Émile Benveniste and Linguistic Necessity
In 1966, the Societies of French Language Philosophy invited linguist Émile Benveniste to speak at their annual conference. He presented his take on “form and sense in language,” underlining that there was “no common point of view” on the subject in his field. One characteristic he raised, and not for the first time, was that of necessity. This essay explores Benveniste’s understanding of necessity, relating it to ideas Saul Kripke developed during the same period that ended up in his language-oriented Naming and Necessity.
A short description of what Kripke was trying to do will start us off. He argued that all proper and some common (e.g. natural kinds) names were “rigid designators” for the linguistic community in which the naming existed. They designated, that is to say, the same entity across all possible worlds. As such, naming set a necessary relation.
This was limited to “denotation” and did not cover “connotation.” The entity was merely named and not described; there were no characteristics or cluster of characteristics entailed. For the linguistic community to which I, and Kripke, belong, “Napoleon” denotes, without specific contextual cues such as brandy or France’s Second Empire, one entity commonly understood by the group. As a member, I can imagine a possible world where Napoleon won at Waterloo or was in fact a Russian peasant girl. In both scenarios, a single named entity exists despite its characteristics more or less radically changing. It is still Napoleon being referred to, even as a peasant girl.
Naming can be based on description, as in some cases of what Kripke called “initial baptism.” In that case, naming “fixes the reference by some contingent marks of the object.” The referent/reference pairing was not affected by subsequent changes in those marks, inaccuracies, and so on. Even limited to this world, my understanding of Napoleon has little to do with the initial naming and is likely vague, if not outright mistaken, in many ways. Yet, within the linguistic community, I can still meaningfully use “Napoleon.”
This year’s bicentenary of Napoleon’s death underlines the point. In France, the event has sparked a discussion about who Napoleon was and the manner in which he influenced French society. After two hundred years, there is no consensus as to his salient characteristics and contradictory portraits are painted across the political spectrum. Yet the name, and thus necessarily the object denoted, at the centre of the debate is shared by all.
Kripke’s argument was far more developed, including identity relations, ontological versus epistemological necessity and other nuances. Still, the above gives us a sense of the level on which he was working, that of a language-object relation.
Benveniste began at a more basic level by setting what he saw as the nature of language:
“Before everything, language signifies, such is its primordial character, its original vocation that transcends and explains all the functions it assumes in the human environment.”
With that in mind:
“Sense is the notion implied by the term of language itself, as a collection of communication processes identically understood by a speaker group; and form is, from the linguistic point of view (to distinguish it from the point of view of logicians), either the material of linguistic elements when sense is separated out or the formal arrangement of these elements at a certain linguistic level.”
It appears Benveniste viewed language, following Paul Riceour’s interpretation of the talk, as a closed formal system. Sense and other functions in the human environment were accessory—implied but not entailed. That is not strictly true. Instead, the idea was to establish that “very much before serving to communicate, language serves to live.”
Language is part of being human. Jumping straight to communication leads to misunderstandings, such as that people created the tool of language to better interact and that there is such a thing as primitive languages. Leaving out xenolinguistics, the most fundamental necessity associated with language is its coincidence with humanity.
Linguists nonetheless study languages as formal systems:
“In semiology [semiotics], that which the sign signifies does not have to be defined. For a sign to exist, it is necessary and sufficient that it is received [by speakers of the language] and that it is linked in one way or another to other signs. Does the entity considered signify? The response is yes or no. If it is yes, it is recorded; if no, no.”
Back in a 1939 article, “Nature of the Linguistic Sign,” Benveniste pointed out an ambiguity in Saussure’s ideas stemming from the fact “Saussure taught that the nature of the sign is arbitrary.” Insofar as there is no reason, possibly excepting onomatopoeias, why a certain utterance signifies a particular concept, the sign is indeed arbitrary. There are however two ways, so long as poetic language is set aside, in which it is necessary:
- Grammatical necessity: the utterance must be made up of the finite basic building blocks (e.g. phonemes, morphemes) of the language and reasonably relate to other signs in the system.
- Analytic necessity: a sign is a signifier/signified pairing. Even if the details of the signified concept may not matter for linguistic study, the utterance must signify something. Both the utterance and that something rely on this necessity: the something cannot be grasped as a discrete entity, and thus does not exist as that entity, without the signifier.
In the 1939 article, the second point was the main focus. In the 1966 talk, both made an appearance.
These sorts of necessities bring us closer to Kripke’s level, though are still more basic. While Kripke was cognizant of the particularities of naming and expressing his argument in ordinary language, he glossed over the details with the expression “linguistic community.” The initial baptism of an object was generally “fixing the reference by some contingent marks” or by literally pointing it out. The process described assumed the referent would be a name compatible with the community. Grammatical necessity spells out what would be required for a new name to fit into the preexisting shared language.
Equally, while there is some overlap between the referent/reference and signifier/signified pairings, they are not the same thing. Magritte played with the difference with his famous “ceci n’est pas une pipe.” His painting of a pipe, insofar as it was a sign, signified the concept of a pipe. It may have referred to an actual pipe or a pipe as a kind, but that goes beyond its semiotic role. And, of course, it was not itself a pipe.
The difference is complicated by Kripke’s inclusion of unicorns, along with gold, water and Napoleon, in his examples of rigidly designated objects. A sign signifies the concept Napoleon, an idea common to the group, which refers to the designated object. The etymological link between “designate” and “sign” here is unfortunate. There is, in any case, no pertinent difference between concept and object when it comes to unicorns.
Benveniste’s point was that, at base, language signifies. Describing a language, then, is to determine what utterances have meaning, or sense, for speakers. The determination is an empirical process, part of linguistics as a science. Positing that language signifies is however not entirely empirical. This is why Benveniste had to exclude poetic language, a sort of language that did not follow the rules set out, and why the term “analytic necessity” is appropriate.
As Kripke addressed imaginary objects and not imaginative language, it is reasonable to suggest the titular naming would conform to the grammatical and semiological framework of the linguistic community. Necessity, then, arises not only in relation to the denoted object, but also within the language itself.
A small detour into structuralism and post-structuralism seems appropriate here. Benveniste noted in his 1962 “‘Structure’ in Linguistics” that Saussure never wrote about structure or structuralism, only about a given language as a system. During Saussure’s lifetime, linguistics was almost exclusively diachronic, meaning it isolated words and expressions from the context of their time and addressed them only in regards to their etymological history. In the wake of Saussure, because his ideas only came into vogue after his passing, the discipline started seriously studying components of a language as framed by the whole language at a given time.
The use of “structuralism” within and outside linguistics dismayed Benveniste, since it was so broad and ill-defined as to be dismissed for being too broad and ill-defined. “In French-language European linguistics […] the fundamental principle is that [each] language is a system, of which all the parts are united by a relation of solidarity and dependence.” This can be contrasted with Chomsky’s Cartesian approach to language, in which the individual plays a more central and independent role in its development.
Benveniste did not help his cause in defining sense “as a collection of communication processes identically understood by a speaker group.” It is doubtful understanding is strictly “identical” and what can be said about components of a language is exhausted by looking at them synchronically.
At the same time, Kripke’s names, or referents, are not arbitrary or contingent on an intra-linguistic level. It is perhaps not the most important level, philosophically speaking, and it is understandable why he glossed over it. We would nonetheless run into a problem if the object denoted by “Napoleon” was named “shoe,” “Robespierre” or a series of sounds the community did not differentiate. While it is important to move beyond language to fully understand language, it is unhelpful to dismiss how signs in a given language, along with the basic building blocks of those signs, interact.
The second part of Benveniste’s talk introduced another aspect of language:
“That language signifies, that means signification is not something given to it in addition, or in a larger way than another activity; it is its very being; if it was not that, it would be nothing. But it also has a completely different character, though one equally necessary and present in all real language, though subordinated, I insist on this, to the first: that of being realised by vocal means, to practically consist of a group of sounds emitted and perceived, that is organised in words endowed with sense. It is this double aspect, inherent to language, that is distinctive.”
With some reservation, Benveniste named this second aspect “semantics”:
“Semiotics is characterised as a property of language, semantics results from an activity of the speaker who puts the language into action. The semiotic sign exists in itself, is the basis of the reality of language, but it does not entail particular applications; the sentence, the semantic expression, is only particular. With the sign, one reaches the intrinsic reality of language; with the sentence, one is linked to things beyond language; and while the sign has the signified as a constituent part that is inherent to it, the sense of the sentence involves reference to the circumstances of the discourse, and the attitude of the speaker.”
Here we seem to have jumped over the level of Kripke’s argument, though in a way that makes that level difficult. Kripke painted a picture of the relation between a noun and its referenced object, or kind of object, in the basic context of ordinary use within the relevant linguistic community. When he uttered “Napoleon” without supporting sentences in his 1970 lectures to an anglophone audience at Princeton University, the noun referred to one object for everyone.
If we need a sentence in order to link “to things beyond language,” it appears that “Napoleon” merely signified and did not reference an object. It is unclear as to whether the lectures themselves constituted enough of a circumstance to allow for a default reference. They were an instance of putting “language into action.” Further circumstantial cues, such as brandy or the Second Empire, may then have only been necessary to move beyond the default.
Benveniste’s awareness of the difficulty was evident in how he described “sentence”: “After all, it is in that way that we communicate, by sentences, even truncated, embryonic, incomplete, but always by sentences.” This was a conscious break from, though subordinated to, the formal semiotic approach and Saussure’s framework. Its aim was to get at “that which one can call the intended, that which the speaker means, the linguistic actualisation of their thought.” The minimal semantic unit “and necessary unit of the coding of thought” was the word.
The focus is therefore on use in “ordinary use within the relevant linguistic community.” Kripke both described and demonstrated such use in his lectures, though in a truncated manner that did not move beyond the minimal unit. Had he abstracted the name from use, as a noun with necessary meaning in the given language, then he would have only been talking about a signifier/signified pairing.
In the discussion after Benveniste’s talk, cases of ambiguous or missing intent such as sentences in grammars and quotations were raised. In the same way we can set aside poetic language, such marginal cases are important to note but need not be addressed here.
What does need to be addressed is Benveniste’s account of the referent/reference pairing. Kripke’s “reference” is more akin to “sense” in the following:
“All the while understanding the individual sense of words, one can very well, outside the circumstances, not understand the sense that results from the assembly of words; a common experience that shows the notion of reference is essential.
If the “sense” of a sentence is the idea that it expresses, the “reference” of the sentence is the state of things that bring it about, the situation of discourse or the fact to which it is connected and that we can never predict or guess. In most cases, the situation is a unique condition, the understanding of which nothing can replace. The sentence is therefore each time a different event; it only exists in the instant when it is pronounced and disappears instantly; it is a vanishing event.”
This sets up the necessity of the “here and now” as well as that of “sounds emitted and perceived.” Benveniste argued that the contingencies of when and where an utterance occured, who spoke and who was the interlocutor were all semantically essential. The coincidence of humanity and language, only this time with the focus on actual instances of both, had once more come to the fore. Language was part of the human experience and ordinary speech acts were “a subtle mix of liberty in the utterance of the idea, and the constraint in the form of that utterance, that is the condition of the realisation of language.”
Kripke’s necessity is not particularly sensitive to the expanded understanding of reference. As naming only denotes and says nothing about the characteristics of the object, integrating the word in a multi-word sentence does not make the designation less rigid. That is why the Napoleon of historical record and the Napoleon of “Imagine the possible world where Napoleon was a Russian peasant girl” refer to the same object despite the radical change of sense.
The “unique condition” that formed Benveniste’s reference takes the word beyond its default reference in ordinary use. The utterance of “Napoleon” in a Princeton lecture circa 1970 on the modernisation of Paris during the Second Empire would have likely been referring to Napoleon III and not Napoleon I. A poem including “Napoleon” may have only used the word for its oral or graphic form. Examples of the varied uses of the word abound, yet the common understanding of the Kripke-style reference remains.
Benveniste only lightly sketched out semantic necessity in the talk, so we will turn to his 1958 article “Of Subjectivity and Language” to flesh it out. In the article, he aimed to respond to the question, “What about language makes it apt for communication?” Effectively, it was because “it is only in and by language that man constitutes himself as subject; because only language creates in reality, in his reality that is that of being, the concept of ‘ego.’”
Language was coincident with being human. Through the ordinary use of language, the speaker set themself as the subject who spoke and the “I” of the utterance. “From this fact, I fixes another person, someone who, completely exterior to ‘me,’ becomes my echo to whom I say you and who says you to me.” Just as individual human experience required language, “language is only possible because each speaker fixes themself as subject.”
While Benveniste was talking about language as an overarching concept, he used the example of the ubiquity of the personal pronouns “I” and “you” in actual languages to argue the point. Though the pronouns were omitted in some societies due to “conventions of politeness“ and the like, “a language without expression of the person does not happen.” He then went on to show how terms of “here and now” were dependent on this intersubjective point of reference.
Semiotically, this meant a subset of signs the system could not do without were meaningful while being conceptually “empty.” “I” only finds its signifier and, in Kripke’s parlance, its reference when someone utters a sentence. Perhaps, despite their differences, this aspect of language could strengthen the argument for meaningful denotation free of characteristics. A challenge for another time.
Benveniste was far from comprehensive in his 1966 talk. Linguistic analysis has many more layers and a portion of what he called semantics could have been split off into pragmatics. He limited himself to giving a sense of the “subtle mix” involved in the individual expression of ideas in a common system; the use of a natural part of being human as an instrument of communication both with oneself and others. This essay limited the scope further to focus on necessity, which allowed a comparison to an argument expressed in ordinary language by a logician at about the same time.
The “subtle mix” could be rephrased as a tension between necessities. In Kripke’s argument, ordinary language use in the relevant linguistic community glossed over many of them. It was reasonable, given his aim to show how denoting set a rigid and meaningful referent/reference pair without characteristics in merely three lectures. Benveniste’s perspective reminds us of the broader linguistic framing; of how the argument was sandwiched uneasily between contextless signifier/signified pairings isolated from the object and a more robust notion of reference that limited the amount of denoting a name alone could do.