For good reasons, Nicholas (“Nick”) Evans is one of the world’s most prominent linguists. Among his numerous achievements, the best known among experts may be the outstanding grammatical descriptions of Kayardild (1995) and Bininj Kunwok (2003), as well as his work on insubordination (e.g. Evans & Watanabe 2016) and reciprocals (e.g. Evans et al. 2011). But to a wider group of readers, he is of course well-known through the overview paper Evans & Levinson (2009), and the 2009 book “Dying words”, which gives a beautiful nonspecialist overview of worldwide linguistic diversity, its documentation, and its comparison.
Like many great representatives of our discipline, Evans has contributed both to highlighting the structural uniqueness of each language and to efforts at comparing languages at a larger scale. There is a certain tension between the uniqueness of languages and their comparison, and in the last section below, I will say a few more words about why I chose to focus on Evans’s contribution to these issues in this blogpost.
Each language has unique properties and categorizations, and our descriptions must respect these. Somewhat poetically, we can perhaps (using Sapir’s words) talk about a “genius” of a language. It is in particular Franz Boas who gets credited with emphasizing this since the early 1900s:
“Boas made explicit many of the tenets that have become axiomatic in the best descriptive work: the importance of describing each language and each culture on their own terms rather than importing inappropriate European models, the need to discover the inner design of each language inductively through the study of texts, and the scientific responsibility to produce as undistorted a record as possible by setting grammar alongside a comprehensive dictionary and text collection” (Evans 2009: 36)
In another passage of the same book, Evans expresses the same idea in similar terms:
“No matter how hard the linguist has tried to be comprehensive in their grammatical description, and to represent the functionings of the language in a way that does justice to its own unique genius, there will always be phenomena they have overlooked or failed to understand.” (Evans 2009: 223)
This uniqueness is of course reflected in Evans’s own descriptions of Kayardild (1995) and Bininj Kunwok (2003). (A beautiful example involving a Kayardild kinship term is cited below.)
The dangers of carrying over our traditional concepts to all languages
The Boas-Sapir-Whorf skepticism toward the traditional method of carrying over European categories to other languages is well known among comparative linguists. Authors such as Dahl (1990), Haspelmath (2001), and Cysouw (2011) have noted that it may actually be the European languages that are unusual in a number of ways, so we should not take the structures that are familiar from them as a baseline. We have a constant tendency to do this, but we should resist it. As Gil (2001) noted, a good primary-data linguist (“fieldworker”) needs to “unlearn” our ingrained prejudices and be open to surprises.
That any influence from European models should be minimized is now widely recognized among fieldworkers. For example, Epps (2011: 648) says that describers should “produce descriptions in formats that will enable and facilitate comparison across languages, but also remain true to the languages themselves, without forcing them into ill-fitting predetermined categories (Gil 2001; Haspelmath 2007; 2010).” Mithun (2001: 51) highlights the importance of collecting natural texts if we want to “learn about the language in its own terms”, and Chelliah & De Reuse (2011) simply presuppose the need to liberate oneself from a Eurocentric perspective when they assess 19th century linguists, who had not done this fully yet:
“While some early fieldworkers perceived that it was necessary to describe the language as much as possible in its own terms, rather than making descriptions conform to the mold of say, Latin or Sanskrit grammar, it remains true that the large majority of early writers were limited by known grammatical patterns.” (Chelliah & De Reuse 2011: 64-65)
These days, we should no longer be limited in this 19th-century way, and let the languages speak for themselves. As Evans (2009) writes about Wilhelm von Humboldt, one of the early advocates of the systematic and open-minded study of diverse languages:
“The sum total of all cognitive processes enacted in all existing languages at all times of human history constitutes the sum total of all ‘world recognition’ accessible to man.” This is no drifting and facile relativism, but sets out an immense challenge to philological scholarship: that of gradually drawing nearer to reality through the sensitive and far-reaching study of the different vantage points that each language can give us on the world, through the legacy of what has been forged by each people’s distinctive attempt to grasp reality, blazing their own thought-paths down through the millennia.” (Evans 2009: 3)
Describing different languages from our narrow perspective is thus also a kind of lack of respect for other people’s cultural heritage.
The metalanguage-theory conflation
Dryer (2006) has pointed out very clearly that it is a programmatic feature of generative linguistics that the metalanguage (the “formal framework”) for description/analysis should at the same time be explanatory. This seems quite wrong, as it is different from what most other sciences do (except perhaps physics?), and Evans & Levinson (2009) arrive at the same diagnosis. Concerning the idea of Optimality Theory (OT) that “the grammar of one language inevitably incorporates claims about the grammars of all languages”, they say, quite rightly:
“Instead of putting the filtering where it belongs, in cultural system evolution across generations, OT effectively burdens each individual mind with a précis of the functional history of all known human languages, and loads the entire optimization process onto on-line grammatical computation… This conflation of the metalanguage with the object of description is a peculiar trick of the generative tradition… The complex representational structures look undermotivated, and covert processes proliferate where alternative models deftly avoid them… Once explanatory theories hook external factors (e.g. psycholinguistic or evolutionary factors) into the account, this conflation of cognition and metalanguage can be dropped.” (Evans & Levinson 2009: 474)
The conclusion is thus that the metalanguage for description must primarily be adequate for the language in question. Explanatory theories can be built in a next step, with whatever concepts are appropriate for them. Such theories will usually involve comparison between languages, so we first need comparative concepts before we can then link our generalizations to broader explanatory factors.
Comparison and the traditional categories
Comparison has sometimes been attempted on the basis traditional categories, but this is problematic. Linguistics has many traditional terms that everyone knows from the textbooks and from salient exemplars, e.g. “verb” and “complementizer”, but also terms that are primarily used for non-European languages such as “classifier” (e.g. Aikhenvald 2000) and “ideophone” (e.g. Voeltz & Kilian-Hatz 2001). However, we cannot assume that these traditional terms exhaust the possible categories in human languages:
“Further unfamiliar word classes are continuously being unearthed that respect only the internal structural logic of previously undescribed languages. Even when typologists talk of “ideophones”, “classifiers”, and so forth, these are not identical in nature across the languages that exhibit them… no two languages have any word classes that are exactly alike in morphosyntactic properties or range of meanings (Haspelmath 2007).” (Evans & Levinson 2009: 435)
Indeed, since the beginning of modern comparative linguistics (such as Wilhelm von Humboldt’s work of the 1820s), new phenomena have been found and have been given new labels (Humboldt coined the terms incorporation and agglutination). And here’s a similar quotation, again citing my (2007) paper approvingly:
“[Traditional terms like subject, adjective, inflection, syllable, pronoun…] are not absolute universals… Rather, they re descriptive labels, emerging from structural facts of particular languages… Consequently, for the most part they do not have precise definitions shared by all researchers, or equally applicable to all languages (Haspelmath 2007).” (Evans & Levinson 2009: 439)
As a result, it is not immediately clear how rigorous comparison between languages should be carried out. What kinds of concepts work across languages?
Comparison through innate categories?
Many linguists have tried to solve the comparability problem by adopting the idea that there is a small set of building blocks which is part of the innate blueprint for grammars (I have also called this the “Mendeleyevian Vision”). In the responses to Evans & Levinson (2009), this is advocated, for example, by Nevins (2009: 461), who says that comparison should be done by means of “formal universals”, which in phonology “are constituted by the analytic elements that human minds employ in constructing representations of sound structure. … [they] refer to the set of available data structures (e.g., binary features, metrical grids, autosegmental tiers) and the possible operations on them that can be used in constructing a grammar of a language.”
But as Levinson & Evans (2010: 2738) note, there is no reason to make this assumption, because “even the strongest universal patterns could emerge from multiple cnstraints in populations of language users and learners, without residing in the heads of speakers”. Evans & Levinson regard the innateness claim for the formal categories of a grammatical system as “a fundamental category mistake, like confusing mathematics with the object it describes – a nautilus shell exhibits a logarithmic spiral, but it doesn’t use logarithms to produce its shell.” More recently, Evans put it like this:
“[For comparative purposes, we have been] developing an ontology for calibrating coding (and for quantifying uncertainty/ambiguity of analysis), and functional accounts of distributions. Though typologists have been steadily working away at this, this remains a work-in-progress, leaving linguists “like chemists without a list of the elements, or physicists with no account of particles” (Corbett 2012: xiv). “ (Evans 2016: 507)
Since we do not have a list of elements, we are very far from the Mendeleyevian Vision, that there is a smallish sets of elements which can be used both to describe and compare languages. Thus, we have to make our own comparative concepts, as Levinson & Evans (2010) recognize:
“Language-specificity of categories raises problems for [comparison and explanation], as the typologists have become increasingly aware (Dryer 1997; Haspelmath 2007; 2010). It doesn’t follow that comparison is impossible, only that it has to be undertaken in an auxiliary language designed to generalize over language-specific categories.” (Levinson & Evans 2010: 2737).
There are different types of comparative concepts in this special “auxiliary language”. Let us start with etic comparative concepts.
Etic comparative concepts: Nonverbal stimuli
One way to avoid the pitfalls of traditional terms is to gather fresh data (rather than to rely on existing descriptions), based on a set of comparative concepts created by the comparativist: Nonverbal stimuli (such as pictures or videos) that can be held constant across languages. This is the well-known “Nijmegen method”, illustrated nicely by the field manuals which can be found on the MPI Nijmegen site). One such project was led by Nick Evans, on “Reciprocals Across Languages”, which was based on video clips shown to speakers of 20 different languages (Evans et al. (eds.) 2004; 2011). In the introduction of the resulting book, the editors write about the methdology of semantic typology:
“Within semantic typology, there is growing realisation of then need to use standardised stimuli in studies of categorisation, so as to place all languages on a level playing field, rather than privileging one as the source from which translations from the others are sought. …this also provides a way of subjecting phenomena to cross-linguistic comparison without having to predefine the categories of interest in a way that may import accidental characteristics of some familiar reference languages.” (Evans et al. 2011: 24)
By using nonverbal stimuli, one makes sure not to confuse the etic concepts for comparison with emic categories of particular languages (especially with emic categories of well-known European languages). But on the other hand, describing such emic categories is also an interesting aspect of research on smaller languages:
“We stress that in no way does the gathering of carefully matched data stand in the way of discovering language-specific categories, since the structure of this book balances systematic multifactorial comparison with traditional language-specific analyses in Chapters 3-18. Again and again, these chapters emphasise language-particular semantic preoccupations… These differences in the meaning ranges of constructions emerge from the intersection of language-internal exploration with the comparative grid made available by a set of systematic stimuli – the two approaches complement each other, together yielding more than the sum of the parts.” (Evans et al. 2011: 25)
So while description and comparison complement each other, they do not depend on each other. Comparison depends on a comparative grid which is as independent of any particular language as possible in practice, and description “emerges from the structural facts” of each language. While the comparative grid certainly provides useful data for an emic description, this grid can be neither sufficient nor necessary for determining the language-particular structures.
Etic comparative concepts that can be described verbally
Reciprocals are a particularly varied domain, and there are areas of semantic typology where verbal description is in fact possible. One such domain is kinship, where researchers often use nonverbal notation (the well-known kinship diagrams used by anthropologists, e.g. Murdock’s six patterns of kinship), but this nonverbal notation is quite useless for the speakers. If you are an anthropologist who wants to ask me how I describe my father’s sister’s son in my home language, it is best to use ordinary language (e.g. the target language, in this case German, though other languages would work very well, too; incidentally, the answer is the loanwords Cousin [kuzɛŋ], though my mother still used the older native term Vetter).
Likewise, for comparison of kin terms, we can get quite far by using ordinary-language terms like “parent/child”, “male/female”, “older/younger”. This is how Evans (2011: 509-510) describes the contrasts between the kinship systems for siblings in English, Indonesian, Japanese and Kayardild:
We can see that each language has its own system, and using the comparative concepts for the description of each language would be quite nonsensical: “Older/younger” plays no role in English, the sex of the referent plays no role in Indonesian (kakak simply means ‘older sibling’), while Japanese has for terms where both of these concepts do play a role. However, this does not mean that there is a closed set of concepts by which all languages can be described and compared at the same time. To describe Kayardild, we need a still different concept: the relation between the sex of the speaker and the sex of the referent. The word kularrind refers to a sibling of the opposite sex, and just as it makes no sense to say that English sister has two meanings (‘older sister’ and ‘younger sister’), it makes no sense to say that Kayardild kularrind has four different meanings (‘ani’, ‘ane’, ‘otooto’, and ‘imooto’, to use the Japanese terms).
Again, we see clearly that comparison is separate from description, even when the comparative concepts can be readily expressed verbally, and thus might be mistaken for categories of a language. Levinson & Evans (2010: 2736) are very clear on distinguishing four “basic kinds of linguistic undertaking”:
(a) accessible language description,
(b) formal analysis of a language,
(c) language comparison, and
(d) the development of explanatory theories.
These are related, but distinguishable tasks – and (c), the “task of mapping the world’s diverse linguistic structures”, crucially involves “developing cross-linguistically viable ontologies”, i.e. universally applicable comparative concepts.
Thus, languages are comparable because we make them comparable, through artificial comparative concepts – either of the etic type, as with video stimuli, or of the verbal type, as with the typology of kinship terms. They are not comparable by themselves – each language has a different set of emic categories. Among linguists who do not subscribe to the Chomskyan hope of finding an innate blueprint for languages consisting of a specific toolkit for grammars, this is really uncontroversial.
But the stereotypes of the terms and categories that we inherited from the Greek and Roman grammarians is powerful, and we sometimes create new stereotypes (e.g. “incorporation” or “serial verb construction”, which are often treated as if they were entities that exist independently of particular languages or their comparisons). For this reason, linguists such as Dryer (1997), Croft (2001), Gil (2001), Lazard (2005), Cristofaro (2009) and Haspelmath (2010) have reasserted the old insight, without making any claims of novelty. There was some discussion of these issues in 2016 on the Lingtyp list (and in a number of 2016 papers in the journal LT), and at the 2017 ALT conference in Canberra, there was an interesting plenary panel on comparability organized by Nick Evans, with nice talks, among others, by the ANU scientists Lindell Bromham (bringing in the perspective from biology) and Kim Sterelny (bringing in a more general perspective from philosophy), and also by typologists Ulrike Zeshan and Grev Corbett. After the talks, David Gil asked how these talks related to the issues discussed in 2016 (by Dryer, Croft, Gil, Moravcsik, and others), and the answer was: “No relation – we wanted to discuss comparability in different terms here.” That was a puzzling answer, because Nick Evans, at least, has long been emphasizing the same kinds of issues that led Dryer, Lazard and others to insist on the description of each language in its own terms, and he has proposed the same kinds of solutions. So part of the reason I wrote this blogpost is to dispel any misconception that there are any substantive disagreements among comparative linguists (who are not committed to the Chomskyan ideas) and fieldworkers about how to compare and how to describe languages. There are none. (The serious disagreement with the Chomskyans persist, however, and I wish I understood them better…)
Aikhenvald, Alexandra Y. 2000. Classifiers: A typology of noun categorization devices. Oxford: Oxford University Press.
Chelliah, Shobhana & De Reuse, Willem. 2011. Handbook of descriptive linguistic fieldwork. Dordrecht: Springer.
Corbett, Greville G. 2012. Features. Cambridge: Cambridge University Press.
Cristofaro, Sonia. 2009. Grammatical categories and relations: Universality vs. language-specificity and construction-specificity. Language and Linguistics Compass 3(1). 441–479.
Croft, William. 2001. Radical construction grammar: Syntactic theory in typological perspective. Oxford: Oxford University Press.
Cysouw, Michael. 2011. Quantitative explorations of the world-wide distribution of rare characteristics, or: the exceptionality of northwestern European languages. In Simon, Horst J. & Wiese, Heike (eds.), Expecting the unexpected: Exceptions in grammar, 411–432. Berlin: De Gruyter Mouton.
Dahl, Östen. 1990. Standard Average European as an exotic language. In Bechert, Johannes & Bernini, Giuliano & Buridant, Claude (eds.), Toward a typology of European languages, 3–8. Berlin: Mouton de Gruyter.
Dryer, Matthew S. 1997. Are grammatical relations universal? In Bybee, Joan L. & Haiman, John & Thompson, Sandra A. (eds.), Essays on language function and language type: Dedicated to T. Givón, 115–143. Amsterdam: Benjamins.
Dryer, Matthew S. 2006. Functionalism and the theory-metalanguage confusion. In Wiebe, Grace & Libben, Gary & Priestly, Tom & Smyth, Ron & Wang, Sam (eds.), Phonology, morphology, and the empirical imperative: Papers in honour of Bruce Derwing, 27–59. Taipei: The Crane Publishing Company. (http://www.acsu.buffalo.edu/~dryer/foundation.htm)
Epps, Patience. 2011. Linguistic typology and language documentation. In Song, Jae Jung (ed.), The Oxford handook of linguistic typology, 634–649. Oxford: Oxford University Press.
Evans, Nicholas. 1995. A grammar of Kayardild: With historical-comparative notes on Tangkic (Mouton Grammar Library 15). Berlin: Mouton de Gruyter.
Evans, Nicholas D. 2003. Bininj Gun-wok: A pan-dialectal grammar of Mayali, Kunwinjku and Kune. 2 vols (Pacific Linguistics, 541). Canberra: Australian National University.
Evans, Nicholas. 2009. Dying words: Endangered languages and what they have to tell us. Hoboken, NJ: Wiley.
Evans, Nicholas. 2011. Semantic typology. In Song, Jae Jung (ed.), The Oxford handbook of linguistic typology, 504–533. Oxford: Oxford University Press.
Evans, Nicholas. 2016. Typology and coevolutionary linguistics. Linguistic Typology 20(3). 505–520. (doi:10.1515/lingty-2016-0023)
Evans, Nicholas D. & Levinson, Stephen R. & Enfield, Nick J. & Gaby, Alice & Majid, Asifa. 2004. Reciprocals. In Majid, Asifa (ed.), Field Manual, vol. 9, 25–30. Nijmegen: Max Planck Institute for Psycholinguistics.
Evans, Nicholas & Levinson, Stephen C. 2009. The myth of language universals: Language diversity and its importance for cognitive science. Behavioral and Brain Sciences 32(5). 429–448. (doi:10.1017/S0140525X0999094X)
Evans, Nicholas D. & Gaby, Alice & Levinson, Stephen C & Majid, Asifa (eds.). 2011. Reciprocals and semantic typology. Amsterdam: Benjamins.
Evans, Nicholas D. & Gaby, Alice & Levinson, Stephen C & Majid, Asifa. 2011. Introduction. In: Reciprocals and semantic typology. Amsterdam: Benjamins.
Evans, Nicholas & Watanabe, Honoré (eds.). 2016. Insubordination. Amsterdam: Benjamins.
Gil, David. 2001. Escaping Eurocentrism: Fieldwork as a process of unlearning. In Newman, Paul & Ratliff, Martha (eds.), Linguistic fieldwork, 102–132. Cambridge: Cambridge University Press.
Haspelmath, Martin. 2001. The European linguistic area: Standard Average European. In Haspelmath, Martin & König, Ekkehard & Oesterreicher, Wulf & Raible, Wolfgang (eds.), Language typology and language universals: An international handbook, vol. 2, 1492–1510. Berlin: Walter de Gruyter.
Haspelmath, Martin. 2007. Pre-established categories don’t exist: Consequences for language description and typology. Linguistic Typology 11(1). 119–132.
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.
Levinson, Stephen C. & Evans, Nicholas. 2010. Time for a sea-change in linguistics: Response to comments on ‘The myth of language universals.’ Lingua 120(12). 2733–2758.
Mithun, Marianne. 2001. Who shapes the record: The speaker and the lingust. In Newman, Paul & Ratliff, Martha (eds.), Linguistic fieldwork, 34–54. Cambridge: Cambridge University Press. (doi:10.1017/CBO9780511810206.003)
Nevins, Andrew. 2009. On formal universals in phonology. Behavioral and Brain Sciences 32(5). 461-462.
Voeltz, F. K. Erhard & Kilian-Hatz, Christa (eds.). 2001. Ideophones. Amsterdam: Benjamins.