How useful are retro-definitions for (typological) linguistics? (Maybe not very.)

Like all sciences, linguistics needs technical terms, and we generally treat the grammatical terms that we inherited from our ancestors (such as syllable, affix, compound, dative, imperative, subordination) as technical terms: We list and define them in terminological dictionaries (such as Matthews’s  “Concise Oxford Dictionary of Linguistics”), and we expect students to learn them (textbooks often have appendixes with lists of terms). So in a practical sense, we take them quite seriously.

But from the point of view of our larger general-theoretical ideas, we tend to be less sure. Different phonologists have different views on “syllables”, different morphologists have different views on “affixes”, and different syntacticians have different views on “subordination”. When linguists move beyond the textbooks, they often dismiss traditional terminology, because many of us think that only our theories can provide a basis for defining terms in a proper way. Traditional usage is thought to be “pre-theoretical” and therefore of little relevance to modern theoretical linguistics.

This seems to be a widespread view, and until about a decade ago, I shared the view and did not regard it as particularly important to think about consistent and widely understood definitions of traditional terms. But then my views changed, and I began to advocate retro-definitions (see also this blogpost).

There are two main reasons why I now think that linguists should have clear, simple and sharp definitions of widely used terms:

– We should take our terms seriously as technical terms, being fully aware that they are not the last word.

– By providing clear and sharp definitions, we make it clear that determining what a term means is not a research question.

Linguists often seem to use terms only in a preliminary way (reflected in caveats such as “working definition”), but in addition to preliminary terms used in research contexts (subject to revision as research advances), we also need some non-preliminary terms, for example (i) for explaining the preliminary terms, (ii) for teaching purposes, and (iii) for summaries of our research to outsiders.

It may well be that research on grammar is not advanced enough to engage in the sort of terminological rigor that other disciplines are practicing. Since the mid 19th century, botanists have treated the Linnean conventions for species naming as standard, and other displines followed in the later 19th and early 20th century. What linguists did during that age was create a standard phonetic alphabet, and one might wonder whether grammatical terminology could also be standardized (I first suggested this possibility in a 2017 blogpost). But clearly, any such effort is still quite far in the future. There have been attempts at creating “ontologies” for grammatical research (e.g. Farrar & Langendoen 2003; Klimek et al. 2017), which would be similar to standard sets of terms, but these have not been widely discussed (let alone adopted) in the field.

Maybe one problem is that linguists typically think that the field is not advanced enough to settle on definitions. But if more research led to greater clarity of terms, we would find clear examples of this. Instead, it seems that grammatical terms get less clear as time passes. The lack of clarity of the terms serial verb construction and existential construction is lamented by the following authors, and they seem to imply that this is because there has not been enough research on these constructions:

“Despite the by now impressive literature on serial verb constructions…, there is still surprisingly little agreement on what exactly defines serial verb constructions.” (van Staden & Reesink 2008:21)

“Despite the extensive literature on [existential constructions], there is to date no agreement on how to define “existential sentences”…” (Bentley et al. 2015: 1)

However, one may also suspect that it was precisely the additional research that led to less clarity about the meanings of the terms, because each new paper used them in a slightly different sense. More generally, does terminological clarity really depend on deeper theoretical understanding? Or is it perhaps the other way round, that deeper understanding (possibly coming from fruitful interactions of linguists who do not talk past each other) depends on terminological clarity? (I wrote about this earlier in this 2021 blogpost)

It seems to me that linguists often have a fairly good shared understanding of the extension of a term, but they cannot agree on a definition, because even some of the most basic terms (such as morpheme, word, feature, or formal) are unclear. Take the example of the term construction, which is extensionally fairly clear: In practice, a construction is a schema that allows one to create a complex expression by filling one or more slots. But the definition of construction is not clear, because of the movement of Construction Grammar, where constructions are taken to be an important theoretical entity. Constructions have been taken to include not only schemas, but also expressions without an empty slot (words and morphs; Goldberg 1995), and there have been discussions about the degree of entrenchment required before something is considered a construction. But these discussions are not really relevant for the definition of construction for grammatical research – because de facto, linguists usually deal with grammatical patterns as social conventions, not with grammatical patterns in the minds of individual speakers. Thus, the following definition (Haspelmath 2023) is not really controversial:

Some linguists are skeptical about sharp definitions because they think that the phenomena of languages are due to grammaticalization, which is necessarily gradual. I thought so for a long time myself, as illustrated by my 1992 paper on “heads in morphology” (see §1, p. 71), and again in my 2011 paper on “the gradual coalescence into “words” in grammaticalization”. A more recent reflection of this idea is found in Lehmann’s (2020) paper on univerbation (a special subtype of grammaticalization), e.g.

“univerbation is itself a gradient process which displays phases of weaker and stronger univerbation. In principle, univerbation takes place at the moment that a construction is converted into a word. Since, however, the concept of word itself does not have neat boundaries, this process is not, in fact, an instantaneous conversion, but rather a transition.” (Lehmann 2020: 209)

However, in recent years I have come to recognize that it is not possible to measure the gradualness of grammaticalization if we do not have fine-grained sharp comparative concepts. Perhaps these sharp concepts could be different from the traditional terms of grammar, but in any event, we need some basic sharply defined terms. If all our terms are fuzzy, there is no way we can demonstrate the fuzziness of reality.

One of the most basic terms in grammatical research is the notion of boundness, because the distinction between morphology (word-internal structure) and syntax (word-combining patterns) depends on it. Is there a “gradient of fusion”, as suggested by Bickel & Nichols (2007: 180)? (See this blogpost for further discussion.) This would probably mean that the syntax-morphology distinction is gradient, but in any event, we would be able to show the gradience of “fusion” (or “boundness”) only if we had more basic sharply defined terms.

Thus, we need basic sharply defined terms, and clearly defined traditional terms – which we can achieve by retro-defining the traditional terms (I first talked about retro-definitions in my 2021 paper). Thus, I have proposed definitions for terms such as affix, clitic, compound, negative polarity item, passive, and quite a few others.

Now to get back to the usefulness of these definitions (mentioned in the title of this blogpost): What is it that we can do with them? Himmelmann (2016; 2021) emphasizes the need for comparative concepts to be “useful”, and he therefore rejects the idea that they are “arbitrary” (in the sense of Lazard 2002; 2005). And in my original 2010 paper on comparative concepts, I said myself that…

“…comparative concepts are concepts created by comparative linguists for the specific purpose of crosslinguistic comparison. … They are not psychologically real, and they cannot be right or wrong. They can only be more or less well suited to the task of permitting crosslinguistic comparison.” (Haspelmath 2010: 665)

So how “well-suited” for comparison is a concepts such as “incorporation” when it is defined (rather arbitrarily) as “an event-denoting verb-noun compound construction in which the noun occupies an argument slot of the verb and occurs in a position where nominal patient arguments cannot occur”. (Haspelmath 2025: 2)

I frankly don’t know, and in my more recent papers, I have not repeated the suggestion that comparative concepts should be well-suited for comparison. On the contrary, I sometimes say explicitly that we do not know how “useful” a traditional concept is. For example, in the concluding section of my forthcoming paper on antipassive constructions, I note that antipassives are quite heterogeneous (citing Heaton’s 2020 extensive research), and that

“…because of this heterogeneity, it is quite likely that the prominence of the term antipassive is due to an accident of the history of linguistics, not due to a special importance of the constructions that fall under this label.”

Most of our traditional terms have become well-known in linguistics because of such accidents of history. Terms such as “article”, “genitive”, “word” and “subordination” go back to earlier periods of our discipline, when linguists were primarily interested in Greek and Latin, and comparative perspectives going beyond Indo-European were added only slowly (with typological linguistics really taking off only in the 1970s). With some of the more recent terms, the accidental origins are particularly clear (e.g. with “clitic”, going back to Nida 1949; with “serial verb construction”, going back to Stahlke 1970; with “negative polarity item”, going back to Baker 1970), but the earlier terms are not any less accidental. It could be that with a different history, linguists would rely on rather different basic terms now.

So why should we bother about these accidental traditional terms and their definitions? I already said it above: Because they are treated as technical terms, and without clearly and sharply defined technical terms, it is hard to say anything useful about grammatical patterns in general linguistics – and because linguists have a tendency to think of these terms as corresponding to something real (and of their definitions as a task for research). By providing fairly arbitrary definitions, we can make it clear that the research tasks lie elsewhere (comparing languages and finding universals, so that we can learn about the nature of human language)

In a recent Facebook discussion, Egbert Fortuin asked me:

“My question is whether there are any guiding principles for a good comparative concept. Are there a set of criteria that need to be met in order to arrive at a successful one? Or is your only criterion that you are clear about the term? Or that the term is simple and memorable? I can hardly accept that this is really all there is to it.”

Indeed, when it comes to retro-defining, the only criteria are clarity, simplicity and memorableness (plus of course the need for a retro-definition to correspond largely with the traditional extension of the term). However, these are the criteria for retro-definitions of traditional terms, not for all comparative concepts.

Comparative concepts are very diverse (see Haspelmath 2018) – even a Bible verse can serve as a comparative concept for parallel-text study using parallel Bible texts (e.g. Mayer & Cysouw 2014). Many of them are custom-made for the purposes of a special study, but some of the terms are taken off the shelf”, using traditional terms – and these need retro-definitions, so that it is clear what is meant by them. But since I am not super optimistic about the usefulness of many of the grammatical terms, the analogy with IPA is only partial – because IPA was clearly designed to cover the space of possibilities and to make all (or most of) the important distinctions (it was not primarily designed to give clear definitions of traditional letters, though that was part of it). In morphosyntax and grammatical semantics, we seem to be quite far from a consensus idea of what “the space of possibilities” is (maybe because the space is larger than in phonology?).

References

Baker, C. L. 1970. Double negatives. Linguistic Inquiry 1(2). 169–186.

Bentley, Delia. 2017. Copular and existential constructions. In Dufter, Andreas & Stark, Elisabeth (eds.), Manual of Romance morphosyntax and syntax, 332–366. Berlin: De Gruyter. (doi 10.1515/9783110377088-009)

Bentley, Delia & Ciconte, Francesco Maria & Cruschina, Silvio. 2015. Existentials and locatives in Romance dialects of Italy. Oxford: Oxford University Press.

Bickel, Balthasar & Nichols, Johanna. 2007. Inflectional morphology. In Shopen, Timothy (ed.), Language typology and syntactic description, vol. III: Grammatical categories and the lexicon, 169–240. Cambridge: Cambridge University Press.

Farrar, Scott & Langendoen, D. Terence. 2003. A linguistic ontology for the semantic web. Glot International 7(3). 97–100.

Goldberg, Adele E. 1995. Constructions: A construction grammar approach to argument structure. Chicago: The University of Chicago Press.

Haspelmath, Martin. 1992. Grammaticization theory and heads in morphology. In Aronoff, Mark (ed.), Morphology now, 69–82. Albany: SUNY Press. (https://zenodo.org/records/225299)

Haspelmath, Martin. 2010. Comparative concepts and descriptive categories in crosslinguistic studies. Language 86(3). 663–687. (doi:10.1353/lan.2010.0021)

Haspelmath, Martin. 2011. The gradual coalescence into “words” in grammaticalization. In Narrog, Heiko & Heine, Bernd (eds.), The Oxford handbook of grammaticalization, 342–355. Oxford: Oxford University Press.

Haspelmath, Martin. 2018. How comparative concepts and descriptive linguistic categories are different. In Van Olmen, Daniël & Mortelmans, Tanja & Brisard, Frank (eds.), Aspects of linguistic variation: Studies in honor of Johan van der Auwera, 83–113. Berlin: De Gruyter Mouton. (https://zenodo.org/record/3519206)

Haspelmath, Martin. 2023. On what a construction is. Constructions 15(1). (doi:10.24338/cons-539)

Haspelmath, Martin. 2025. Compound and incorporation constructions as combinations of unexpandable roots. Zeitschrift für Wortbildung / Journal of Word Formation 9(1). 1–23. (doi:10.21248/zwjw.2025.1.124)

Heaton, Raina. 2020. Antipassives in crosslinguistic perspective. Annual Review of Linguistics 6(1). 131–153. (doi:10.1146/annurev-linguistics-011619-030412)

Himmelmann, Nikolaus P. 2016. What about typology is useful for language documentation? Linguistic Typology 20(3). 473–478. (doi:10.1515/lingty-2016-0020)

Himmelmann, Nikolaus P. 2022. Against trivializing language description (and comparison). Studies in Language 46(1). 133–160. (doi:10.1075/sl.19090.him)

Klimek, Bettina & McCrae, John P. & Lehmann, Christian & Chiarcos, Christian & Hellmann, Sebastian. 2017. OnLiT: An Ontology for Linguistic Terminology. In Gracia, Jorge & Bond, Francis & McCrae, John P. & Buitelaar, Paul & Chiarcos, Christian & Hellmann, Sebastian (eds.), Language, Data, and Knowledge, 42–57. Cham: Springer International Publishing. (doi:10.1007/978-3-319-59888-8_4)

Lazard, Gilbert. 2002. Transitivity revisited as an example of a more strict approach in typological research. Folia Linguistica 36(3–4). 141–190. (doi:10.1515/flin.2002.36.3-4.141)

Lazard, Gilbert. 2005. What are we typologists doing? In Frajzyngier, Zygmunt & Hodges, Adam & Rood, David S. (eds.), Linguistic diversity and language theories, 1–23. Amsterdam: Benjamins.

Lehmann, Christian. 2020. Univerbation. Folia Linguistica 54(s41–s1). 205–252. (doi:10.1515/flih-2020-0007)

Matthews, P. H. 2007. The concise Oxford dictionary of linguistics (Oxford Paperback Reference). Oxford, United Kingdom: Oxford University Press.

Mayer, Thomas & Cysouw, Michael. 2014. Creating a massively parallel Bible corpus. Proceedings of the international conference on language resources and evaluation (LREC), 3158–3163. Reykjavik: European Language Resources Association (ELRA). (https://www.cysouw.de/home/presentations_files/cysouwmayerPARALLELLREC.pdf)

Nida, Eugene A. 1949. Morphology: The descriptive analysis of words (2nd edition). Ann Arbor: University of Michigan Press.

Stahlke, Herbert F.W. 1970. Serial verbs. Studies in African linguistics 1(1). (http://elanguage.net/journals/index.php/sal/article/viewArticle/929) (Accessed March 13, 2014.)

van Staden, Miriam & Reesink, Ger P. 2008. Serial verb constructions in a linguistic area. In Senft, Gunter (ed.), Serial verb constructions in Austronesian and Papuan languages (Pacific Linguistics 594), 17–54. Canberra: Pacific Linguistics.


OpenEdition suggests that you cite this post as follows:
Martin Haspelmath (November 28, 2025). How useful are retro-definitions for (typological) linguistics? (Maybe not very.). Diversity Linguistics Comment. Retrieved December 13, 2025 from https://doi.org/10.58079/158lw


2 thoughts on “How useful are retro-definitions for (typological) linguistics? (Maybe not very.)

  1. Interesting post as usual!
    A question: you mention that
    In morphosyntax and grammatical semantics, we seem to be quite far from a consensus idea of what “the space of possibilities” is (maybe because the space is larger than in phonology?).

    I’m sure there’s no consensus view on the “space of possibilities”, but do you happen to know of any works that have attempted to define this space?

    • Not really. Linguists have often said that the space must be limited, but there are no concrete proposals along the lines of the proposal for a distinctive feature set in phonology (by Halle & Jakobson in the 1950s, and Chomsky & Halle 1968), or the proposals for a set of semantic primitives (by Wierzbicka in many different works since the 1970s).

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.