The right thing to say, from a Wittgensteinian point of view, is that communication runs no risk of collapse, and that this is one of the things we can be sure of, even though our certainty rests on nothing at all.

Dummett, The Logical Basis of Metaphysics

How do we manage to grasp the meaning of a sentence, and how can we be sure that whoever is listening to us understands the sentences we utter just as we want them to be understood?

This is the question answered, at the beginning of the twentieth century, by the founder of modern logic, the German philosopher Gottlob Frege, when he formulated the well-known and much-debated principle of compositionality. According to this principle, the meaning of a proposition depends on the meaning of the words that compose it and on the way they are combined in the sentence. To understand the utterance “The cat chases the mouse” I must know the meaning of “cat”, “chase” and “mouse”, and be able to tell it apart from the sentence “The mouse chases the cat”, which, in English, has a different meaning solely by virtue of the order of the words. The mechanism of comprehension would follow a bottom-up path, that is, from the component parts to the structured whole, and everyday practice would seem, at first sight, to confirm this hypothesis. For if utterances were not structured wholes, from both a syntactic and a semantic point of view, there would be no explaining how we understand new utterances never heard before, starting from a limited number of words and from a relatively narrow set of syntactic combinations. Were it otherwise, we should have to learn every utterance (potentially infinite in number) one by one, something our limited memory would not allow. Underlying this principle, moreover, is the assumption that the meanings of the constituent parts are stable. The same word must, that is, keep its meaning stably, within certain limits, across different utterances: we expect the verb “to fly”, for example, to have the same meaning both in the utterance “I fly with Alitalia” and in the utterance “The canary has flown out of the cage”. But how, then, do we understand the expression “Offensive words flew”? The first limit to the principle of compositionality is set by all the metaphorical, idiomatic and rhetorical uses of language, but not only. None of the deictics (here, now, then, you, she) could make any contribution of meaning to the sentence except in the context in which the utterance itself is produced, to say nothing of vague or indeterminate expressions, which disambiguate their meaning only in the context of the utterance: in the sentence “she is a tall girl”, tall is a vague term: tall by how much? compared with whom?
The pervasiveness of these phenomena means that the principle of compositionality is, as Picardi notes (2009:69), a necessary but not a sufficient condition for determining what utterances mean. An utterance is not simply made up of strings of words, and understanding an utterance cannot be reduced to understanding its constituent parts. As Dummett puts it, the relation between the word and the utterance appears to be far more subtle: “Understanding an utterance implies understanding its constituent words, but is not reducible to it” (Dummett, 1996:142). Knowing the meaning of a word is like knowing which way the switches go to turn on the light. Normally I turn on the light without being aware of the right direction, not because I do not know it, but simply because under normal conditions there is no need to keep it in mind. It is only under exceptional conditions (the bulb has blown, and to change it I have to turn off the light) that a higher level of awareness comes into play.

Our knowledge of the meaning of any given utterance derives from a prior (implicit and general) knowledge of the words that compose it and from our observation of the particular way they combine within the utterance. “For the purposes of recognition, then, the sense of the word is prior to that of the utterance” (Dummett, 1983:46). It is only by taking this aspect into account that one can explain how we manage to understand and to think new utterances, never heard before. We manage to do so because the sentence carries a familiarity with respect to each word that occurs in it. When, however, it is a matter of explaining what it is for words to have a sense, what it is for us to grasp the sense of a word, it is the sense of the utterance that is prior to that of the word. For Dummett the meaning of the word is a generic and partial meaning, one that evokes familiarity and not identity; it is the meaning given by dictionaries, so to speak, which is as it is by convention, by social agreement, but this meaning is disambiguated in the particular act of utterance.

In fact Frege himself had set the context principle against the principle of compositionality, saying precisely that “the unit of meaning is not the word but the utterance”. We can grasp the meaning of an utterance on the basis of the words that compose it, but the meaning of the word is given only in the context of its use. The two processes, however, are by no means in contradiction with each other. They are two different processes of analysis: analysis into constituents presents the utterance as it is structured, its internal constitution, whereas analysis by decomposition reveals how we see, break down and understand the utterance. But how are the two processes related? Is the analysis that goes from the parts to the whole equivalent to the analysis that goes from the whole to the parts? Leaving aside here the difficulty, already pointed out by Wittgenstein in the Philosophical Investigations § 47, of defining what is simple and what is composite, that is, the very difficulty of objectively identifying the parts of a supposedly composite element, let us dwell a little longer on the dynamics that emerge as we proceed from the whole to the particular and from the particular to the whole.

The use of a word in one utterance is connected to the whole set of possibilities of its use in other utterances. In § 199 of the Philosophical Investigations Wittgenstein had stated that “to understand a proposition means to understand a language”, and on the strength of this sentence Wittgenstein has been cast in the role of forerunner of semantic holism, which pushes Frege’s context principle to its extreme consequences. Instead of proceeding from the parts to the whole, one proceeds from the whole to the parts, in a top-down fashion. The work of identifying the parts is in reality relative to the hypotheses formed by a hypothetical radical translator faced with utterances in a language he does not know at all, as Quine maintains. The final result of a mental procedure of this kind is that there exist countless conjectures, all equally valid, which do not coincide with the meanings as we know them from dictionaries, but which work the moment the interlocutor gives his assent. Meaning, then, would be nothing other than an agreement between the parties involved in the mechanism of communication.

But how can I be sure of being understood in the way I mean, if, in every communicative exchange, everything is always called back into question? Might not what Pirandello has one of his characters say then be true: “And how can we understand each other, sir, if into the words I speak I put the sense and value of things as they are within me, while whoever listens to them inevitably takes them with the sense and value they have for him, of the world as he has it within himself? We think we understand one another; we never do!” (Six Characters in Search of an Author).

The criticism Dummett levels at semantic holism is that it gives up investigating how language works, which, in everyday life, proves to work perfectly well, even when we misunderstand one another, indeed we misunderstand one another precisely because we normally understand one another. Dummett therefore starts again from the principle of compositionality, which […] is not reducible to the mere truism, which even the holist cannot but accept, that the meaning of an utterance is determined by its composition. The bite of the principle derives from the thesis that understanding a word consists in the capacity to grasp characteristic elements belonging to a particular range of utterances that contain it. (1996:315)

We may call dependence the relation that holds between one expression and another when understanding the first requires understanding the second. What the principle of compositionality essentially requires is that the relation of dependence between expressions and sentence-forms be asymmetrical. More precisely, in order to allow for the existence of sets of mutually dependent expressions, the relation must be conceived as holding between ranges of expressions rather than between single expressions. […] Compositionality requires that the relation of dependence impose on the sentences of the language a hierarchical structure that deviates only slightly from a partial ordering. The deviation allows there to be ranges of expressions the understanding of whose meanings can be acquired only simultaneously; as regards these ranges, however, the understanding of an expression of a given range can depend only on the understanding of expressions that stand a step lower in the hierarchy. (1996:312)

There would thus exist independent, progressive and interconnected fragments of language. To understand an utterance of a given language it is enough to know some fragment of it, which could in itself constitute a complete language. This system makes it possible to explain how we acquire language as children and how we learn a foreign language as adults, how we manage to formulate and understand utterances without ever attaining a total knowledge of the language, not even of our own mother tongue.

Within the framework of this system there are no isolated expressions, but expressions that we can understand only together with others, for example pairs or sets of contrary predicates. Colour words are a plausible example of this: we can tell one colour from another only when they are set side by side; the same holds for the pairs “male” and “female”, or for words indicating relations, “father of”, “mother of” and “son of”. Or again, to understand a trait of character one must first understand what it means to attribute that characteristic to actions: to understand what “to be vindictive” means, one must understand what it is to perform “a vindictive action” (Dummett, 1996:311). The order of priority of an utterance over the word will further require that understanding a word consist in the capacity to grasp certain ranges of the first type in which the word appears, before being able to grasp certain others. Understanding the word “fragile”, for instance, as it emerges in its abstract meaning in the sentence “He is a fragile person”, will normally require the prior understanding of the word “fragile” in its concrete meaning, as it may be, for example, in the sentence “This plate is fragile”, without, however, being a condition of its comprehension. (Dummett, 1996:313). It is precisely on the basis of this mechanism that, for example, syllabuses for learning foreign languages are structured, syllabuses which, not by chance, follow well-defined sequences of acquisition.

Unlike formal languages, in which the hierarchical order is determined a priori, in natural languages it is by no means easy to determine what the various ranges of utterances are and in what complex intersection of levels they are placed. Holistic theories give up this distinction: in them the condition for understanding any utterance whatever of a language is that one be able to understand the entire language. But this leads, according to Dummett, to a stalemate in which no further investigation of how the language works would any longer be possible.

Understanding, like knowledge, is a matter of degree: I can understand the meaning of an utterance partially and at various levels. I may understand the sentence “GDP has collapsed” as an ordinary citizen and worry about the consequences that will follow, but I will understand its meaning at a level of complexity certainly lower than that of an expert economist. If, however, the understanding of the expressions of a language rests on the prior understanding of other, lower-level parts of the language, one will reach a point (unless one wishes to fall into a process ad infinitum) at which there must be some part of the language that the speaker has to understand first, in a non-verbalizable way or, as Dummett puts it, implicitly. “For a word to be of this kind, there must be some part of the language that the speaker has to understand before he can understand the word. It follows from this that not all words and not all expressions can be of this kind, and therefore not all the knowledge that is constitutive of understanding the sense of a word can be verbalizable knowledge”. (Dummett, 1996:213).

What kind of knowledge is this capacity to intuit the meaning of a word in a non-verbalizable way? If, as Dummett says, it is the task of the meaning-theory (a meaning theory) to explain what the processes of formation and signification of the individual words and of the expressions belonging to a particular language are, it is the task of a semantic theory (theory of meaning) to investigate the general principles, that is, to ask what kind of semantic value an expression ought to have: whether it is a value of the implicit-intuitive type or a verbalizable value, and therefore of the logical-deductive type. (Dummett, 1996:214). If, then, the principle of compositionality and the context principle appear to lie at the base of the meaning-theory, semantic theory, as Dummett understands it, will have to rest on principles that trace a system of far greater complexity. Always assuming, of course, that one presupposes the existence of a system to be investigated.

And it is precisely at this point that we seem to catch an analogy between Dummett’s semantic theory and the theory of complex systems. The concept of a complex system, as defined by James Grier Miller, is a set of units that interact by establishing relations among themselves, where the word set indicates that the units have common elements. (Miller, 1978:16).

What characterizes a complex system, then, is the concept of structure understood as a network of relations among the elements that constitute it; the interaction among these elements determines its stability, however temporary: the system persists as long as the elements interact with one another while retaining their properties; if the elements cease to interact, the system degenerates or is transformed. To these fundamental concepts two others are added that define a complex system: the concept of emergence, that is, the arising of phenomena that cannot be explained simply by reference to the constitutive aspects of the individual parts, in other words the sum of the parts is not equivalent to the whole; and, further, the concept of logical openness proposed by Peirce. According to this concept, in a complex system a creative process is set in motion, one of inventing hypotheses, which presupposes the possibility of choosing, among a range of possible variables, the most effective ones. (Minati, 2010:15-46)

To the atomistic vision (which emerges from the principle of compositionality) and the holistic one (which pushes the context principle to its extreme consequences), Dummett adds the molecular vision of language, which seems to us to meet the criteria for describing a complex system. There are, that is, fragments of language that can be regarded as autonomous and progressive systems: the acquisition of certain fragments is the condition for understanding subsequent fragments. The fragments are not closed, however, but interact with the other fragments/systems within a network of non-linear and dynamic interactions. At the base of the system lies the meaning of the word, whose nature is manifold: depending on its surroundings it is explicit-deductive or implicit-intuitive meaning, and it follows dynamics that run from the whole to the parts and, at the same time, from the parts to the whole, with a high margin of unpredictability in the results. The pervasiveness of phenomena of vagueness and indeterminacy of meanings in fact makes of language a creative code, as the linguist De Mauro calls it, and therefore an unstable one, that is, a system that entails the continual redefinition of its object. The gaps in language of which Dummett speaks would be nothing other than the very conditions of possibility for the system to exist as a dynamic and adaptive system, and in this resides the confidence that we understand one another and, sometimes, that we do not.

Bibliography

De Mauro, T., Minisemantica, Laterza, Bari 1982.

Dummett, M., The Logical Basis of Metaphysics, Duckworth, London 1991 [It. trans. edited by E. Picardi, La base logica della metafisica, Il Mulino, Bologna 1996].

Dummett, M., Frege: Philosophy of Language, Duckworth, London 1973 [It. trans. edited by C. Penco, S. Magistretti, Filosofia del linguaggio. Saggio su Frege, Marietti, Casale Monferrato, 1983].

Miller, J.G. (ed.), Living Systems, New York, McGraw-Hill, 1978.

Minati, G., “Sistemi: origini, ricerca e prospettive”, in Strutture di mondo. Il pensiero sistemico come specchio di una realtà complessa, edited by Lucia Urbani Ulivi, Il Mulino, Bologna 2010, pp. 15-46.

Picardi, E., Le teorie del significato, Laterza, Bari 2009.

Wittgenstein, L., Philosophische Untersuchungen, Basil Blackwell, Oxford 1953 [It. trans. edited by R. Piovesan, M. Trinchero, Ricerche filosofiche, Einaudi, Turin, 2014].

Since 2005 Rosa Errico has taught Italian language and language teaching methodology at the Ludwig-Maximilians-Universität in Munich. With the publishers Guerra Edizioni and Klett Verlag she has published articles, textbooks and grammars for the teaching of Italian to foreign learners.