The necessity of grammatical structures

A great deal of digital ink has proliferated (I won’t say has been ‘spilled’ because that would imply it was done in waste) about the question of linguistic complexity, and whether it is possible to show in a meaningful way that some languages are more or less complex than others. After reading DeGraff’s (2001) and others’ commentary on McWhorter’s (2001) well-known article, ‘The world’s simplest grammars are creole grammars’, I have recently come to reject the question as having any meaning, except perhaps in the impressionistic sense of complexity as being ‘harder for adult language learners to acquire’. At the very least, I have yet to see a precise formulation of complexity that I am convinced captures the idea that linguists are attempting to pin down, and since I myself have no alternative definition to offer, I shall refrain from using the concept until such a definition presents itself. I also wonder what an investigation into the relative linguistic complexity of different languages would actually yield, since no one to my knowledge has suggested that the relative simplicity or complexity of a language has any significance for our understanding of the faculty of language, its acquisition, how languages change over time, etc. Many have of course suggested that complexity or simplicity is the result of these various factors, but not that complexity in turn has an impact on any other aspect of language, for example the rate of first language acquisition (but I admittedly have not read all of the relevant literature on complexity to know this for certain).

While I am therefore wary of wading into the discussion of linguistic complexity, there is an idea mentioned in both McWhorter’s paper and several of the commentaries that was accepted rather uncritically and otherwise discussed very little, and which I think it would be productive to examine more critically. This it the idea that certain grammatical structures are unnecessary to communication, and it forms a key part of McWhorter’s position on linguistic complexity, as the following quote from the abstract of his 2001 paper shows:

“creole languages display less overall grammatical complexity than older languages, by virtue of the fact that they were born as pidgins, and thus stripped of almost all features unnecessary to communication” (abstract)

McWhorter’s (and others commenting on the paper) idea of necessary versus unnecessary features of language is strongly informed by his belief in an innate Universal Grammar, as seen in the following:

“in older grammars, millennia of grammaticalization and reanalysis have given overt expression to often quite arbitrary slices of semantic space, the result being a great deal of baroque accretion which, while compatible with Universal Grammar, is incidental to it, as well as to even nuanced human expression.” (126)

McWhorter considers complex constructions, and therefore much of the grammatical structure of languages generally, to be little more than unnecessary and accidental grammatical fluff, not part of the communicative core of language. He illustrates his point by discussing grammatical gender:

“Grammatical gender affixes are so starkly devoid of semantic substance or syntactic function that their fundamentally happenstance essence is particularly clear, and they are useful in illuminating the similarly contingent nature of other features often found in grammars.” (129)

It should also be noted that McWhorter sees features like grammatical gender as an affront to functionalist approaches to language, and quotes Lass (1997:367-368) to this end:

“Not only does a language have by definition a sufficient set of structures and categories available for doing anything that a speaker “needs” to do; it will also have a lot of material that simply makes speakers do things, whether or not there is any (functional, discourse, pragmatic) “need” to do them. […] We live perpetually with “decisions” of past generations. Somebody, somewhere (as it were) decided in the eighteenth century or thereabouts that the expression of progressive aspect should be obligatory in English, and as an English speaker I’m simply stuck with it.” (Lass 1997:367-368, quoted in McWhorter 2001:130).

The point that I would like to question here is that certain grammatical structures are necessary for communication and others not. This theory is only tenable if one adheres to an innatist understanding of language where certain aspects of grammatical structure are part of the genetic faculty of language itself. In this understanding, these features of language are ‘necessary’ in the Aristotelian sense of ‘necessary and sufficient conditions’. That is, they are definitionally part of language as defined by Universal Grammar. Any grammatical structures which are not proscribed by Universal Grammar are therefore merely contingent or incidental features of the languages that have them.

The first major problem with this approach to understanding creoles is that, if we reduce the necessary features of language to only that which is shared by every extant language, it’s not clear there would be anything left, except perhaps hierarchical relationships between words (the exact center of the debate between Daniel Everett and the innatists regarding Pirahã). More directly relevant, creoles obviously contain many more ‘unnecessary’ features than this irreducible minima of structures requires; not even the most ‘simplified’ of creoles comes close to matching this core of necessary features.

The more serious problem with this approach is, from a functionalist perspective, the idea that any specific grammatical structure can be part of the language faculty in the first place. For functionalists, the language faculty affords the creation of grammatical structures, but does not itself proscribe them. In other words, all grammatical constructions are contingent rather than necessary features of language. What is necessary (in the Aristotelian sense) about language lies in humankind’s ability to create symbolic representations and associations between those representations, which then give rise to the structures of grammar in a complex-adaptive and emergent way.

McWhorter uses the concept of grammatical necessity in non-Aristotelian ways as well, when talking about the morphosyntactic coding of features that are not always relevant to what the speaker is attempting to communicate, such as definiteness in English (161), or grammatical gender as mentioned above. In his view, these are things that speakers are “forced” to communicate, and are thus unnecessary and even anti-functional to the communicative ends of the speaker. They are part of the the “leftover baggage” of historical change:

“Obviously, grammatical gender marking is unnecessary to human communication, given how very many human languages lack it. Grammatical gender affixes, beyond the extent to which they distinguish natural (biological) gender, do not mark any real-world entity or category or serve any communicative need (see Trudgill 1999 for a useful discussion of this point). To be sure, the free classifier morphemes (Greenberg 1978) or former pronouns that gender markers usually arise from did mark real-world categories. However, as affixes, they no longer do, or at best correspond to perceivable categories only very approximately, their fundamental superfluity to communication being illuminated further by the fact that they sometimes even arise from happenstance phonetic correspondences (Nichols 1992: 141–142).” (129)

While McWhorter acknowledges that “happenstance accretion” does begin as a matter of so-called communicative necessity, this misses the broader point that communicative necessity doesn’t just vanish. It is not as though linguistic innovation happens and is then suddenly over. Andersen makes this point forcefully:

“each and every step in such a development is an innovation. not only the initial act. through which a new linguistic entity comes into being. It is through innumerable individual acts of innovation – of acceptance. adoption. and acquisition – that any new entity gains currency and enters into competition with traditional entities in the usage of a linguistic community.” (Andersen 1989:14)

All grammatical constructions continue to hold some type of functional load for as long as speakers use them, and the the linguistic system is perpetually adjusting to reflect that load. If any grammatical construction ever became truly a-functional, I suspect it would disappear overnight.

If McWhorter wants to say that certain grammatical structures have become communicatively unnecessary, then at what point did these ‘unnecessary’ features stop being necessary? Can it really be said that such features are any less necessary today then they were when they first appeared? Were they ever really ‘necessary’ in the first place?

The necessity of grammatical structures from a functional-discourse perspective is that they follow a community of practice by speakers, and if one wants to be understood one must participate in those practices, i.e. follow the socially accepted conventions of the language (so-called ‘grammatical rules’). So if speakers were following the conventions of their language when the feature first appeared, and are still following the conventions of the language now that that structure has become grammaticalized or fossilized, is it any less necessary now than it was at its inception?

Reframing this slightly, the functional need that so-called complex or unnecessary grammatical structures fulfill is the need to communicate using the conventional set of signs most appropriate to the communicative goal of the speaker. In other words, what is ‘necessary’ about a particular grammatical construction is the fact that it is shared between multiple members of the speech community, and thus the only one available for communicating what is intended. One needs to use the constructions of a language if they want to communicate in it, just as one needs to use the words of Navajo if they wish to be understood by Navajo speakers. Words are no more or less necessary than inflectional marking or any other type of construction, especially when we understand that, from a constructional perspective, there can be no clear demarcation between grammar and lexicon in the first place. From a constructional perspective, then, McWhorter’s claim can be rephrased as saying that certain constructions are necessary for communication, and others are not. But since there is no principled way in which one could draw a distinction between necessary and unnecessary constructions, McWhorter’s dichotomy should be rejected as a false one.

In conclusion, from a definitional perspective of what constitutes language, all grammatical structures are contingent. None are inherently contained within the language faculty. But from a functional perspective of the communicative purpose of language, all grammatical structures are necessary, in that they all aim at the communicative ends of the speaker using them.


  • Andersen, Henning. 1989. Understanding linguistic innovations. In Leiv Egil Breivik & Ernst Håkon Jahr (eds.), Language change: Contributions to the study of its causes (Topics in Linguistics, Studies and Monographs 43), 5-27. Berlin: Mouton de Gruyter.
  • DeGraaf, Michel. 2001. On the origin of creoles: A Cartesian critique of Neo-Darwinian linguistics. Linguistic Typology 5:213-310.
  • Greenberg, Joseph H. 1978. How does a language acquire gender markers? In Joseph H. Greenberg (ed.), Universals of Human Language, Vol. 1, 47-81. Stanford: Stanford University Press.
  • Lass, Roger. 1997. Historical linguistics and language change. Cambridge: Cambridge University Press.
  • McWhorter, John H. 2001. The world’s simplest grammars are creole grammars. Linguistic Typology 5:125-166.
  • Nichols, Johanna. 1992. Linguistic diversity in space and time. Chicago: Chicago University Press.
  • Trudgill, Peter. 1999. Language contact and the function of linguistic gender. Poznaø Studies in Contemporary Linguistics 35:133-152.

Daniel W. Hieber

I am a graduate student in linguistics at University of California, Santa Barbara, specializing in typologically-informed language documentation and description in North America and East Africa. I previously worked at the language-learning company Rosetta Stone, where I created software for Navajo, Inupiaq, and Chitimacha as part of the Endangered Language Program, and also did research for our commercial language products. I received my B.A. in linguistics and philosophy from The College of William & Mary in 2008.

More Posts - Website

3 thoughts on “The necessity of grammatical structures

  1. I disagree, in that completely illogical and functionless distinctions can ‘stick around’ in language easily – e.g. the denn/weil difference in word order in German (one taking SVO order and the other SOV), or wh-fronting in English, or suppletions, etc. These things persist because of the speaker’s need to conform, and the surprising lack of power that people have normally to change their language (e.g. it would difficult to get English speakers to regularise plural forms or to abandon wh-movement).
    As for the issue of complexity, it has a fairly straightforward definition in principle, which is counting the number of rules (or amount of information) needed to for the speaker to correctly produce a certain sentence type. e.g. Mandarin only requires one word order rule to produce SVO sentences and questions, whereas English requires the additional rule of wh-fronting. Quantifying rules is of course a difficult task, but there is nothing in principle wrong with this definition of complexity.

  2. Agreed. If some gram is wholly useless, it’ll have a hard time sticking around. It has to pass through the bottlenecks of production and comprehension every single time it is used and that cost has to be paid by some measure of utility. In this connection, am fond of the following quote from Keller 1998:

    “From a logical perspective, the categories produced by a natural language often are rather confused and crazy. What counts in evolutionary processes is not logic, but utility.”

    There is a slightly more charitable interpretation of McWhorter’s basic idea though, and this is that some types of grams may be less useful than others in the particular communicative situations in which pidgins originate.

    We know that gender marking is actually quite useful in communication for e.g. reference tracking (cf. work by Ellen Contini-Morava). So it is emphatically not merely ballast; it serves a purpose. Yet if I speak a language with ten genders and you don’t, and we find ourselves in a situation where I want to you hand me that wrench, I’ll likely focus more on getting you to understand that I want you to GIVE me that WRENCH you MORON than bothering about sharing the intricacies of my grammatical agreement system. On this view, the fact that pidgins may start out as ‘lowest common denominators’ has little to do with a supposed universal grammar and more with the constraints on communication when you’re doing real-time negotiation of mutual understanding without a shared code.

    Finally, like you I wouldn’t equate the idea of ‘lowest common denominator’ (seen merely from the perspective of the two systems you start out with) with ‘lack of complexity’. I think we know way too little about the processes of sense-making involved in the amazingly complex communicative situations in which shared codes evolve out of previously separate languoids. You could almost say that communication in the absence of a shared code is more complex, as you don’t have the well-worn conventionalised grams of countless generations to build on and there is a much heavier load on processes of pragmatic inferencing and multi-modal integration.

Comments are closed.