The Natural Semantic Metalanguage approach: an interview with prof. Bert Peeters
We conducted an interesting interview about Natural Semantic Metalanguage with one of its main promoters, Bert Peeters, associate professor at the Australian National University. He’s also an adjunct associate professor at Griffith University, Brisbane. His main research interests are in French linguistics, intercultural communication, and language and cultural values. Publications includes Diachronie, phonologie, et linguistique fonctionnelle (1992), Les primitifs sémantiques (ed., 1993), The lexicon-encyclopedia interface (ed., 2000), Semantic primes and universal grammar (ed., 2006), Tu ou vous: l’embarras du choix (ed. with N. Ramière, 2009), Cross-culturally speaking, speaking cross-culturally (ed. with K. Mullan and C. Béal, 2013), and Language and cultural values: adventures in applied ethnolinguistics (ed., 2015).
As a linguist, you have used “Natural Semantic Metalanguage” in several publications. But what, exactly, is “Natural Semantic Metalanguage”?
Natural Semantic Metalanguage – or NSM, as it is called by those who use it or have at least heard of it – is a tool or an instrument made out of language and used to talk about language, which is why it is called a meta-language. Whenever we need to talk about language (linguaggio) or the properties of language, or as soon as we want to say something about one or more languages (una o più lingue), on the other, we have no choice but to do so by means of… a language (una lingua). The latter is then referred to as a metalanguage (una metalingua – not un metalinguaggio, I’ll come back to that in a moment), and the former could be called the object language, i.e., the object of the metalanguage.
Natural Semantic Metalanguage is used for the purpose of ‘doing semantics’, as understood by linguists, i.e. for the purpose of describing meaning: meaning of words, meaning of word segments, meaning of phrases, meaning of entire sentences. It is also used for the purpose of describing ways of thinking that are culturally specific, i.e. so-called cultural scripts. I won’t go into that here, but I’ll limit myself to an example of word meaning. Since I am talking to medical specialists, let’s consider the word trauma. It has more than one meaning, but the one I want to focus on is defined by the Merriam-Webster dictionary as ‘a disordered psychic or behavioral state resulting from severe mental or emotional stress or physical injury’. This is not how NSM goes about it. The NSM definition (or explication, as we call it, since it is meant to make things explicit) goes as follows:
That’s quite a mouthful…
Yes, it is – and for good reasons. Have another look at the Merriam-Webster definition. It is short, but is it clear? For medical specialists, it probably is. But come to think of it, what this dictionary has done is replace one complex word – trauma – by a string of others (disordered, psychic, behavioural, state, resulting, severe, mental, emotional, stress, physical, injury) linked together by three ‘little words’ (a, or, and from). The complex words are not necessarily any clearer than the word that is being defined. The NSM explication is very different: it is long, and provided you read through it carefully, it is easy to grasp. Reading it, you get a real feel for what a trauma is and what it feels like, something you can’t say about the definition provided by Merriam-Webster. This brings me to the adjective natural in ‘Natural Semantic Metalanguage’: NSM explications, when done properly, are rather good at what they are meant to do, because they are couched in natural, or naturally occurring, language, as opposed to artificial, or technical, or esoteric language.
Unlike many other metalanguages, Natural Semantic Metalanguage is also natural in a different way. It is as natural as its object language. It shuns all kinds of formalism and relies entirely on existing vocabulary (other metalanguages often use symbols, abbreviations or newly developed concepts, or use existing vocabulary in idiosyncratic ways) as well as on existing grammar (other metalanguages often have no grammar whatsoever, or use logic instead; in Italian they would presumably be referred to as metalinguaggio rather than metalingua, as from a linguistic point of view they are very much a matter of vocabulary only – no naturally occurring grammar). Its vocabulary and grammar are rigorously defined, on the basis of empirical research on dozens of typologically and genetically unrelated languages that stretches back decades. The only vocabulary and grammar used is vocabulary and grammar that, until proof to the contrary, is truly universal, i.e. expressible in all the languages of the world – except in very particular circumstances on which I will say a little more later on. This means that NSM has as many guises as there are languages in the world, and all these guises are fully translatable into each other, without semantic distortion. What can be said in one version can also be said in any of the others, and end-users understand exactly the same thing, irrespective of the actual language that is being used. To put it in more technical terms, all external guises of the NSM are meant to be fully isomorphic.
For practical reasons, the language most often used for formulations phrased in NSM is English, but several strictly isomorphic versions have been developed relying on the resources of other languages, including those that are being used by practitioners of narrative medicine. The words and phrases used in NSM are known as primes. On current count, on the basis of the empirical research I referred to, the number of primes stands at 65: 65 words or phrases that are translatable in all the languages of the world, and that are semantically simple – which is why they are called primes. These 65 words or phrases cannot be defined any further, unless we allow the definition to be more technical than the word or phrase we are trying to define, which goes against the very spirit of a definition. The primes are the bedrock that underpins everything else: all semantically complex words and phrases can be defined – and ought to be defined – in terms of semantic primes and their universal combinations (i.e., their grammar). The only concession that is made is the occasional reliance, not on primes, but on so-called semantic molecules, which are fully definable in terms of other molecules and/or primes. This concession is made to prevent explications from becoming unwieldy and uninterpretable. But I am concerned I am becoming uninterpretable myself. Am I?
Not at all. What would be an example of an explication containing molecules?
Let’s stay in the medical world and look at the word temperature – as used in body temperature. The explication that follows is based on that of the word temperature as used in sentences such as What’s the temperature? and The temperature was dropping fast.
There are two molecules in this explication, identified by means of a lower case m added between square brackets. One is the word hot, the other one is the word number. These words are not semantically simple, they can be explicated further, something I will not attempt to do here. Hot and number are used in the explication for (body) temperature as it would probably be impossible to work the explications for them into the explication for (body) temperature without rendering the latter unintelligible. Molecules are not created or posited haphazardly…
This is exactly what I was going to ask next. What is the process for discovering the primes and the molecules?
Let me show you the current table of primes first. This table is taken from a 2017 paper published by Anna Wierzbicka.
In Table 1, tildes (~) are used to link together exponents that belong to one and the same prime. Such exponents are known as allolexes. Whether, say, much or many is used depends on the number (singular or plural) of the word that follows. Not all allolexes are explicitly identified in the table: i, for instance, has an allolex me. Many of the words and phrases in the table appear to be polysemous, i.e., have more than one meaning. This is certainly the case if the words and phrases are considered to be genuine English words and phrases. But they are not: what you see in the table are the English exponents (or lexicalisations) of the semantic primes. The primes themselves have only one, indivisible, meaning, and are used in NSM explications in only that one meaning. We have ways of clarifying which meanings are ‘primitive’, i.e., belong to the primes, and which are not – but I don’t think I need to illustrate these here.
You wanted to know more about the process for discovering the primes and the molecules. For the primes, that process has been ongoing for several decades. It is not a scientific process, if by ‘scientific process’ you mean a process that is outlined from the start and then rigorously followed until the end. Instead, NSM scholars have relied – there really is no other way – on trial and error. We have always tried to find the simplest concepts, which we have then tested against other languages, to see if these concepts exist there as well, i.e., are lexicalised in those languages as they are in the languages we have already looked at. Over the years, we have abandoned some presumed primes, because we found they could after all be defined in simpler terms; but above all, we have added many more (we started off with about a dozen or so…), which we discovered were needed to define – or explicate – more complex concepts. And we have always cross-checked against other languages to see whether the newly added primes were lexicalised there as well. To our great relief, we have mostly found that they were, and we haven’t been obliged too often to give up on a prime we thought we needed, but that turned out not to be lexicalised in language X or language Y.
For semantic molecules, the process is different. Here, we do have some criteria we can use, such as usefulness (we don’t want to posit a molecule unless we will really need to use it on a number of occasions), frequency (we don’t want to posit a molecule that hardly occurs in the language), and cultural salience (we don’t want to posit a molecule language users cannot relate to). The latter criterion points to a crucial difference between primes and molecules: the latter are not necessarily universal, but – as I mentioned before – they must lend themselves to explication in terms that are ultimately universal.
As a professor of linguistics, how do you envision the use of the Natural Semantic Metalanguage in health care?
I don’t expect NSM as such to be used in health care. It is too limited for that purpose. For practical applications, it needs to be tweaked somewhat. The important thing is that this tweaking needs to be done responsibly, and not in some sort of arbitrary way. It is with the idea of responsible tweaking for the purpose of practical application in mind that NSM scholars, in recent times, have come up with the idea of Minimal English, which is a by-product of the natural semantic metalanguage. It is largely derived from it, but it also allows for some concessions to be made, all in the interest of usability in the real world. Different practical applications may require different concessions, which means Minimal English is not cast in iron. Moreover, just like NSM has many guises, there is scope not only for a Minimal English, but also for a Minimal Italian, a Minimal Spanish, a Minimal Chinese, etc. The various Minimal Languages are not as rigorously translatable among themselves as the various isomorphic guises of NSM are, but the idea is to preserve as much translatability as is achievable. We tweak NSM into Minimal X or Minimal Y, we don’t disfigure it. I believe you interviewed Cliff Goddard a little while ago; as the editor of a forthcoming volume on Minimal English, he is closer to the action than I am.
Let’s not forget, though, that you are the first author of a paper,  to be published in exactly that volume, that summarises the encounter between narrative medicine and Minimal English…
Am I? Gee, I did forget. Just kidding… You are quite right. I recently worked with the Fondazione ISTUD towards developing a ‘universal’ illness plot. The idea was to guide patients to talk about their illness (rather than their disease) in a semi-structured way, so that they would all focus on the same sorts of things: how they felt before the illness, what happened next, what they did, what they were told, what they felt, what they knew, how it changed their lives, what they are doing/feeling now, what their hopes are for the future, what they feel when telling or writing their story. The patients, of course, wouldn’t use NSM or Minimal English (or Minimal Italian, etc.), but they would all be completing the same prompts, irrespective of the language in which the prompts were formulated. This allows for researchers to more readily compare the feedback they receive from patients. Also, simple prompts are easier to complete then those formulated in medical jargon. And it is in the elaboration of the plot that we used Minimal English rather than NSM. We allowed some words that are absolutely essential in conversations about health, such as doctor/nurse/specialist – but we still avoided words such as hospital or medical practice. We also took some liberties with NSM grammar.
What are your personal expectations about this work?
Well, I think that’s a question I have already answered: the use of a ‘universal’ illness plot, which can be translated in many other languages with only the slightest distortion (for no distortion at all, we would need to use NSM, and that would not be easy), allows for comparison of patient feedback across languages, and it also facilitates doctor-patient interactions, since the prompts are clearer and easier to understand if they are formulated in a simple, close-to-universal, fashion. The use of such a plot will inevitably be beneficial in terms of the quality of the care provided. Of course, we all know that the evidence-based approach in health care is irreplaceable; but it is not the only approach and it should not be used exclusively. In health care like anywhere else, there is room for a plurality of approaches, all of which will contribute to a good quality outcome.
This is quite an “odd” question to finish. Do you think that decomposing words using NSM might have some therapeutic effect, since the words it uses are simpler and go back to essentials?
You clearly left the trickiest question until last… What I can say, as an NSM practitioner who also has other interests in the area of the language sciences, is that, for me, formulating an explication in NSM often remains a daunting and intimidating task. I guess that is something I don’t share with the colleagues who have devoted their entire careers to NSM research. To be perfectly honest, although Goddard has referred to me in writing as an “old hand” (meaning, I presume, I know the tricks of the trade), I often put explicating off for as long as I can, but then, when I finally give it a go, I get hooked and want to keep trying and trying… and trying, until I think I have got it right. And the intellectual pleasure derived from being able to submit my explications to other NSM scholars, who will either enthusiastically endorse them, or pull them to pieces, or anything in between, is immense. When an explication is criticised, not to say demolished, by my peers, it is of course always a bitter-sweet moment: bitter, because my efforts seem to have gone nowhere, sweet because I know some progress has been made regardless and further improvements are just around the corner. In that sense, yes, you might say “doing NSM” has a therapeutic effect. I am constantly aware of the fact that existing definitions (formulated without any reference to NSM) are often unsatisfactory, for all sorts of reasons, and the fact that I can try to do better and that I have a tool that allows me to do it is a powerful incentive to keep going.
- Cliff Goddard & Anna Wierzbicka, Words and meanings: lexical semantics across domains, languages, and cultures. Oxford, Oxford University Press, 2014, p. 215.
- Cliff Goddard & Anna Wierzbicka, Words and meanings: lexical semantics across domains, languages, and cultures. Oxford, Oxford University Press, 2014, p. 227.
- Cliff Goddard & Anna Wierzbicka, NSM analyses of the semantics of physical qualities: sweet, hot, hard, heavy, rough, sharp in cross-linguistic perspective, Studies in Language 31(4), 2007, pp. 765-800; Cliff Goddard, The conceptual semantics of numbers and counting: an NSM analysis, Functions of Language 16(2), 2009, pp. 193-224.
- Anna Wierzbicka, Terms of address in European languages: A study in cross-linguistic semantics and pragmatics. In Keith Allan, Alessandro Capone & Istvan Kecskes (Eds.), Pragmemes and theories of language use, Berlin, Springer, 2017, pp. 209-238.
- Cliff Goddard (ed.), Minimal English for a global world: Improved communication using fewer words. London: Palgrave Macmillan.
- Bert Peeters & Maria Giulia Marini, Narrative medicine across languages and cultures: using Minimal English for increased comparability of patients’ narratives. In Cliff Goddard (ed.), Minimal English: Improved communication using fewer words (note 5).