguistics and g-linguistics is helpful, in a way in which the terms “formal”, “framework”, “theory” and “description” are not.

What Good is saying is that there have been earlier satisfactory p-linguistic analyses/descriptions of gender in the Nguni languages (especially Zulu and Xhosa), and that what Taraldsen’s work adds is a re-analysis in terms of a specific hypothesis about the innate grammar blueprint (or “universal grammar”): If the claims made by the Nanosyntax framework (about innate features and the organization of grammar) are correct, then Taraldsen’s description may be the true description of Nguni speakers’ mental grammars, and it provides further support for the universal claims of Nanosyntax. Taraldsen does not offer “descriptive insights” into the Nguni languages (i.e. p-linguistic insights, or additions to their p-theory), but his contribution does provide support for a general theory of the domain-specific component of human linguisticality.

But can we say that Creissels is reporting on p-linguistic findings, while Good is presenting findings from g-linguistics? Is this how the two communities differ? The answer is, of course, NO – Creissels is just as interested as Good in general questions, and the authors whose work Good summarizes are mostly working on particular languages. Morever, almost everyone is interested in typology these days. In practice, all of them (and all of us) are p-describers, g-theorists and typologists at the same time, though with different foci. Saying that “typology” is somehow opposed to “theory” makes no sense.

So what is it that divides us? Why can’t we talk to each other, all too often?

Here is an attempt at an answer, in three parts:

(1) We are all theoretical linguists (unless we are applied linguists), i.e. the prestigious label “theory” is not reserved for one particular methodological choice. And to the extent that we are interested in language in general, we are general linguists. (Many theoretical linguists are interested primarily in a single language, or in a group of related or adjacent languages – they typically focus on historical questions, and they are usually happy to say that they do not engage in general linguistics.)

(2) We recognize that p-analyses must be complete, and that general theories are permanently preliminary. This creates an asymmetry: P-descriptions should be informed and inspired by, but cannot rely on general theories, because of their preliminary status. If we write a (hopefully complete) dictionary of a langage, we cannot wait until general lexicology has resolved its disputes, and a comprehensive grammar cannot be based on a grammatical framework that keeps undergoing revisions at a rapid pace. Making progress on general theories has proved extremely hard, while writing a dictionary or a grammar is eminently feasible, and the degree of completeness of such works depends primarily on the available funding. This means that p-analyses/theories are not preliminary in the same way as general theories (at least if they are seen as theories of the social conventions  that guide the language users).

(3) We recognize that there are (at least) two types of general theories (or theory components) that cannot be reduced to each other: cultural-evolutionary theories that explain general properties of human languages as resulting from functional-adaptive pressures, and biocognitive theories that explain general properties of human languages as resulting from innate constraints on possible grammars. It is open to debate for each general phenomenon which approach explains it better, but some phenomena are clearly due to functional-adaptive forces, and some are clearly due to innate predispositions. Cultural-adaptive and biocognitive theories can (and must) thus coexist peacefully, and linguists should not think of their own theoretical orientation as somehow excluding the other (or competing with it wholesale).

Am I too optimistic in thinking that these three points need not be controversial? It is sometimes hard to say what is controversial and what isn’t because linguists keep talking past each other, due to unclear meanings of many terms (e.g. Leivada & Murphy 2021).

Theoretical Linguistics  is an open peer commentary journal, so the issue with my paper on general linguistics also contains commentaries by D. Adger, B. Bickel, R. D’Alessandro, D. Forker, S. Fuchs & L. Paschen, J.-L- Mendívil-Giró, A. Tallman, and D. Zaefferer, to which I got the chance to respond (my reply is also available on LingBuzz). These commentaries partly reflect the problems with terminology, and in particular the problem that arises when one equates “theory” with “general theory”, as is often done.

Thus, I am somewhat hopeful that by distinguishing clearly between general and particular (g-/p-) linguistics on the one hand, and general and theoretical linguistics on the other, we can move a bit closer to reunifying linguistics.

(Or maybe there is a better way – in any event, I don’t think it will work to pretend that “only my community is following the right approach”, or to hope that “my community will get stronger and stronger and the other community will disappear”. Reunification will remain on the agenda until it happens.)


Creissels, Denis & Good, Jeff. 2018. Current issues in African morphosyntax. In Güldemann, Tom (ed.), The languages and Llnguistics of Africa, 709–881. Berlin: De Gruyter Mouton. (doi: 10.1515/9783110421668-007) (Accessed June 16, 2021.)

Dryer, Matthew S. 2006. Descriptive theories, explanatory theories, and basic linguistic theory. In Ameka, Felix K. & Dench, Alan & Evans, Nicholas (eds.), Catching language: The standing challenge of grammar writing, 207–234. Berlin: Mouton de Gruyter. (

Haspelmath, Martin. 2021a. General linguistics must be based on universals (or nonconventional aspects of language). Theoretical Linguistics. (

Haspelmath, Martin. 2021b. How to tear down the walls that separate linguists: Continuing the quest for clarity about general linguistics. Theoretical Linguistics. De Gruyter Mouton 47(1–2). 137–154. (doi:10.1515/tl-2021-2011)

Leivada, Evelina & Murphy, Elliot. 2021. Mind the (terminological) gap: 10 misused, ambiguous, or polysemous terms in linguistics. Ampersand 8. 100073. (doi:10.1016/j.amper.2021.100073)

Taraldsen, Knut Tarald. 2010. The nanosyntax of Nguni noun class prefixes and concords. Lingua 120(6). 1522–1548. (doi:10.1016/j.lingua.2009.10.004)

Cite this blog post
Martin Haspelmath (2021, September 12). Can linguistics be reunifed? How the “general vs. theoretical” distinction might help. Diversity Linguistics Comment. Retrieved April 23, 2024, from

3 thoughts on “Can linguistics be reunifed? How the “general vs. theoretical” distinction might help

  1. > I don’t think it will work to pretend that (…) “my community will get stronger and stronger and the other community will disappear”

    Very true if we define communities purely by their membership or social clustering. On the other hand, it may indeed happen if we define communities instead by their views. No two mutually exclusive theories can come to a conciliation where both will be maintained as such. One or both will have to be concluded incorrect and abandoned. The process could look like a kind of reunification where maybe aspect A of theory 1 and aspect B of theory 2 are maintained to create a yet-new theory 3… but by the working definition this would, too, mean for both earlier theory-defined communities to “disappear”. (Or maybe some will decide we “really” only meant aspect B and not the whole theory, and therefore community 2 is the one that has succeeded over community 1.)

    The $1000 sociology-of-science question of course is: which one of these drives the other? Do linguists first sort themselves into social clusters and then end up adopting different theories from their milieus — or do linguists first adopt different theories and then sort themselves into communities of the like-minded? These would imply different paths for reunification. Considering that linguists are mostly trained, not born, I think it’s primarily the first, but I don’t rule out that the second makes a degree of contribution as well.

  2. Interesting article!
    A question: you state that “In the case of linguistics, there were no (strong) ideological divisions until the 1960s, so what we should aim for is a “reunification” of the “camps”, to restore the pre-1960 situation.”
    I’m not great with history, so I’m curious: what happened in the 1960s that led to the formation of strong ideological camps? [I *think* this is associated with the rise of generative grammar, but I just wanted to make sure.]
    Perhaps analyzing this in a bit more depth might help to understand *why* the divisions formed. With this understanding, perhaps we would understand what specific issues need to be resolved in order for reunification to be possible.

    • Yes, generative grammar came along in the 1960s and emphasized that a break from the past was needed. Not everyone agreed, and then there was even a split among generative grammarians (which was so dramatic that it came to be known as “the linguistics wars” These developments have often been analyzed, but too rarely from a conciliatory point of view. Newmeyer (1998; 2005) was making a serious attempt to compare diverse approaches, and I am trying to build on his work.

Leave a Reply

Your email address will not be published. Required fields are marked *

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