“Semantic Primitives”, fifty years later

Full Text

1. Introduction

Are there any concepts that all human beings share? Three centuries ago, Leibniz was convinced that indeed there are such concepts, and he called this hypothetical set of universal concepts “the alphabet of human thoughts”. He wrote, for example: “the alphabet of human thoughts is the catalogue of primitive concepts, that it, those concepts which cannot be made clearer by means of any definitions”, and “the alphabet of human thoughts is the catalogue of those concepts which can be understood by themselves and from whose combinations our other ideas arise” (Couturat 1903: 430, 435, cf. Wierzbicka 2001, 2011).

The idea was still widely known and discussed in the 18th century (see, for example, D’Alembert 1759), but in the 19th century it faded from philosophical discourse and eventually it was largely forgotten. In 1963, however, it was revived by the Polish linguist Andrzej Bogusławski.¹ A few years later, it was taken up in my own work, and in 1972, in my book Semantic Primitives, a first hypothetical set of universal human concepts – “semantic primitives”, as they were then called – was actually proposed. It included 14 elements.

At the time when Semantic Primitives was published, I was living in Warsaw and drew my inspiration from the European linguistic tradition. Apart from Bogusławski in Warsaw, my main interlocutors were in Moscow and they included Aleksander Žolkovskij, Igor Mel’čuk, Jurij Apresjan and Elena Paducheva. I also spent a year in America, at the MIT, listening to lectures by Noam Chomsky and his associates, but I wasn’t attracted by their ideas, and when I returned to Poland in 1967 I was confirmed in my goal: to search for Leibniz’s “alphabet of human thoughts”, through linguistic study of meaning, embodied in the languages of the world.²

“Search” is the operative word here: for me, it was not a matter of constructing a system that would “work”, but of searching for the truth, in accordance with the long European tradition epitomised by titles such as “La recherche de la vérité” (Descartes, 1684) and “De la recherche de la vérité” (Malebranche, 1674).

From the start, I thought, as did Bogusławski, that in principle, it should be possible to find the truth about the ultimate elements of human thinking through in-depth exploration of a single language – any language. At the same time, it seemed clear that in practice, a focussed semantic study of many different languages would be a necessity too – if only because a single human life would not be sufficient for the experimentation (the process of trial and error) needed to identify the semantic primes of one language without investigating many others at the same time.

From this point of view, emigrating to Australia and joining the Australian National University in 1973 was a great blessing, as it led to many diverse languages being studied from the “semantic primitives” point of view and, after a decade or so, brought about a radical expansion of the inventory of primes.

Graduate students and other scholars at the Australian National University sought to apply the semantic primitives approach to Australian Aboriginal languages such as Yankunytjatjara (Cliff Goddard) and Arrente (David Wilkins, Jean Harkins), to Chinese (Hilary Chappell), to Ewe (Felix Ameka), to Mbula (Robert Bugenhagen), and many others.

Crucially, my emigration to Australia resulted in a close collaboration with Cliff Goddard. In fact, it was he who suggested the name under which our theory, and the practice based on it, is now generally known: NSM, from “the Natural Semantic Metalanguage”. Since the mid-1980s Goddard and I have been developing the NSM theory as equal partners.

Thus, from the early 1980s, more and more linguists, experts in many different languages and language families joined in testing the expanding set of semantic primes held as universal against an increasing range of languages and domains. As a result, for three decades or so, the set steadily grew. (In Peeters’ 2006 book, Semantic Primes and Universal Grammar it included 61 primes). The expansion stopped in 2014, when the number of 65 primes was reached (see Goddard 2018: 33–37, Gladkova & Larina 2018).

Two books appeared in that year: my own Imprisoned in English and another, co-authored by Cliff Goddard and myself, entitled Words and Meanings: Lexical Semantics Across Domains, Languages and Cultures. Both books announced that the number of 65 primes was reached, and both expressed the authors’ confident conviction that that was it. Thus, in Imprisoned in English I wrote:

Extensive semantic investigations conducted over many years, by many scholars, in the NSM framework, have led to the conclusion that there are sixty five primes, the same in all languages (p. 34).

Similarly, in Words and Meanings Cliff Goddard and I wrote:

After nearly forty years of sustained research, both within selected individual languages and across many languages, linguists in the NSM program are prepared to claim that they have discovered the complete inventory of simple universal concepts that are embedded in the lexicons of all (or most) human languages. To say this is not to deny that much further work is necessary, nor does it rule out the possibility of further revisions to the current inventory. The claim is, however, that a plausible, stable, and well-evidenced set of “universal words” have been identified (…) strictly speaking, the units we are talking about are not words as such, but word meanings. These putatively indefinable word-meanings are known as semantic primes and they are 65 in number (p. 12).

In this paper, I will review the developments which have taken place since those words were written. I will re-affirm our belief that we have identified, in full, the shared “alphabet of human thoughts” and that it includes 65 semantic primes. I will also examine the recurring claims that one of these primes, which we now call have parts, does not pass the test of universality.

2.   What is at stake

Many scholars who debate the plausibility of the existence of a shared “alphabet of human thoughts” treat the question as purely theoretical: one of countless “academic questions” discussed in universities, without any great significance “in the real world”.

They are mistaken. A shared set of human concepts makes it possible to establish a shared human lingua franca, a “Basic Human” in which messages of global significance can be formulated and exchanged, across all parts of planet earth. In particular, if a charter of global ethics is ever to be agreed on – or even meaningfully discussed – by representatives of different traditions, it needs to be formulated in cross-translatable words.

There is a vital connection between shared human concepts and cross-translatable words. Those who believe in shared human concepts but not in shared human words often miss the point that if there were any shared human concepts not embodied in actual words, they could not be used for a global exchange of messages and views.

For example, a charter of global ethics requires not only the universality of the concepts good and bad, but also the availability, in all languages, of some words embodying these concepts. Representatives of different traditions cannot sit around a table and discuss what is good and what is bad if they don’t have some words for the concepts good and bad. And if they are going to accept English as their working language, they need to rely in their discussion on those English words which are cross-translatable into other languages of the world. Otherwise, the dialogue will degenerate into an exercise in what Carsten Levisen called “conceptual colonialism”.³

The wide-spread assumption that the Anglo-English concept of ‘fairness’ is a valid tool for global dialogue is a good case-in-point (for a demonstration of the cultural specificity of this concept see Wierzbicka 2006; 2014). Thus, the question is not only: “Do all people on earth have shared concepts?”, but also, “Do all people on earth have cross-translatable words in which those shared concepts can be expressed?” To put it differently, the question is not only: “Does humankind have a shared conceptual mother tongue?”, but also “Can people speak to each other in that shared mother tongue?”

For example, if there is to be an international round-table discussion about the issue of the sale of human body parts, the participants need to have a shared concept of “parts of the body” and some cross-translatable words or phrases to express that concept. What is at stake, then, is not only the question of human unity in some theoretical sense, but also human solidarity and human communication in a very practical sense.

One of the most memorable sentences in the King James Bible comes from a line in the Acts of the Apostles, from St Paul’s speech to the Greeks in Athens (Acts 17:26): “[God] hath made of one blood all nations of men to dwell on all the face of the earth.”

If we accept that we are all related by blood, as members of the same human family, then we are, as Pope Francis puts it in his new encyclical (of 3 October 2020), “Fratelli Tutti” (the title of the official English translation is “Brothers and Sisters All”). But as many modern thinkers have pointed out, a deep awareness of human unity requires something else as well. In the words of the German philosopher Karl-Otto Apel, it requires that we see all people on earth as a “universal community of communication” (1972).

But there are six or seven thousand different languages in the world. Can a universal community of communication be established in a world divided by thousands of different languages?

As I see it, the answer to this question depends on the availability of shared concepts which can be expressed in cross-translatable words and phrases.

3. An example: talking about the trade in human body parts

Consider again the issue of the trade in human body parts, which has already been alluded to and which is an important topic in international discourse. For example, in his encyclical “Fratelli Tutti” Pope Francis speaks against “all conditions akin to slavery” with special reference to “an abomination that goes to the length of kidnapping persons for the sake of selling their organs” (Section 2.4, Online). Can this issue be discussed in all languages, or only in some? Assuming for the moment that the discussants have a word meaning “to sell” at their disposal, can one say something like this in any language:

“It is bad if people want to sell parts of people’s bodies”?

Some linguists have claimed that in the languages of their expertise there is no word for PARTS (see e.g. Nash and Wilkins 2021). Could the speakers of such languages discuss the trade in human body parts?

My own expectation is that they could. Before I show how I think they could do it, I will present three other tenets of a hypothetical “charter of global ethics” formulated in NSM (for many other such tenets, see Wierzbicka 2018):

1. It is bad if people want to do bad things to other people.
2. It is bad if people want to do bad things to other people’s bodies.
3. It is bad if people want other people to feel something very bad in their bodies.

Suppose that we want to add to these three (and many others like them) a tenet condemning the trade in human body parts; and that we want to formulate this tenet in a way that would make it cross-translatable – even into languages without a special word corresponding to the English word “part” as used in the phrase “part of the body”. How could we do it? To put it differently, how could the speakers of such a language condemn such trade?

Here is my hypothesis, based on a trail of evidence going back to 1994: they could say the equivalent of the following sentences:

Human bodies [or: our bodies] have many “things”,
some of these “things” are inside the body
(heart is one of them, liver is another, there are others).
It is very bad if someone wants to sell these “things”.

It seems uncontroversial that in this context the word glossed as “things” expresses the same meaning as the English word “parts”. So, in some languages it may not be possible to talk about the trade in human body parts as succinctly as in English. This doesn’t mean, however, that a word like “things” doesn’t do the job in a specific lexico-grammatical frame. (For earlier discussion, see e.g. Goddard & Wierzbicka 1994: 46, Wierzbicka 1996: 60, 2007: 25-27, Goddard 2002: 30).

In this connection, it is interesting to note that the outstanding Warlpiri lexicographer Paddy Patrick Jangala opens many of his definitions of Warlpiri body part terms with the phrase glossed in the Warlpiri Dictionary as “that which we all have”; and that he emphasises that a given body part term applies to human bodies generally. For example:

Pirlkiri

Pirlkiri that’s this (one) that we all have at the top of our shoulders, Aboriginal and White people, that’s pirlkiri” (quoted in Nash and Wilkins 2021 p. 9).

In other words, pirlkiri is not just the upper part of a particular person’s shoulders, but the upper part of the shoulders in the human body as such.

In this context, it is worth emphasising that when Paddy Patrick Jangala says “that [which] we all ‘have’ [in Warlpiri mardarni]” he is not talking about ownership or “possession” but about the structure of the human body. In our article “Talking about bodies and their parts in Warlpiri” (2018), Cliff Goddard and I proposed two main lexical exponents for the prime which we now prefer to call have parts, and we emphasised the important role of the verb mardarni ‘to have’ as used in sentences in which the subject is not a person but a thing, or a body. Schematically, we proposed that in sentences like “the body has (mardarni) many things, head, arms, legs, and others” the verb mardarni does not indicate “ownership” but “having parts”.

This is in fact consistent with what the Warlpiri Dictionary (Laughren et al. 2006) says, since it attributes a separate sense to this usage of mardarni: “Definition: Y is a part of X”. In their critique of our treatment, however, Nash and Wilkins reject our interpretation and affirm: “It is not that mardarni ‘have’ is the Warlpiri reflex of part; possession is clearly the relevant notion.” (2021, footnote 12).

But objects and bodies cannot “possess” or “own” anything, in the ordinary sense of these words. To my mind, a sentence like “our bodies have [mardarni] many things, head, arms, legs, and others” clearly refers to the part-whole relationship, and not to “ownership” or “possession”. As Cliff Goddard and I have discussed in a recent article on the meta-category of “possession” (2019), this fictitious category created by linguists represents an aggregation of three diverse semantic schemas which centre on three conceptual anchor points: ownership, body parts, and kinship. Bodies do not “own” or “possess” their parts. I will come back to this shortly.

4.   Generalisations versus exemplars

It is undoubtedly true that while some languages (e.g. English and other European languages) favour abstract generalisations such as “the body has many parts”, many others (e.g. Warlpiri and other Australian languages) avoid such abstract generalisations and favour the use of exemplars (either instead of explicit verbal generalisations or in addition to them). It seems obvious that this difference in ways of speaking has profound cultural underpinnings. But avoidance of abstract generalisations unsupported by exemplars is one thing and the absence of lexical resources for making such generalisations is another (see Wierzbicka 1996: 61, 2007: 26, Goddard 2002: 30).

Consider for example how the translators of the Warlpiri Bible have rendered the famous sentence about the body having many parts in St Paul’s First letter to the Corinthians (1 Cor 12:12):

English

Just as a body, therefore, has many parts, but all its parts form one body, so it is with Jesus Christ.

Warlpiri (the text and the gloss provided by Warlpiri Bible translator Steve Swartz)

Yapa-kurlangurlu palkangku-ka mardarni jurru, rdaka-jarra, wirliya-jarra manu panu-kari. Panu-juku kalu jinta-jarrimi yapangka palkangka jintangka-juku. Ngula-piya-yijala Jijaji Kirajiji manu ngalipa yapa nyanungu-nyangu.

(A person’s body has head, two arms, two legs and many other (things). The many (all) of them become/are one in a person’s body in one. Just like that also (are) Jesus Christ and us His people.”

Strikingly, the Warlpiri translator has added some exemplars (head, arms, legs) which are not mentioned in the English version, or in the Greek original (although other exemplars – the hand, the eye, the ear – are mentioned in the immediate context). At the same time, the combination of the words palka ‘body’, mardarni ‘have’ and manu panu-kari ‘many other (things)’ does convey the same generalisation that would be expressed in English with the phrase “many other parts of the body”.

Furthermore, the sentence which follows makes a generalisation referring to all parts of the body without any exemplars. Swartz glosses this sentence as follows: “the many/all (of them) are one in a person’s body in one”. Nash and Wilkins (2021), who also cite these sentences form St Paul’s letter, gloss this sentence slightly differently: ‘A person’s body has a head, two hands, two feet and many others. The many are united in a person’s single body.’

As I see it, in this context, the phrase glossed by Swartz as “many other (things)” and by Nash and Wilkins as “many others” means exactly the same as the English phrase “many other parts”, and the phrase glossed as “the many” means exactly the same as the English phrase “the many parts”. This conclusion is fully consistent with the “folk definitions” of many “body part words” included in the Warlpiri Dictionary (Laughren et al. 2006) and the glosses provided for them. For example, the word pawiyi ‘spine’ is glossed in the Dictionary as follows: “Pawiyi is the part of our body that is down below the back of the neck and between both our shoulders.” As I see it, the material cited in the Dictionary strongly supports the view that Warlpiri does have lexical resources enabling the speakers to refer to “parts of our [human] bodies”. (For further discussion, see Wierzbicka & Goddard 2018, in press).

5.   The set of universal semantic primes in 2020: is this it?

Seven years after the set of 65 universal semantic primes was first presented as the answer to Leibniz’s question about the “alphabet of human thoughts”, I am happy to repeat what I said in Imprisoned in English: “Extensive semantic investigations conducted over many years, by many scholars, in the NSM framework, have led to the conclusion that there are sixty five primes, the same in all languages” (p. 34).

Does this mean that the table of 65 primes is exactly the same in 2020 as it was in 2014?

No, not exactly; but it is very close now to what it was then. There are still 65 primes, and only one of them shows a new face: it is the prime mine, as in the sentence “it is mine”, with which we replaced the prime that we earlier designated, for many years, with the word “have”. In 2014, have was briefly replaced with be someone’s, but after a short time it stabilised in the “egocentric” version mine or be mine. The reasons for this replacement are discussed in detail in an article entitled “It’s mine!” co-authored by Cliff Goddard and myself and published in 2016 (Goddard & Wierzbicka 2016) and will not be discussed here.

Apart from mine, however, there are no new primes in the current table of primes, and mine itself is not an entirely new prime but an older prime
re-conceptualised. Thus, from the point of view of NSM researchers, the table with 65 elements has now been stable for many years, and as more and more domains were addressed in NSM-based work, it has proved sufficient as a toolkit for dealing with them all.

How have these ideas been received by those outside the NSM research community? Have linguists sceptical about the NSM theory, or downright hostile to it, been able to throw serious doubt on any of the 65?

A good deal of scepticism has indeed been expressed, at different times, by different authors. We have sought to consider all such critiques as carefully as possible. See, for example Goddard and Wierzbicka’s 2014b response to Nicholas Evans’ doubt about know and think and to Daniel Everett’s claims about one and two, my 2012 response to Daniel Everett’s claims about all and several others in Pirahã (2012) and Goddard’s (2008) response to George Van Driem’s (2004) claim about the absence of feel in Nepali. Not all our responses have yet been published. In particular, Lillian Brise’s doubts about feel, expressed in a careful study of the Nigerian language Igala (2017) still awaits a full answer in print.

Figure 1. Semantic primes (English version), as in Goddard & Wierzbicka 2018 (diagram designed by Lauren Sadow)

The most serious empirically-based attack on the viability of one of the primes which NSM researchers are facing at the moment concerns the prime have parts (part, parts). This is why the bulk of this paper is devoted to this particular prime. Before returning to the complex questions to do with the prime have parts, however, I will review the concept of “semantic prime” itself, acknowledging that our understanding of this concept has deepened and sharpened over the years.

This increased understanding was due largely to the ever expanding range of both languages and domains with which NSM researchers have grappled in their analytical work. In a considerable measure, it was also due to the critique to which NSM work was at times submitted by serious and open-minded outsiders, keen to pursue the truth regardless of trends and ideologies. Among such scholars I would like to single out one: Ken Hale.

6.   Ken Hale: Endorsement and critique of the NSM project

In his contribution to Semantic and Lexical Universals (1994), Hale offered strong support for the NSM project overall, while at the same time questioning one aspect of it: the “strong lexicalisation hypothesis”.

In his introductory chapter in the same volume, Cliff Goddard (1994) had formulated this hypothesis as follows: “Every semantically primitive meaning can be expressed through a distinct word, morpheme or fixed phrase in every language” (p. 13). Having said this, Goddard immediately went on to explain that what was meant was not a one-to-one correspondence between a word and a meaning:

This does not entail that there should be a single unique form for each primitive. Some languages have several forms (allolexes or allomorphs of the same item) functioning as contextual variants expressing the same primitive meaning. Conversely, it sometimes happens that the same form serves as an exponent of different primitives, although their distinct syntactic frames make it appropriate to recognise polysemy (p. 13).

Yet somehow, this explanation was often not heard, and the NSM theory was perceived as expecting every language to have a word (one single, unique word) for each prime. This was even true of Ken Hale.

In his contribution to our 1994 volume Semantic and Lexical Universals Hale wrote:

The comment which I wish to make here is not based solely on the brief and very tentative survey just given of the possible Misumalpan realisations of the lexical and semantic universals. It is based partly on several decades’ work in both practical and theoretical studies of the grammars and lexical resources of a number of Native American and Australian Aboriginal languages. My own experience and the results of many years of study on the part of Anna Wierzbicka and her colleagues, as well as the work on lexical conceptual structure by people like Ray Jackendoff and others, lead me to accept virtually without reservation the notion that there are universal fundamental concepts, or ‘conceptual primitives’ (p. 121).

So Hale accepted the notion that there are universal ‘conceptual primitives’.
But were those conceptual primitives linked with specific linguistic expressions? Hale was not convinced that it was so:

I do have reservations about one aspect of the overall program which this short study of Misumalpan attempts to represent. Specifically, I doubt that all languages ‘have words for’ the conceptual primitives. This in no way challenges the idea of conceptual primitives, since concepts do not have to have names to be real. The ‘reality’ of the concepts can be determined in other ways. And I do not deny that ‘shared words’ exist, of course, nor do I deny the importance of determining what those shared words are or the importance of having a semantic metalanguage, based on universal semantic primitives.

Thus, Hale expressed doubt that all languages “have words for” conceptual primitives, but in fact, as the “strong lexicalisation hypothesis” formulated by Goddard shows, we did not make such a claim. We always recognised the polysemy of both words and phrases which could be plausibly regarded as exponents of primes.

Mentioning, as an example, the polysemy of the English word know Hale states that “the words of a language are not isomorphic with the universal semantic primitives” and that “observations of this nature… cast doubt on the strongest requirement – that is, the isomorphism requirement – on the naming of conceptual universals in the world’s languages” (p. 28).

This statement is entirely consistent with the “strong lexicalisation hypothesis” as it has always been understood by NSM researchers. We never expected that semantic primes would be “named” in the world’s languages. What we did expect then and do expect now, is that they would have some lexical exponents in specifiable lexical and grammatical contexts.

It seems clear to me now that Hale sincerely misunderstood our hypothesis, as formulated in the relevant NSM literature, and also, that we did not always formulate this hypothesis with sufficient clarity and precision. Hale’s example of the prime part (have parts) illustrates this. The fact that at the time (1994) we did not fully understand the semantic profile of part contributed to the confusion. Since part is still the most controversial of NSM primes, Hale’s discussion of it bears closer examination.

Commenting on the Miskitu word pis derived from the English word piece, Hale remarked that it “corresponds well, if not precisely, to the proposed universal concept. This would be a miracle if part were not itself a universal, unnamed before the borrowing” (p. 283).

Hale seemed to accept our claim that speakers of all languages (including Miskitu) have the concept expressed in English with the word “part” in sentences like “the liver is a part of a person’s body”. He assumed, however, that this concept can be “unnamed” and did not seem to accept that every language has a word, or phrase, polysemous or otherwise, with which the concept can be expressed, in a distinct linguistic context.

For example, he observed that while “liver” could be defined in Miskitu as “a thing of our body”, the expression “the thing of our stomach” can be used “in reference to the intestinal worm called liwa/baabil” (p. 281).

How do we know, then, that speakers of Miskitu can distinguish, in their thoughts, between “a thing of the body” such as “liver” or “bladder” and “a thing of the body” such as some intestinal worms? Hale seemed to have no doubt that Miskitu speakers can, and do, distinguish between them, but he did not seem to accept our claim that this is done by means of a particular meaning of the Miskitu word dyara ‘thing’. In particular, he said: “The ‘reality’ of the concepts can be determined in other ways” (p. 282). In what other ways? Presumably by means of linguistic argumentation, based on the analysis of grammatical structures of the kind that he used himself in his article on “Part and whole relationships in Warlpiri” (1981), an argumentation which relies, throughout, on the English words PART and WHOLE, spelled in caps.

But first, linguistic argumentation cannot produce a semantic metalanguage in which meanings and ideas could be explained to anyone, especially across language boundaries; and second, it cannot produce a lingua franca suitable for global communication – for example, at a time of pandemic.

In the conclusion of his chapter, Hale wrote:

In summary, I think that a criterion of terminological isomorphy for universal concepts is too strong. While the proposed universality of fundamental concepts might be contradicted by empirical data at some point, it is not contradicted by the well-known fact that it is sometimes difficult or impossible to ‘find a word for’ some universal concept in a given language (p. 283).

The question is: is it difficult or is it impossible? My own conclusion, after fifty years of grappling with this question, is that while it is certainly difficult, it is NOT impossible – not even in relation to part (have parts­)—provided that by “a word for” we don’t mean “a name” but “a lexical exponent”, and that we recognise the polysemy of words like “thing” and “have”.

I do not claim that in arguing for part (have parts) as a lexical as well as a conceptual universal in the 1994 volume we made our argument sufficiently strong and sufficiently clear. First, we were still confused about the relationship between two primes: part and some, which can both be expressed in English by means of the word “part” (e.g. “part of the ceiling collapsed”, “part of the meat was burnt”). Second, we made a mistake in choosing part rather than have parts as our preferred way of referring to the prime in question. And third, as it seems to me now, one piece was missing in our argument in favour of this prime.

As discussed in my 2007 paper “Bodies and their parts”, there is a universal cognitive model of the human body which presents the body as having many parts, located in different places in the body. The universality of this model can be captured in the component: “people’s bodies are like this”.

Speaking in ordinary English, we could say that “things” like “liver” and “bladder” are inherent to the human body, whereas intestinal worms are not. In Minimal (and therefore cross-translatable) English, we could speak about it like this (I will continue with Hale’s example of liver):

People’s bodies are like this:
they have many “things”, some of them are inside the body,
liver is one of these “things”.

It seems to me that in this context, the word glossed as “things” means exactly the same as “parts” and cannot refer to worms. Thus, I agree that by itself, a sentence like “liver is a thing of the body” (or: “in the body”, or: “belonging to the body”) does not fully disambiguate the word glossed here as “thing”. It is only a combination like “the human body has many things, the thing called liver is one of these things” which fully disambiguates it.⁴

Essentially, the same applies to so-called “folk definitions” from the American Indian language Papago, published by Casagrande and Hale (1967). For example:

cimamag “horned toad”: “and those also go around which one small (…),
it has some things sort of standing on its head, they are sharp (…) (p. 170).

The native speaker who is offering this definition of a particular word appears to be describing the kind of creature called by this word. The implied starting point is: “a creature of this kind is like this: …”; and the description which follows includes the component “it has some things on its head”. In this lexicogrammatical context, the word glossed by the authors as “things” can only refer to parts of the creature’s body, not to any extraneous objects.

The same applies to Durie et al.’s observation that in the Austronesian language Achenese ‘The knife has a blade’ and ‘The knife has a sheath’ are expressed in exactly the same way (1994: 194). The examples they offer, however, are glossed: “That knife has a sheath” (knife-that-BE-sheath) and ‘That knife has a blade’ (knife-that-BE-blade) (p. 155). This shows that in order to disambiguate the construction in question a somewhat larger lexico-grammatical context is needed, for example:

sikin (‘knife’)
a thing of one kind
things of this kind are like this:
they have two “things” [one is sharp, the other is not sharp, etc.].

To return to Hale, Hale recognised and endorsed three tenets which are the keystones of NSM theory, without quite believing that they could be integrated. First, following Weinreich (1962) and Casagrande and Hale (1967), he accepted the idea that every language can be its own metalanguage. Second, he accepted, “virtually without reservation”, “the notion that there are universal fundamental concepts, or ‘conceptual primitives’” (Hale 1994: 282). Third, he accepted that “‘shared words’ exist” and he recognised “the importance of determining what those shared words are” and “the importance of having a semantic metalanguage based on universal semantic primitives” (1994: 282).

But writing about these things in 1994, he didn’t see his way to integrating these three ideas. It seems likely that the uncertainties about part, more than anything else, prevented him from recognising that the search for “universal words” and the search for universal conceptual primitives can be two sides of the same coin; and relatedly, that in principle every language can be an adequate metalanguage not only for itself, but also for every other language.

To return to the example of the trade in human body parts, such a practice can only be condemned by all people on earth if all people on earth have the concepts of people, body, and parts; and that they have translatable words or phrases in which the topic can be discussed in international fora.⁵

7.   Talking about “parts” of animals, plants and artefacts

Essentially, what applies to the human body applies also to animals, plants and artefacts: they can be seen as having either two or many “things” (parts) – “things” which are often seen as comparable to “things” in the human body (for earlier discussion see e.g. Wierzbicka 2007: 37).

For example (partial sketch explications only):

trees

things of one kind, there are many kinds of things of this kind (etc.)

things of this kind grow in the ground; they are big

a thing of this kind has many “things” (parts), one is long, grows in the ground; above it there are many others, they are like the arms in people’s bodies

mushrooms

things of one kind, there are many kinds of things of this kind (people can eat these things, etc.)

things of this kind grow in the ground, they are not big

a thing of this kind has two “things”, one is long, it grows in the ground; the other is above it, it is like the head in people’s bodies

insects

living creatures of one kind, there are many kinds of creatures of this kind; they are very small (etc.)

the body of a creature of this kind has many “things”, like a human body has many “things”

one is like the head in people’s bodies, some are like legs, one is like the big “thing” below the head, above the legs

two are like the wings in birds’ bodies

knife

a thing of one kind called “knife”, things of this kind are made by people (etc.)

people can cut many things with things of this kind

a thing of this kind is long, it has two “things”, one is sharp, the other is not sharp

chair

a thing of one kind called “chair”, things of this kind are made by people (etc.)

someone can sit on a thing of this kind

a thing of this kind has many “things”, like a human body has many “things”

some (of them) are like the legs in people’s bodies, one is like the back

It is well known that in many languages, words used to refer to human body parts are also used to refer to parts of living creatures, plants and artefacts. For example, in Warlpiri, the word which refers to the human head is also used to refer to a comparable part of a boomerang or a spear-thrower; and a word which is used to refer to the human nose is also used to refer to the front part of a car (Laughren 1984). This fact strongly supports the idea that Warlipiri speakers perceive the similarity in structure and think of the “things” so named in creatures, plants and artefacts as analogous to the “things” (parts) of a human body.

As I wrote in 1985 in my Lexicography and Conceptual Analysis, with reference to Giambattista Vico (1744) and Baudouin de Courtenay (1929), as well as to my own painstaking empirical research into the semantics of the concrete lexicon presented in that book, “things around us are conceptualised to a remarkable degree with reference to the human body” (1985: 343). As our empirical knowledge about the lexicons of the languages of the world increases, Vico’s claim that the human body is a vital conceptual model for human conceptualisation of the world gains stronger and stronger empirical support. The “partonomic” structure of the human body is an essential reference point for people everywhere on earth, as they try to make sense of the world they live in. Evidence suggests that (despite claims to the contrary, see e.g. Majid et al. 2006: 145) both body and ‘things’ (parts)
of the body are essential conceptual tools for all people on earth. (For further discussion of the universality of the concept ‘body’, see Wierzbicka 2007, Goddard 2008).

8.   Conclusion

More than a century ago the great American anthropologist and explorer of American Indian languages Franz Boas affirmed the “psychic unity of mankind” (following his teacher, German anthropologist Adolf Bastian). Afterwards, for a long time, this tenet was widely accepted in anthropology, and as a critic, Le Pan (1989: 2), lamented thirty years ago, “the most influential anthropologists of the past fifty years have all been in agreement that the peoples of the world all think in the same way”.

The last thirty years, though, saw another swing of the pendulum. Among the most influential proponents of this swing, I would single out the leading anthropologist and founder of the new discipline of cultural psychology, Richard Shweder, who contraposed “cultural pluralism” to the “principle of psychic unity”, and called the belief in the principle of “psychic” (psychological) unity “pious” (Shweder & Sullivan 1990: 400). To his credit, however, Shweder remained open-minded and later accepted the NSM claim that know, think, want and feel, and also good and bad, are universal human concepts (Shweder 2004: 82).

After fifty years of investigations, both empirical and analytical, I submit that the same applies to eleven out of the fourteen “semantic primitives” which I posited half a century ago, and to the full set of sixty five, which Cliff Goddard and I posited seven years ago (2014a, b), including have parts (part, parts).

Yes, we need to be always conscious of the danger of taking categories of our own language for universal and attributing them to speakers of other languages. In particular, in the present era of the global domination of English there is an ever-present danger of taking concepts lexicalised in English for universal. No one has sought to highlight this danger over the years more strongly and more consistently than NSM researchers, to mention only my own books English: Meaning and Culture (2006) and Imprisoned in English (2014), and Carsten Levisen’s “Biases we live by” (2019). More than that, we have consistently exposed the “pervasive Anglocentrism entrenched in the language of contemporary science” (Goddard and Wierzbicka 2014b: 155).

But this is precisely why we have always insisted on finding, roughly speaking, “universal words” (or “lexico-semantic universals”, cf. Goddard 2001), and not only theorising, in English, about conceptual primes lexically embodied in some languages but not in others. Thus, for parts – as for any other hypothetical prime – it is critical to establish whether or not it is lexically embodied in all the languages sampled.

As we have seen, according to Ken Hale, a word for a prime (for example, part) could not be borrowed from another language unless it was already there, in the speakers’ minds. But the fact that a language borrows a word for a particular prime does not necessarily mean that before this word was borrowed, the prime was “nameless”. For example, the fact that Miskitu has borrowed the English word “sort” (in Miskitu, sat) and that sat functions now (according to Hale) as the Miskitu exponent of the universal prime kind, does not mean that Miskitu did not have another exponent for that prime before the borrowing.

Of course genuinely new concepts are borrowed all the time, often via loanwords. For example, the English word “fair” has been borrowed by German because it brought with it a new concept (‘fair’) which many speakers of German had learned from English and found useful. But complex concepts like ‘fair’ can be borrowed precisely because they are complex and composed of simple concepts (primes) which are already present in the speakers’ minds. A prime, on the other hand (i.e. a concept that is simple and not decomposable into simpler concepts) cannot be borrowed because there are no other concepts out of which it could be built in the speakers’ minds.

After fifty years of working with parts (have parts) as a hypothetical semantic prime I acknowledge that it would have been better to label this prime, from the outset, as “to have parts”, rather than just “parts” and especially “part” in the singular. Possibly, many misunderstandings could then have been avoided. This is not a new thought, though, since in 1994, in the chapter “Introducing lexical primitives”, Cliff Goddard and I wrote:

Though the concept of ‘parthood’ has always been an element in the NSM inventory of fundamental concepts, there has been a change in expectations about the kind of exponent that can be expected to be found cross-linguistically. Rather than it taking a form analogous to English part of, recent research suggests an exponent with the converse orientation is more widely attested, that is, an element like have parts (p. 46).

For clarity’s sake, I will now contrast here three different positions on the universality of conceptual and lexical primes.

1. As we have seen, according to some linguists, such as Hale (1994), there is a universal set of conceptual primes that all humans share, but there is no set of “universal words or expressions” diverse in form but identical in meaning in all languages (at least, not one which could serve as an adequate semantic metalanguage for them all). In effect, then, Hale seemed to accept that there is a shared conceptual “alphabet of human thoughts” for all people on earth to think with but not to speak (‘write’) with.
2. According to some other linguists, there is no universal set of conceptual primes shared in its entirety by all languages: а language may have its own set of “semantic primes”, that is conceptual primes expressed in this language in identifiable words and meanings; and such sets of lexically embodied primes may overlap, but there is no complete “alphabet” of lexically embodied primes common to all languages (again, not one that could serve as an adequate semantic metalanguage for them all). This is, as I understand, Nicholas Evans’ position⁶.
   Thus, for Evans, part “is not a prime in languages like Warlpiri, Kayardild, or a number of others”. In his view, there may be a “solid core [of primes] which are equivalent in all languages, and then an outer set where different languages compose things differently (…) ‘Part’ would be such a case” (personal email, October 6, 2020). Or, “to stick with the alphabet metaphor, (…) the total alphabet isn’t shared and (…) there will be a common set of letters for a subset supplemented by some language-specific ones” (personal email, October 22, 2020). This means, in effect, that even if there were large overlaps between the sets of semantic primes lexically embodied in different languages, there is no shared complete “alphabet of human thoughts” for all people to both think and speak (‘write’) with. In fact, according to Evans, a language may not have sufficient lexical resources to serve entirely as its own metalanguage, and verbal explanations may need to be supplemented by ostention and by gesture⁷ (personal email, October 9 and October 22, 2020).

3. According to NSM linguists, on the other hand, there is a universal “alphabet” of conceptual primes lexically embodied in all languages, an “alphabet” that people can both think and speak (‘write’) with. Leibniz’s metaphor of alphabet is wonderfully apt here, because an alphabet is not just any set of “letters” but a complete set, sufficient for writing anything that one might want to write. Four consequences follow from this.

First, every language can be, in principle, its own metalanguage. In his 2008 overview “NSM: The state of the art” Cliff Goddard called this the “belief in the meta-semantic adequacy” of natural languages:

This fundamental conviction is the conviction that ordinary natural languages are adequate to represent their own semantics via language-internal paraphrase; that is, belief in the “meta-semantic adequacy” of natural languages. This entails the view that every language has an irreducible semantic core with a language-like structure, with a mini-lexicon of indefinable expressions (semantic primes) and associated syntax (p. 11).

Second, speakers of all languages share a particular, specifiable, set of concepts, and have a set of words or expressions with which they can express these concepts.

Third, every language can, in principle, be a metalanguage for every other language: if the sets of expressions in terms of which different languages can be described match, then any such set can, in principle, serve as an adequate metalanguage for all other languages.

Fourth, speakers of all languages can discuss some topics of common interest using words different in form but identical in meaning. For example, there can be an international discussion about a charter of global ethics, free of Anglocentrism but based on words and expressions which correspond to shared human concepts.

I find it wonderfully symbolic that the last prime whose universality has been repeatedly questioned in serious linguistic work—HAVE PARTS—can be unambiguously identified in contexts which underscore human unity, such as this:

“All people’s bodies are like this: they have many THINGS (=PARTS);
the head is one (of them); many are inside the body.”

This need for a reference to people’s bodies complements and rounds off Boas’ emphasis on the psychic (psychological) unity of humankind. We all think, know, want, and feel; we all think in terms of good and bad; and we all have bodies, with many parts. Evidence suggests that, despite all our diversity, linguistic and cultural, we people all think about the world with sixty five shared “semantic primitives”; and we all know that we have bodies, bodies with many parts (“things”). Consequently, we all have linguistic resources necessary for condemning not only genocide, torture, infanticide, and rape, but also trade in human body parts.

To have a global discussion on matters of global importance we need more than a set of shared conceptual primitives; we also need a shared semantic metalanguage in which those primes – shared human concepts – can serve as tools for human communication, potentially including all people on earth.

According to Pope Francis’ encyclical “Fratelli Tutti” (mentioned earlier), “In today’s world the sense of belonging to a single family is fading” (section 30). From this point of view, it seems particularly important to recognise that the principle of psychological unity of all people on earth is not just a pious slogan, or a well-meaning declaration not based on evidence, but a truth supported by empirical findings; and that these findings can enhance our sense of belonging to a single family and a universal community of communication.

As the article “Psychic unity of humankind” in the Encyclopedia of Anthropology (Job, 2006) says, “Ineluctably, the idea has ethical significance. For attempting to inform humans about what they are and what they have in common is not a neutral act. By contributing its share, anthropology becomes part of the world-historical process by which human unity comes to exist in a new sense in virtue of being known to exist” (online).

What applies to anthropology applies also to linguistics: if, as the same article says, “the idea [of “the psychic unity of humankind”] remains at the very heart of anthropological enterprise”, the idea of a common human “lingua mentalis” (“language of the mind”) (cf. Wierzbicka 1980) must remain an integral part of the “linguistic enterprise”.

The current one-sided emphasis on diversity without  acknowledgment of the fundamental unity of all languages undermines the truth about the unity of the human mind and  of the “human race”.

The emphasis that many influential linguists place today on linguistic diversity is such that the underlying conceptual unity of all languages tends not to be mentioned at all. When it is mentioned (which is very rare) it is mentioned only in general terms, without any concrete examples. Typically, both in scholarly linguistic works and in the publications for the general reader, numerous examples of astounding diversity are offered, without a single example of something that all languages share. Not even the universality of the concepts “you” and “I”, “people” and “body”, and “good” and “bad”, long argued for in the “NSM” literature, is mentioned or acknowledged.

The message implicitly (if not explicitly) conveyed is that the unity of the human mind is only a pious slogan. There are no shared human concepts, there can be no “universal human community of communication”. The thing to do is to celebrate the diversity of languages, and not to seek what we humans share.

By contrast, the NSM approach, which was initiated by the publication of Semantic Primitives fifty years ago, has always seen the diversity of human languages as combined with, and undergirded by, a shared conceptual core, and has sought to determine what that shared core was, regarding this search as a task of utmost importance.

As cross-linguistic investigations of the last fifty years show, despite the phenomenal diversity of human languages and cultures, a shared “alphabet of human thoughts” was not just a figment of Leibniz’ imagination. In fact, we can now affirm with confidence that there is not only a shared “alphabet of human thoughts” but a common language, Basic Human, with a specifiable vocabulary and grammar, which can be seen as humanity’s “shared mother tongue”⁸.

I believe that this common language, Basic Human, represents the deep truth about the “genetic code of the human mind” (cf. Wierzbicka 2010; Goddard, Wierzbicka and Fabrega 2014); and that for this very reason, it can provide a secure basis for a non-Anglocentric global discourse about questions that concern us all, such as ethics, the earth and its future, and the health and well-being of all people on earth.

^{1 I first heard Bogusławski’s ideas on the subject in a talk he gave at Warsaw University in 1964 (“O założeniach semantyki”). Before that, he had presented them in 1963, in a paper submitted to the journal Voprosy jazykoznanija, but not accepted. (Eventually the paper was published in Lingua Posnaniensis XLV, 7–18, in 2003).}

^{2 To this day, there is a strong synergy between the NSM approach, anchored in universal semantic primitives and the Moscow School of Semantics (see e.g. Apresjan 2000, chapter 8; Mel’čuk and Milićević 2020, chapter 3).}

^{3 Levisen (2019: 4) characterised “conceptual colonialism” and “conceptual Anglocentrism” as follows:
When speakers of “languages other than English” are reported to live without some of the important emotions in the world of Anglo English, for example “sadness” (Levy 1973), “depression” (Obeyesekere 1985) or “happiness” (Wierzbicka 2004), the standard response in Anglophone scholarship seems to be: “maybe they don’t have the word, but surely they have the concept.” This dogma seems so strong in current thinking, that apparently no empirical evidence is needed to support the claim. In my view, this is where ethnocentric bias can turn into conceptual colonialism. The problem is the unidirectional nature of the claim: the concepts “we” live by, must somehow be present in other people’s discourse, but the argument is never made the other way around. No one, for instance, would argue that English speakers live by the Bislama concepts of kros, les, and sem, or that the interjections awo! or dipskin! are tacitly present in English speakers, when they clearly have no words for exactly these concepts (…).
Conceptual Anglocentrism is the imposition of Anglo semantic concepts on non-Anglo conceptual words and worlds, to which these Anglo concept might be foreign and meaningless, and which, when used to conceptualize these words and worlds, inevitably lead to distortion, and pseudo-precision. Conceptual Anglocentrism is a near-synonym of interpretative Anglocentrism. The worst form of interpretative or conceptual Anglocentrism is, de facto, conceptual colonialism.}

^{4 Accordingly, the terms for specific body parts could be explicated along the following lines (A and B):}

1. ^{head (someone’s head; with the word “part”)
   people’s bodies have many parts, this is one of them
   when people think about their bodies, they can think about this part like this:
        it is round [m], it is big, it is above everything else
        I can move it when I want
   because people's bodies have this part, people can think}

^{head (someone’s head; with the word “thing”)
people’s bodies have many “things”, this is one of them
when people think about their bodies, they can think about this “thing” like this:
     it is round, it is big, it is above everything else
     I can move it when I want
because people’s bodies have this “thing”, people can think}

^{5 In a careful semantic study of the Algonqian language East Cree, Marie-Odile Junker (2008) showed that in that language people generally don’t talk about the “part-whole” relationship in the way speakers of English do, and also, that there is no word that would be used in the same way as the word “part” is used in English and no phrase matching the English phrase “part of the body”. Further, Junker suggested that “part-whole relationships are conceptualised in East Cree… from an opposite direction from that operative in English. From a Cree perspective we should speak of part-whole relationships starting from the view that something is first perceived as a whole, and then divided or broken into (specific) parts” (p. 187). Junker concluded that “the status of this prime [PART] must be… reconsidered” (2008: 189).
In a sense, the choice of HAVE PARTS rather than PART as the main exponent of the prime in question does represent a reversal of the direction from which the prime in question is considered: we start from the whole (the body) and recognise that “it has many parts”, or, as one would say in some languages, “it has many things”, “it has many (things)”, “it is many (things)” or “there are many (of it)”.
East Cree appears to rely on this last strategy. For example, according to Marie-Odile Junker, to say “the knife has two parts, one is sharp, the other is not sharp” one would say something like this: “the knife, there are two (of it); one is sharp, the other is not sharp” (personal email, 30 September 2020).
When I recently asked Marie-Odile how people could speak in East Cree about the issue of the trade in parts of human bodies, she reported (personal email of October 20, 2020) that three of her French-speaking Cree consultants said that they would need to be more specific and offered the following response (which she conveyed to me in English):
About the body, some people sell kidneys, eyes, etc. This is very bad.}

^{The Latin tag “et cetera” means, of course, “and others” or “and other things like this”. So this response from French-speaking Cree consultants appears to suggest that in order to speak in East Cree about “parts of people’s bodies” in general, the speaker may need a supporting lexicogrammatical context including the words “body”, “things” and “other”, plus a mention of some exemplars which themselves are conceived of as “things in the body”.
One other piece of relevant information. When asked about St Paul’s sentence in Ephesians (Eph 5:30) rendered in the King James Version as “We are the members of his [Christ’s] body”, the SIL Bible translator Bill Jancewicz replied that in one of the earlier translations, “Legacy (1862) Western Cree”, the word “members” (in Greek, mele) has been rendered with the word paskessiwiniw, and he offered the following gloss:
“For all of us are parts [paskessiwiniw] of his body (paskessiwiniw = limbs)”.
(Personal email from M.O. Junker, 30 October 2020). The matter requires further investigation.}

^{6 In an email of December 22, 2020 he clarifies his position further as follows: “I would say that IF a set of primitives can be found in a given language, it need not be entirely the same (though it would be likely to overlap) with the set in another. [...] I also would claim that gestures cannot be discounted in the total expressive setting.”}

^{7 For example, in his article in the Oxford Handbook of Linguistic Typology Evans (2010a: 515) asserts that in the Australian language Dalabon the concepts KNOW and THINK, regarded in the NSM research as universal semantic primes, lack specific exponents, and that a single word, bengkan, “covers both ‘know’ and ‘think’”. He also affirms that Dalabon has a Dalabon-specific semantic prime, which he represents as √beng. Both these claims (made also in Evans’ book Dying Words (2010b: 59)) have been examined in detail and, I believe, refuted in Cliff Goddard’s and mine joint paper “Semantic fieldwork and lexical universals” (Goddard and Wierzbicka 2014a).}

^{8 The expression “humanity’s shared mother tongue” is of course a metaphor: nobody speaks Basic Human on a daily basis, and it is not anyone’s first acquired (“native”) language in a literal sense. Yet it is interesting to note how much Basic Human can be actually heard in young children’s speech in many languages, as the literature on child language reflects (see e.g. Braine 1976; Bloom 1991; Bowerman and Levinson eds. 2001; Tien 2010; Slobin 2017). In the English version of Basic Human, examples could include “No!” (“I don’t want”), “More!”, “Peek-a-boo!” (cf. “You can’t see me”, “I see you!”), “I can’t”, “It’s mine!”, “Show me!” (“I want to see”), “I don’t know”, “What’s that?” and so on. Early utterances include of course many concrete nouns, such as “ball” or “duck”, but these could be interpreted along the lines of: “something of one kind, called ‘ball’ (‘duck’).” For discussion, see Goddard 2011: 385-391). And of course even before any words are uttered, babies’ faces and hands can express very clear messages in Basic Human, such as “I feel something good”, “I feel something bad”, “I see you!”, “I want this!”, “I don’t want this!”, and so on.}

About the authors

Anna Wierzbicka

Author for correspondence.
Email: anna.wierzbicka@anu.edu.au
ORCID iD: 0000-0002-6074-7865

Professor of Linguistics (Emerita) in the School of Literature, Languages, and Linguistics at Australian National University. Her work spans a number of disciplines, including anthropology, psychology, cognitive science, philosophy and religious studies as well as linguistics, and has been published in many journals across all these disciplines. She has published over twenty books. Her latest book, “What Christians Believe: The story of God and people in Minimal English”, was published by Oxford University Press, New York in 2019. Professor Wierzbicka is a Fellow of the Australian Academy of the Humanities, the Australian Academy of Social Sciences, the Russian Academy of Sciences and the Polish Academy of Arts and Sciences. She is the winner of the International Dobrushin Prize for 2010 and of the Polish Science Foundation’s 2010 prize for the humanities and social sciences.

References