Types of pronouns: Beyond the stereotype

Everyone knows what a stereotypical pronoun is: they, she, he; and all linguists know that there are also interrogative pronouns (who, what), relative pronouns (who, which), and demonstrative pronouns (this, these). Or should we say “demonstrative adjectives”, because they are typically used in adnominal function (this room, these chairs)? And is a pronoun a type of noun, i.e. a “pro-noun” in the literal sense (standing for a noun)? If so, are there “pro-adjectives” and “pro-adverbs” as well?

There is a fair amount of variation in linguists’ use, and since we generally give specific examples when we talk about such phenomena, we usually understand each other nevertheless. However, if we want to have technical terminology as a disciplinary standard, we need precise definitions of such commonly used terms. We need more than stereotypical examples that give us a rough sense of what these terms mean.

First of all, I don’t think we want to move from the familiar “pronoun” to “pro-noun”, and to using “pro-adverb” and “pro-adjective” consistently whenever we refer to interrogative pronouns like where or indefinite pronouns like some. We don’t want to say “pro-form” whenever we don’t care about the difference, because this would be a major shift in usage. (It is true that “pro-form” exists, and Wikipedia even has a page on it, but linguists rarely use this term.)

For my 1997 study of indefinite pronouns, I considered the issue and decided to use “pronoun” in a broad sense (not in the etymological sense) (Haspelmath 1997: 10-11):

Note also that a “nominal” pronoun usually stands for a nominal (phrase), not for a noun, so to be precise, one would have to say pro-nominal anyway. (Actually, pronominal is sometimes used in the literature, at least for anaphoric pronouns, and was made especially popular by Chomsky’s (1981: §3.2.3) “binding theory”; however, it is unclear why pronominal replaces the more usual term pronoun in this literature.)

In grammars of Ancient Greek and Latin, there is a tradition of presenting a table of “correlative pronouns”, with demonstratives, interrogatives, indefinites and relatives. I found a particularly nice example of such a table here: http://www.cultus.hk/latin_lessons/declension/correlative1.html

The Wikipedia section on “table of correlatives” attributes this mode of presentation to L.L. Zamenhof  and his work on Esperanto, but he was building on an older tradition. For example, Kaegi https://la.wikipedia.org/wiki/Adolphus_Kaegi (1889) has the following table for Ancient Greek, and he did not invent the “table of correlative pronouns” either:

Thus, even though the term pronoun is often used as an abbreviation of “personal pronoun” (which are the most stereotypical pronouns), there is a long tradition of using it more broadly.

And this tradition is still very much alive. In May 2020, I did a Twitter poll asking my colleagues what they wanted to include in the term “pronoun”, and a clear majority did indeed include not only personal pronouns, but also interrogative and indefinite pronouns (for simplicity, I left out relative and demonstrative pronouns).

So I conclude that personal pronouns are just one type of pronoun, and we should not use “pronominal” exclusively for personal pronouns. (Thus, I would reject the common expression “pronominal affix” for person indexes, as I discussed in my 2013 paper. Likewise, the expression “pro-drop” for situations in which a person index does not need a conominal is confusing and is best avoided. Thre is more to pronouns than the stereotype.)

As a first approximation, we can work with the following definitions:

locuphoric personal pronoun (e.g. I, you, we): a form that denotes a speech role (speaker/producer and/or hearer/comprehender)

anaphoric personal pronoun (e.g. he, she, they): an anaphoric form that does not contain a noun and that can be used in a complement clause coreferentially with a matrix clause argument

interrogative pronoun (e.g. who, what, where, when): a form that stands for the open parameter in a content question

indefinite pronoun (e.g. someone, something, somewhere): a form that can be used for indefinite reference in the absence of a noun

relative pronoun (e.g. who, which, where): a form that occurs at the periphery of a relative clause and that represents the relatum in the relative clause

demonstrative pronoun (e.g. this (one), that (one), there): a form that can be used for deictic reference in the absence of a noun

(In this blogpost, I am not going to talk more about the definition of anaphoric personal pronoun, which was discussed on the LINGTYP list in July 2021. But it should be noted that locuphoric and anaphoric personal pronouns are semantically quite different, so there is no general definition of “personal pronoun”.)

One might ask whether there is a general definition of “pronoun”, such that the six pronoun types are  special kinds of pronouns. I don’t think there is, and I won’t pursue this further. (Saying that a pronoun is “a form that substitutes for some other form” does not work, of course, because one can also go in the reverse direction and replace a pronoun by a noun.)

Now the next issue is the relationship between demonstrative pronouns (‘this one’) and demonstrative determiners (‘this N’). As was shown by Holger Diessel in his WALS chapter (from 2005) on “Pronominal and adnominal demonstratives”, there are quite a few languages which have different forms (either different stems or different inflectional markings) for these two types of demonstratives. For example, French distinguishes between demonstrative determiner ce/cette and demonstrative pronoun celui/celle.

Similarly, languages also have interrogative determiners (which N?), indefinite determiners (some N), and perhaps even relative determiners (?They couldn’t open the door, which problem they had not foreseen; some languages allow such constructions). These three kinds of determiners are somewhat marginal, and in my 1997 book, I simply swept the indefinite determiners under the rug of the term indefinite pronoun. It did not seem worthwhile to create a new cover term only because one member of a series of indefinite pronouns (someone, something, somewhere, sometime, some, …) is not a pronoun in the strict sense. But the situation is not quite satisfactory, and ideally one would have a broader cover term for pronouns and determiners that one can use when one needs to be precise.

Next, what about demonstrative adverbs, like here and so, – are they demonstrative pronouns, too? Diessel (1999: Chapter 4) distinguishes between pronominal, adnominal, and adverbial demonstratives, so he does not include them. As a cover term for all three, he uses simply “demonstratives” – which is fine, because there are no other kinds of demonstrative expressions. But this does not work so well for interrogative, indefinite and relative pronouns. As Diessel (2003) emphasizes, demonstrative and interrogative pronouns show many similarities, and one might say that English which is an “adnominal interrogative” (or “interrogative determiner”). However, there are other kinds of interrogative expressions, in particular interrogative particles and interrogative clauses (as well as interrogative phrases https://wals.info/chapter/93; e.g. which house is an interrogative phase, while that house would not be called a “demonstrative phrase”). There are also other kinds of indefinite expressions (indefinite articles and indefinite nominal expressions), and other kinds of relative expressions (in particular relative clauses). Thus, saying that somewhere is an “indefinite adverb” does not work well, and saying that the time when includes a “relative adverb” when does not work so well either.

[Finally, a brief comment on the terms possessive pronoun and possessive adjective. In the earlier tradition of Indo-European linguistics, linguists often said that adnominal genitive personal pronouns like my/your/her are “possessive adjectives”, while independent genitive personal pronouns like mine/yours/hers are “possessive pronouns”. I agree with Huddleston & Pullum (2002: 471) that this is not a good way of talking about these elements, because they are of course all personal pronouns. Their terms dependent/independent genitive personal pronoun are much better, and Ye (2020) has adopted this terminology for a worldwide comparative study. Similarly, one might also talk about dependent/independent demonstratives, but I have not seen this yet.]

Thus, there is a very good case that all the forms of a correlative table are best called pronouns, even if they are adverbs or determiners. Moreover, many languages have special quality pronouns (quality demonstratives like such, quality interrogatives like ‘what kind’, e.g. Latin qualis) and quantity pronouns (quantity demonstratives like French tant ‘so many’ and quantity interrogatives like French combien ‘how much/many’). These are neither adverbs nor determiners, and they could be called “pronominal adjectives”. Finally, some languages even have interrogative ordinals (e.g. German der wievielte ‘the how-many-eth’; but there is no corresponding demonstrative: *der sovielte). The kinds of ontological categories that are expressed by demonstratives, interrogatives, indefinites and relatives are quite diverse (even including interrogative verbs: Hagège 2008), so as a cover term, the best choice seems to be pronoun. Clearly, “pronouns” are not a “part of speech” like noun, adjective, verb, but represent a different dimension, cross-cutting these classes.

And a brief final remark: In the definitions proposed above, I said that a pronoun is a kind of form; I did not say that it is a “kind of word”, because “word” status is not relevant. Many pronouns are composite forms, e.g. something, and French quelque chose ‘something’. Whether one writes these as one word or not is irrelevant. Instead of “interrogative pronoun”, linguists sometimes say “interrogative word”, but this would seem to exclude multi-word forms. So this is another reason for using pronoun in the broad sense, for the six kinds of pronouns defined above.


Chomsky, Noam A. 1981. Lectures on government and binding. Dordrecht: Foris.

Diessel, Holger. 1999. Demonstratives: Form, function, and grammaticalization. Amsterdam: Benjamins.

Diessel, Holger. 2003. The relationship between demonstratives and interrogatives. Studies in Language. John Benjamins 27(3). 635–655. (doi:10.1075/sl.27.3.06die)

Diessel, Holger. 2005. Pronominal and adnominal demonstratives. In Haspelmath, Martin & Dryer, Matthew S & Gil, David & Comrie, Bernard (eds.), The world atlas of language structures, 174–177. Oxford: Oxford University Press. (https://wals.info/feature/42A)

Hagège, Claude. 2008. Towards a typology of interrogative verbs. Linguistic Typology. De Gruyter Mouton 12(1). 1–44. (doi:10.1515/LITY.2008.031)

Haspelmath, Martin. 1997. Indefinite pronouns. Oxford: Oxford University Press. (https://library.oapen.org/handle/20.500.12657/30819)

Haspelmath, Martin. 2013. Argument indexing: A conceptual framework for the syntax of bound person forms. In Bakker, Dik & Haspelmath, Martin (eds.), Languages across boundaries: Studies in memory of Anna Siewierska, 197–226. Berlin: De Gruyter Mouton.

Huddleston, Rodney & Pullum, Geoffrey K. 2002. The Cambridge grammar of the English language. Cambridge: Cambridge University Press.

Kaegi, Adolf. 1889. Griechische Schulgrammatik: Mit Repetitionstabellen als Anhang. Berlin: Weidmannche Buchhandlung.

Ye, Jingting. 2020. Independent and dependent possessive person forms: Three universals. Studies in Language. John Benjamins 44(2). 363–406. (doi:10.1075/sl.19020.ye)

5 thoughts on “Types of pronouns: Beyond the stereotype

  1. I think this whole discussion about “pronoun” would become clearer by more strongly distinguishing syntactic function from content. As for the content, I actually quite like the general term “pro-form” as discussed on Wikipedia (linked in the article): “function word or expression that stands in for another word, phrase, clause or sentence where the meaning is recoverable from the context.” (I myself would change this definition, but it is a start). Different kinds of pro-forms can simply be named without using the term ‘pronoun’ at all: demonstratives, content interrogatives, relativizers, indefinites, person markers, etc. This should be clearly separated from the syntactic function. For example, content interrogatives can be noun phrases (who, what), adjectives (which, how many), adverbials (when, where, why), etc. In this approach, what is traditionally called “personal pronoun” becomes a “person marking noun phrase” (always using a content term and a syntactic function term).

    • Thanks, Michael, for joining the discussion! You thought a lot about personal pronouns two decades ago, and your 2003 book (on “person marking”) was very helpful – also in putting the focus on “person”, rather than “pronoun”. Sometimes I think that we’d need a special short form (such as “personic”), because these elements are so important in languages (and so frequent in discourse in most languages). Linguists often forget that there are other types of pronouns, and your book (together with Anna Siewierska’s 2004 book “Person”) has reminded us that person forms are really very special, and actually not that similar to demonstrative/interrogative/relative/indefinite pronouns. Yes, the syntactic function (adverbial, adnominal, nominal) is a separate dimension from the “content”, but is there a shared content? That isn’t clear to me. “Where the meaning is recoverable from the context” (the passage that you cite) does not fit for interrogatives or indefinites, I would say. Actually, I’m not sure that anything is shared by all pronouns (or “proforms” if you prefer). In my term database, I currently define “pronoun” as “one of the six types ‘locuphoric pronoun’, ‘anaphoric pronoun’, etc.”. Do you have a better suggestion? – As for your other suggestion, I thought about it, and it does work well for demonstratives (as seen in Diessel’s 1999 book): “Different kinds of pro-forms can simply be named without using the term ‘pronoun’ at all: demonstratives, content interrogatives, relativizers, indefinites, person markers, etc.” This does NOT work for indefinite pronouns (because “indefinites” sounds like “indefinite nominals/articles”), and it does NOT work for relative pronouns (because there are other kinds of relativizers that are not pronouns, e.g. participial affixes), and the term “person marker” is too broad: There are BOUND person form (what is increasingly known as “(person) indexes”, following my 2013 paper), and INDEPENDENT personal pronouns. Both of them are person forms, and Siewierska uses “person marker” in a broad sense to encompass both. However, I would like to use “marker” in a narrower sense, only for bound forms (see https://dlc.hypotheses.org/1826). So given these considerations, there is a preference for using “pronoun” broadly. But it is true that some of the arguments go through if one makes other choices the way I have proposed to make them.

      • I think the term “indefinites” works wonderfully when you distinguish “indefinite noun phrases” from “indefinite determiners”. It’s the syntactic function that distinguishes. One could still debate whether both are the same “content” of course. Likewise for “relativizers”: they might be bound or not, but again, that is morphosyntax, not content. And as for person markers, if you want to use the term “marker” in a specific way, you could try “personiv” 😊

  2. Oh, thanks! Sure, we understand “der sovielte”, but I thought it was never used. – AS for Coseriu’s “Kategorem”, I’m afraid it doesn’t ring a bell…

  3. I object to “sovielte” not being attested


    Eine Anleitung verlangt nach vier Strängen eines Garns mit 275 m Lauflänge bei 100 g. Du möchtest dies nun in DROPS Cotton Merino nacharbeiten und rechnest die benötigte Menge an Knäueln aus. Der *sovielte* Buchstabe des Alphabets ergibt den ersten Buchstaben des Lösungsworts!

    (“Solve some equation. The *so-manieth* letter of the alphabet is the first letter of the solution word”)

    And of course “der so-und-so-vielte” ‘The however-mani-eth’

    I also feel that Coseriu’s idea of Kategorem is relevant here (I have forgotten the details, unfortunately, but since you normally know the history of linguistics so well, I was surprised that you did not mention it)

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.