Marcelo Dascal

Adam Smith’s Theory of Language


Marcelo Dascal


1.      Introduction


Adam Smith’s lasting fame certainly does not come from his work on language. He published very little on this topic and he is not usually mentioned in standard histories of linguistics or the philosophy of language. His most elaborate publication on the subject is a 1761 monograph on the origin and development of languages (FoL). Smith’s monograph joins a long list of speculative work on this then fashionable topic (cf. Hewes 1975, 1996). The fact that he later included it as an appendix to his successful Theory of Moral Sentiments (TMS) indicates that Smith “set a high value” upon this monograph (Stewart 1793: 32) – an appreciation he did not bestow upon his lecture notes on rhetoric and literature, which he consigned to the flames.[1] Although Smith devoted most of his teaching to language-related topics, and certainly developed an organized body of convictions about the subject, it would be an exaggeration to say that he had a “theory of language”.[2]

Nevertheless, as I will try to show, his reflections on language are worth examining, for several reasons. First, they reveal the assumptions about the nature and functions of language underlying the work of an extremely self-conscious writer, whose care for style and clarity of exposition was apparent in all his writings – a fact that was highly praised by his contemporaries and by present day readers alike. Second, they highlight interesting connections between the better known parts of his work, which might help to reconstruct his general underlying epistemology. Third, they help to situate his work in the context of a century where the relationship between language and knowledge often functioned as an indicator of a thinker’s stance on other philosophical and social issues.

In this chapter, we will begin with a very brief summary of the agenda of  language studies legated by the 17th century to the 18th, highlighting the issue of the relationship between language and mind (section 2). We will then analyze Smith’s FoL and explore other aspects of his philosophy of language as expressed in his rhetoric and moral philosophy (section 3). Finally, some suggestions will be made about the significance and eventual influece of his reflections on language (section 4).


2.      Language and mind: from the 17th to the 18th century


The 18th century inherited from the 17th a philosophical agenda where language loomed large. Locke had made language an “official” philosophical topic, by devoting one of the four books of his Essay on Human Understanding (1690) entirely to language. This gesture crowned a century-old burgeoning concern with language –  especially with its relationship with thought and knowledge. Whether in order to warn against the danger of falling prey to the linguistic “idols of the marketplace” (Bacon) or to raise natural language to the position of the reasoning tool par excellence (Hobbes), all major 17th century philosophers addressed the question of language’s epistemic vices or virtues.[3] And the 18th century followed suit, albeit with some noteworthy differences in emphasis and method.

            The central items in the linguistic-epistemic agenda at the turn of the century are, for the most part, represented in Locke’s Essay, which was extremely influential throughout the 18th century. Locke’s views on the relationship language-mind were by no means unanimously accepted, but they provide an excellent vantage point for understanding the issues under discussion.

            Chapter 7 of Book III of Locke’s Essay, devoted to the “particles”, epitomizes the central issue. Particles are words that function as “marks of some action or intimation of the mind” (3.7.4). Their analysis, therefore, lead directly into the study of “the several views, postures, stands, turns, limitations, and exceptions and several other thoughts of the mind” (3.7.4; see also 3.7.6). Leibniz, in the corresponding chapter of  his Nouveaux Essais, aptly summarized, generalized, and somewhat twisted the above suggestion by Locke, while enthusiastically endorsing it: “I truly believe that the languages are the best mirror of the human mind, and that a precise analysis of the meaning of words would provide, more than anything else, knowledge of the understanding’s operations” (Nouveaux Essais 3.7.6; Leibniz 1705: 313).[4] In the formula “language(s) = mirror of the mind” he concisely formulated the most important theme in the linguistic-epistemic agenda of the 17th and 18th centuries. This formula is conveniently ambiguous as to the precise nature, causes, and consequences of the “mirroring” relation it makes use of, as well as of the languages and mental attributes and functions it refers to – an ambiguity that permits it to stand for the variety of positions on the language-mind relationship that thrive in the 17th and 18th centuries.

The main underlying question is whether language mirrors the mind only because it serves to convey to others one’s language-independent mental contents or because it also somehow participates in the mental operations involved in the formation of such contents. The answer to this question entails the choice of a favored mode and direction of explanation. For those who hold the first view, the linguistic should ultimately be explained in terms of the mental, since the latter is the “primary” phenomenon, whereas the former is the “derivative” one. On this view, linguistic observations can be useful in our inquiries about the mind, just as the observation of effects is useful in understanding their causes. For those who hold the second view, language plays a constitutive causal role in mental life, and therefore is part and parcel of the explanation of at least some of its aspects. Locke held the first view, along with most of the thinkers of the period; Leibniz, along with Hobbes and Condillac, held the second, minoritary view.

Although Locke suggests that “knowledge …has greater connexion with words than perhaps is suspected” (Essay, 3.1.6),  this connection – within his theoretical framework – can only have the auxiliary role of supporting his theory of ideas and of helping to avoid the mistaken inferences about the mind suggested by a careless consideration of words.[5] Whatever its supporting role, language, in so far as it mirrors the mind, does so “from the outside”. This view is shared by most 17th century rationalists.Descartes, to be sure, considered human linguistic behavior to be the proof that humans – unlike animals and machines – have a mind, but he and his followers rejected with horror Hobbes’s suggestion that words might play any role whatsoever in what goes on within the mind itself (see Dascal 1994, 1996, and forthcoming). Similarly, the Port Royal logicians and grammarians, although viewing logic and grammar as symbiotic disciplines, did not question the priority of the former. For them, all languages share a ‘deep grammatical structure’ – an idea that is the core of their ‘general grammar’ research program – because the logical structure of thought underlies them.[6]

An alternative – and minoritary – view was proposed by Hobbes (cf. Dascal, forthcoming) and by Leibniz (cf. Dascal 1998). Each in his own way defends the idea that language has, in addition to its communicative function,  a constitutive role in thought. On this view, important cognitive functions cannot not be performed without the assistance of linguistic or other signs. While Hobbes talks of all reasoning as ‘reckoning’ or computation performed by means of words, Leibniz insists on the necessary role of signs in virtually all cognitive operations, from the formation of mildly complex concepts to the performance of any moderately long inference (see Dascal 1998). From this point of view, language ‘mirrors’ the mind in a much more intimate way than as an effect mirrors its cause. Since mental processes are not language-independent, by examining language we are somehow looking at the mental operations themselves, and we are in a position to provide a fuller explanation of  them. The ‘particles’ found in natural languages, in this respect, acquire a special significance, for they both give ‘form’ to discourse  and  weave the relational structure or form of thought (Dascal 1990).

The critique of language and the search for remedies for its deficiencies was part and parcel of the philosophical concern with language. The two camps described were concerned with it, since – whether only communicative or also constitutive – only a duly ‘sanitized’ language could be a reliable epistemic tool. Accordingly, there was considerable agreement as to the sanitation measures required. For example, the exclusion of figurative language was viewed as a sine qua non for language to perform properly its cognitive functions. As a result, rhetoric or dialectic, which had been in the renaissance a major component of the ars inveniendi (method of discovery), was removed from the epistemic agenda and relegated to non-cognitive tasks. Agreements such as these, however, did not obliterate the divergence regarding the explanations given to language’s deficiencies and the steps proposed to overcome them. For the ‘communicativists’, the problem lied not so much in language as in thought itself and linguistic reform per se would be of no avail. The ‘constitutivists’, who viewed language not merely as a tool for the transmission of knowledge but also for its acquisition, were willing to invest more effort in devising suitable notations, careful definitions, and even in designing new, transparent and precise languages which – they hoped – would improve the quality of thought and knowledge.[7]

In the eighteenth century,  where the Newtonian paradigm of science is extended to the ‘moral sciences’ and an unprecedented systematization of knowledge is undertaken by the encyclopédistes, the epistemic significance of language can no longer be treated as a relatively marginal topic. Academies offer prizes for the best essays on the relationship between language, mind, and the advancement of knowledge, and the best minds of the century comply. The discussion of these issues becomes so intense in the second half of the century that Kant’s silence about them in his Critique of Pure Reason (1781) is sharply criticized by his contemporaries Herder and Hamann (cf. Dascal and Senderowicz 1992). Some of the 17th century’s concerns arouse less interest[8] and other themes – such as inquiries about ‘the origin of language’ become more fashionable; but the issue that organizes the whole field remains that of the precise role of language in cognition.[9]

Although initially repudiated as dangerously anti-scientific, atheistic and materialistic, the idea that language is an indispensable constituent of cognitive processes slowly gains adherents, even among the champions of the ideationist theory of mind,[10] and by mid-century the view that language is an indispensable component of cognition is no longer minoritary. It is Condillac, who describes himself as a follower of Locke, that articulates forcefully and influentially this view (cf. Aarsleff 1982: 146-209; Dascal 1983; Formigari 1993: passim). In the Essai sur l’Origine des Connoissances Humaines (1746), Condillac examines in detail all the operations of the mind, from perception to reasoning, in order to show “how and in what order” the former engenders (produit) all the others (Essai, Introduction, p. 4). The underlying principle of this progress from the simplest to the most sophisticated mental operations is what he calls the liason des idées.[11] Since signs play a crucial role in this liaison (Essai, Introduction, p. 4), Condillac pursues a parallel inquiry about the development of our semiotic systems, in order to show how full-fledged conventional, articulated languages, writing, poetry, scientific notations, etc. emerge step by step from a natural, inarticulate and primitive langage d’action. The “history of language”, he claims, “will show the true meaning of signs, will teach us how to avoid their abuse, and will leave no doubt about the origin of our ideas” (ibid.). The Essai is thus an account of the simultaneous and interdependent ‘evolution’ of language and mind. It becomes the classical model of a type of inquiry where the emphasis lies more on logical reconstruction than on chronology and where the appeal to super-natural causes is replaced by the search for a ‘natural origin’ and a ‘natural history’ of mind and language. Smith’s monograph on the formation of languages (FoL), as well as several of his other works belong to this type of inquiry.[12]

Condillac reaches the conclusion that the higher mental operations (beginning with “reflection”) couldn’t evolve without the appearance of articulated language (Essai, Unlike other empiricists, he pointed out that sensations – which provide the “materials” for the mind’s operations – are complex conglomerates of stimuli, very far from Hume’s “impressions”, which were taken to be the sensory counterparts of “simple ideas”. In order to reach the building blocks out of which concepts, judgments and reasonings are composed, we need an “analytic tool” that decomposes sensations. This tool we acquire with the emergence of language. Just as the availability of numerical signs is essential for calculating, so too words are essential for thinking in metaphysics and ethics (Essai In spite of his praise for Locke, Condillac blames him for assuming that “the mind makes mental propositions where it connects or separates ideas without the intervention of words” (Essai Condillac, thus, seemed to reinstate – in the new ‘origins’ terminology – the Hobbesian thesis that thinking is nothing but mental word juggling. Horne Tooke, an influential and controversial figure in English ‘philology’ in the second half of the 18th century, put this bluntly by claiming that “what are called [the mind’s] operations are merely the operations of Language” – a thesis he sought to demonstrate with the help of ‘etymological’ evidence.[13]

But not everyone was so eager to accept this reduction of thought to language, for a variety of reasons. Apart from the charge that it led to atheism, it was seen by many (e.g., Shaftesbury, Hutcheson, Harris, Reid) as a materialist reduction, that deprived the mind from its spiritual, active powers. Furthermore, its underlying anti-essentialism or ‘nominalism’, which made thought dependent upon arbitrary linguistic signs, challenged the assumption of the universality of knowledge and raised the specter of relativism and skepticism. This was viewed as particularly dangerous for ethics and politics, and efforts were made to protect our ‘inner moral sense’ from this danger by separating it from language.[14] The same danger was perceived in aesthetics, and was the subject of the extensive debate throughout the century about the ‘standard of taste’ and the principles of criticism (e.g., Shaftesbury, Hume, Burke, and also Smith).[15] Ultimately, the possibility that language – unequivocally a result of social interaction – is deeply and essentially involved in our mental life, brought to the fore the tension between the traditional view of man as an individual, autonomous subject and the emerging awareness of the importance of his social nature. In an age where the very idea of ‘social sciences’ was taking shape, this issue was crucial. It was not by chance that Rousseau (1755) discussed it in one of his social treatises (see note 20). Nor was it by sheer accident that the Berlin Academy in 1769 offered a prize (won by Herder) for the best essay on the topic.


3. Smith on language


In search of a system


In his earliest linguistic publication, a review of Samuel Johnson’s Dictionary of the English Language, published in the Edinburgh Review (1755), Smith praises the pioneering character of this work, but criticizes it for not being systematic enough. Its plan, he says, is not “sufficiently grammatical”; the different significations of a word are “seldom digested into general classes, or ranged under the meaning which the word principally expresses”; and “words apparently synonymous” are not carefully distinguished (EW 5-6). He proposes his own alternative to two of Johnson’s entries, one of which, but, had been analyzed by Locke. According to Smith, but is a particle that holds the place of various grammatical categories “according to the different modifications of the general sense of opposition” (EW 9). Johnson’s eighteen meanings of but are then accommodated into seven grammatical categories and sub-categories and, for the two main senses, ‘adversative’ (= however) and ‘alternative’ (= unless, except), Smith explains the difference between but and its English synonyms. These explanations employ a mixed lot of syntactic, stylistic, and semantic considerations. Although his semantic explanations refer to “mental operations”, he does not seem to follow Locke in taking this to be the main criterion for explaining the particles’ meanings. Unlike Locke, who expresses his dissatisfaction with grammatical terminology and seeks a systematic account in terms of the “several relations the mind gives to the several propositions” (Essay 3.7.5), the systematicity Smith seeks is rather that of a conservative grammarian. Nor is he concerned – as Leibniz (Nouveaux Essais 3.7.5; GP V 312) – with providing a precise definition for each meaning and with subsuming them under a more specific notion than ‘opposition’.

            Eight years later, commenting on the abstract of Ward’s “plan for a Rational Grammar”, Smith praises it as a blueprint for “the best System of Grammar” as well as for “the best System of Logic in any Language” and “the best History of the natural progress of the Human mind in forming the most important abstractions upon which all reasoning depends”.[16] He goes on to describe what his own plan for treating the same subject would look like. He would begin with the verbs, then show how subject, attribute and object were separated from the verb, and likewise investigate “the origin and use of all the different parts of speech and of all their different modifications, considered necessary to express all the different qualifications and relations of any single event” (ibid.). This brief description sums up the essence of FoL, published two years earlier.[17] FoL, however, does not begin with the verb, although at about midway attention is shifted to verbs, which “must necessarily have been coëval with the very first attempts towards the formation of language” (FoL 27, LRBL 215; EW 239). This discrepancy hints at the different principles of organization underlying FoL and perhaps the whole of Smith’s reflections on language.


Language origins: structure and cognition


FoL begins with proper names, “probably one of the first steps towards the formation of language” (FoL 1, LRBL 203; EW 225). By inventing proper names, “two savages, who had never been taught to speak”, would be able to indicate to each other those particular objects which were most familiar to them. Next, they would naturally apply these names to other objects exactly resembling the original ones. This diachronic antonomasia – Smith observes – is parallel to its synchronic counterpart, which we still use when we call an orator a Cicero or a philosopher a Newton. This procedure generates general names and gives occasion to the formation of classes “of which the ingenious and eloquent M. Rousseau of Geneva finds himself so much at a loss to account for the origin” (FoL 2, LRBL 205; EW 227).

            Rousseau's difficulty was a sort of “who came first, the chicken or the egg?” puzzle. Generalization, he said, requires words, but words in turn depend upon the ability to generalize. Smith's solution at first seems to break the circle by espousing a nominalist position: classes arise by virtue of using a name for a multitude of objects. On closer inspection, however, he is claiming that it is not the name that creates the class, but rather the resemblance between its members, which entitles one to apply to them the same name.[18] Just as he assumes ‘familiar objects’ as the natural pre-linguistic candidates for proper names, he also assumes ‘natural kinds’defined in a natural similarly space as pre-linguistically given bearers of general names.[19] It is questionable whether, by side-stepping in this way Locke's anti-essentialism, Smith has indeed addressed Rousseau’s problem.[20]

            The next problem for the savages is finding a way of particularizing a general name in order to refer to an individual object for which no proper name is available. This can be done either through the individual’s qualities or through its relations to other things. Thus come into being two other categories of words – adjectives and prepositions. The invention of the former presupposes the capacities to compare objects and distinguish those having a quality from those not having it, to distinguish the quality from the object to which it belongs, and to conceive the object as subsisting without the quality. “The invention, therefore, of the simplest noun adjectives, must have required more metaphysics than we are apt to be aware of “ (FoL 7, LRBL 207; EW 229). ‘More metaphysics’ means  “more mental operations”,[21] and Smith concludes this paragraph by stressing that these operations – some of which remind Condillac’s ‘analysis’ – “must all have been employed,

before … nouns adjective could be instituted ” (ibid.; my italics). From this he infers “that when languages were beginning to be formed, nouns adjective would by no means be the words of the earliest invention”(ibid.).

            Here too Smith is ambivalent regarding the language-mind relationship. For his inference may be warranted either by the need of an earlier stage of language which was necessary for developing the mental operations required for inventing adjectives, or else simply by the fact that it is more likely that other words, requiring “less metaphysics” were invented earlier. What is clear, at any rate, is the criterion he has been using so far to establish the chronological order of appearance of the different kinds of words: the more mental operations are involved, the later a category of words will be invented.

            The next category to appear is the preposition, which requires more powers of abstraction, generalization, distinction, and comparison than the invention of adjectives (FoL 12, LRBL 209-210; EW 235-236). Not all prepositions are equally “general, abstract and metaphysical”. For example, above and below, which denote specific spatial relations, are less so than of, which denotes “relation in general”. The latter’s interpretation is context-dependent, i.e. the particular relation it denotes “is inferred by the mind”, thus requiring more mental machinery than the more specific prepositions. According to the principle “more metaphysics, later invention”, this means that prepositions such as of, to, for, and with “would probably be the last to be invented” (ibid.). After the preposition comes number which, “considered in general, without relation to any particular set of objects numbered, is one of the most abstract and metaphysical ideas … and, consequently, is not an idea, which would readily occur to nude mortals, who were just beginning to form a language” (FoL 22, LRBL 214; EW 237).

            The sequence of inventions described above amounts to the conjectural development of languages Smith calls ‘compounded’ (later to be called analytic’), e.g. English. But he describes also a parallel sequence that would have led instead to ‘original’ (latter to be called ‘synthetic’ or ‘agglutinative’) languages, e.g. Latin. The latter evolve when, instead of creating separate words for qualities, relations and numbers, the inventors of language resort to the “expedient” of attaching to substantives different terminations. The qualities of having or not having sex and, in the former case, of being male or female are expressed in this way in Latin and Greek; and the expression of relations in these languages by means of cases instead of prepositions “is a contrivance of precisely the same kind” (FoL13, LRBL 210; EW 233). Languages that follow this path are ‘original’ because the expedients they use require “less metaphysics”, so that those “rude” first inventors of language would “naturally” tend to invent them instead of their analytical counterparts.[22]

            Smith is thus employing consistently the same principle, i.e. the amount and degree of sophistication of the mental operations required, in establishing the chronology of appearance of the parts of speech and of the types of languages.[23] Compounded languages would tend to appear earlier because of a ‘least effort’ principle. However, Smith points out, their ‘synthetic’ method has severe limitations: how many qualities, relations and numbers can be represented by nominal declensions without encumbering the linguistic system so as to make it useless? Greek, for example, has five cases, three numbers, three genders, not to speak of the “infinitely more complex” verb conjugations (FoL 24-26, LRBL 214-215; EW 238-239). It seems, thus, that there is a trade-off between parsimony regarding mental operations, i.e., simplicity at the mental equipment level, and complexity of the linguistic system. However diachronically “natural”, easier and useful within the limited range of notions used by primitive men, the latter puts a heavy burden on the synchronic use of language for the communicative needs of more advanced humans. Were they to use only the limited number of declensions and conjugations available for expressing qualities, numbers, relations, persons, time, and aspect, they would in fact need to rely more rather than less on a wider range of mental operations such as memory and contextual inference.

            But, even within the presumed limited scope of the original compounded languages, is it the case, as Smith claims, that the invention of morphological variations of words does not require as much analytic powers as the invention of separate words? In his discussion of the synthetic way of expressing relations, Smith argues that (a) abstraction was not needed because the relation is expressed here “as it appears in nature, not as something separated and detached, but as thoroughly mixed and blended with the co-relative object” (FoL 14, LRBL 211; EW 234); (b) generalization was not needed because, although a word like arboris (‘of the tree’) contains “in its signification” the same relation expressed by the English preposition of, arboris (unlike of) is not a general word that “can be applied to express the same relation between whatever other objects it might be observed to subsist” (FoL15, ibid.; ibid.); and (c) comparison is not needed because the rule assigning the same termination to other nouns in order to express the same relation would arise “without any intention or foresight in those who first set the example, and who never meant to establish any general rule. The general rule would establish itself insensibly, and by slow degrees, in consequence of that love of analogy and similarly of sound, which is the foundation of by far the greater part of the rules of grammar” (FoL 16, ibid.; ibid.).

            Argument (b) emphasizes the grammatical distinction between morphs and lexemes and assigns to it a semantic significance it hardly has: first, because both are equally syncategorematic, i.e. the interpretation of both depends upon the words they are ‘attached’ to, as Smith himself remarked when discussing the extreme generality of prepositions such as of, which “in modern languages hold the place of the ancient cases” (FoL 19, LRBL 212; EW 235); second, because once the same morph is regularly used to express the same relation when attached to other nouns, it becomes as general a symbol of that relation as the corresponding preposition. Smith, perhaps anticipating this objection, argues in (c) that this rule or regularity emerges ‘naturally’, i.e. without design or intention. But this is immaterial to the issue at hand, for as soon as the termination -is  is applied to another word in the genitive case, it (not arboris) has been generalized; and this implies the same power of generalization as that involved in applying the adjective green to two different objects, referring to the same color. Perhaps he has in mind the further step of abstraction (not generalization) involved in the invention of a word like greenness. But it is doubtful that not performing this step differentiates between the inventors of the synthetic -is, and those of the analytic of.  It takes a logician or a metaphysician, not a regular language inventor (nor a later language user) to conceive of such things as ‘ofness’ (or ‘-isness’) and greenness.

 In (a) Smith appeals again to the notion of ‘nature’. Here the idea is that “in nature” what the first inventors of language encounter are “thoroughly mixed and blended” wholes, not yet analyzed into components of various sorts. The synthetic representation of relations fits better this presumed pre-analytic stage of our perception of the world because it is analogous to it, i.e. because its form structurally corresponds to it.  With this iconic conception of the naturalness of a language, Smith is introducing a new criterion of anteriority.  He is assuming that the better a linguistic form mirrors sensory input, the closer it is to nature (at least as it appears to us), and consequently the earlier it is likely to be created. This is a semiotic criterion, based on the type of relation that links signifier and signified. Its correlation with the mental operations criterion presupposes that iconic representation is mentally less onerous than other forms of representation, presumably because it relies upon a pre-existing ability to detect similarities – an assumption Smith had already made use of. The similarity in question here, however, does not require an imagetic or pictorial relation between the sign and its meaning. Smith does not question the arbitrariness of the sign, nor does he appeal to the ononatopoetic speculations about word origins that animated the etymological imagination of his century. He is talking about a structural analogy that could in principle be displayed by ‘analytic’ languages as well, whenever what they represent is itself segmented.

Smith’s use of this semiotic criterion to support the anteriority of synthetic language is thus entirely dependent upon the belief that sensations are “thoroughly blended and mixed” wholes, as Condillac argued, against Locke and the other empiricists.[24] Condillac’s initial language – the langage d’action – was also synthetic. But it was, semiotically speaking, indexical. Its cries, gestures and  “agitations of the whole body” (Essai do not depict the feelings and desires they express – they form with them a single behavioral whole. This, not iconicity, is what makes the langage d’action natural and different from the articulated languages that segment these initial behavioral and sensorial wholes. Furthermore, the savages who used the langage d’action had no leisure to contemplate and describe the world; what they were concerned with was “demanding and giving help to each other” (ibid.).

Condillac’s savages performed directive and expressive speech acts (requesting, warning, ordering, offering, expressing fear, comforting, etc.), while Smith’s language inventors were more concerned with assertive ones: “We never speak but in order to express our opinion that something either is or is not” (FoL 27, LRBL 215; EW 239). But both discover very soon that they need the verb: “No affirmation can be expressed without the assistance of some verb” (ibid.). Since there is no real speaking without verbs, their invention – to repeat – “must necessarily have been coëval with the very first attempts towards the formation of language” (ibid.).

With the verb, FoL returns to the beginnings of language and follows again the  path described for the noun. Generalization, abstraction and analysis led from impersonal verbs “which express in one word a complete event” (FoL 28, LRBL 215; EW 239) to the division of this event into its “metaphysical elements” (FoL 30, LRBL 218; EW 242) and to the institution of new words, or the use of the already available nouns for denoting “not so much the events, as the elements of which they were composed” (ibid.). Thus appear the distinctions of persons (which express “ideas extremely metaphysical and abstract”, especially the first person – FoL 32, LRBL 219; EW 243), subject and complement, tenses, voices, modes. Again, ‘original’ and ‘compounded’ languages diverged in their ways of representing these distinctions, the former morfologically, the latter lexically.

The verb narrative differs from the noun narrative in two important respects. First, it makes use of the semiotic criterion from the outset. Second, it introduces in the discussion, for the first time, a rhetorical comparison between the two types of language. These two considerations are interrelated. Impersonal one-word verbs “preserve in the expression that perfect simplicity and unity which there always is in the object and in the idea” (FoL 28, LRBL 215; EW 239). They mirror, thus, “the perfect simplicity and unity with which the mind conceives in nature … a complete affirmation, the whole of an event” (ibid., LRBL 216; EW 239). Smith mentions here several times the simplicity of the event and of the idea corresponding to it: “… in nature, the idea or conception of Alexander walking, is as perfectly and completely one simple conception, as that of Alexander not walking” (ibid.). He is stressing that, unlike relational complexes, which are mixtures (albeit “thoroughly blended”), events are inherently simple. Whereas the semiotic adequacy of a one-word representation of the former lies in the way they appear to us, that of the latter seems to lie at the deeper level of what ‘really is’. Therefore, to divide an event into parts is an artificial move. It results from the imperfection of a language that employs a multiple-word “grammatical circumlocution” when one word would suffice – a procedure whose “significancy is founded upon a certain metaphysical analysis” (ibid.). This procedure is not only needless but also misleading in the sense of Ryle’s (1932) “systematically misleading expressions”, for the structural analogy suggested by grammar, instead of revealing the logical form of things, hides it.

One can discern here a critical attitude vis-à-vis natural language that can be traced back to the tradition initiated by Bacon. But Smith does not espouse the wholesale baconian critique of the false classifications embedded in the languages created by vulgar unscientific minds – the critique that inspired the research program of a complete reform of scientific language.[25] Nor does what he says indicate that he would endorse Reid’s reliance on natural language in his critique of the empiricists’ ‘way of ideas’ (see Aarsleff 1967: 101). Smith’s attitude is more nuanced. Natural languages, as we know them today, are the result of a process of “mixture of several languages with one another, occasioned by the mixture of different nations” (FoL 33, LRBL 220; EW 244). Consequently they contain at best only some epistemologically and metaphysically reliable elements. These must be carefully distinguished from the unreliable ones, with the help of  – among other means – the kind of conjectural history and typology undertaken by Smith.

This nuanced attitude probably results from Smith’s realization that Girard’s fundamental parallelism parole/pensée/monde does not strictly hold. Girard, whose book (1747) Smith recommended to Ward as the linguistic book from which he “received more instruction than from any other”,[26] defined speech as “the manifestation of thought through words”, thought as arising from “the union of ideas”, and ideas as “the simple images of things” (Girard 1747: 5). As pointed out by Swiggers (Introduction to Girard 1982, p. 32), following Foucault (1966), Girard assumes that the structured world pre-exists thought and language and that its structure is transitively transferred to thought and language. But, in spite of his efforts, this presumed transitivity is precisely what Smith could not find, as we have seen. Consequently, the best System of Grammar, the best System of Logic, and the best History of the natural progress of the Human mind, are not mirror images of each other. The interest of undertaking these three enterprises together, as FoL presumably purports to do, lies in the instructive divergences one discovers between them.

In addition to the epistemic, metaphysical and grammatical perspectives, FoL also displays Smith’s rhetorical concerns as relevant to the assessment of language evolution, structure, and function. When he speaks of metaphysically artificial ‘grammatical circumlocutions’, he is also condemning prolixity as stylistically wrong. When he praises the iconicity and analogy of ‘original’ languages, he is also suggesting the value of apposite metaphors.[27] To be sure, progress in analysis has the functional cognitive advantage of yielding a system of language that is “more coherent, more connected, more easily retained and comprehended” (FoL 30, LRBL 218; EW 242). But its price is a loss of unity and simplicity of expression, which becomes instead “more intricate and complex”. The trade-off between these two desiderata is formulated by Smith as an ‘inverse law’: “In general, it may be laid down for a maxim, that the more simple any language is in its composition, the more complex it must be in its declensions and conjugations; and on the contrary, the more simple it is in its declensions and conjugations, the more complex it must be in its composition”(FoL 36, LRBL 221-222; EW 246). ‘Composition” here is more a rhetorical than a syntactical notion, and for Smith it has a considerable weight vis-à-vis the cognitive advantages of analytic ‘simplicity’.


Towards a theory of language use


This ‘rhetoricization’ of the discussion in FoL reaches its peak in the final paragraphs, which compare the evolution of languages to the evolution of machines – a rather expected simile in the Newtonian era. After showing how English is able to express all the tenses with the help of a few auxiliary verbs, Smith writes:


It is in this manner that language becomes more simple in its rudiments and principles, just in proportion as it grows more complex in its composition, and the same thing has happened in it, which commonly happens with regard to mechanical engines. All machines are generally, when first invented, extremely complex in their principles, and there is often a particular principle of motion for every particular movement which it is intended they should perform. Succeeding improvers observe, that one principle may be so applied as to produce several of those movements; and thus the machine becomes gradually more and more simple [p. 248] and produces its effects with fewer wheels, and fewer principles of motion. In language, in the same manner, every case of every noun, and every tense of every verb, was originally expressed by a particular distinct word, which served for this purpose and for no other. But succeeding observations discovered, that one set of words was capable of supplying the place of all that infinite number, and that four or five prepositions, and half a dozen auxiliary verbs, were capable of answering the end of all the declensions, and of all the conjugations in the ancient languages (FoL 41, LRBL 223-224; EW 249).



This sounds as the description of the ultimate linguistic achievement, especially for a self-proclaimed Newtonian like Smith. But it is surprisingly followed by the claim that, instead of perfection, it marks indeed imperfection:


But this simplification of languages, though it arises, perhaps, from similar causes, has by no means similar effects with the correspondent simplification of machines. The simplification of machines renders them more and more perfect, but this simplification of the rudiments of languages renders them more and more imperfect, and less proper for many of the purposes of language: and this for the following reasons (FoL 42, LRBL 224, EW 249).


These “other purposes” are nothing but the rhetorical virtues of eloquence, beauty, rhyme, mobility of words within a sentence, conciseness. On all these counts the ‘original’ synthetic languages reach “much greater perfection” than the modern languages, whose expressive power is severely limited by their inherent “prolixness, constraint, and monotony” (FoL 45, LRBL 226; EW 251).

            Smith lectured on “rhetoric and belles lettres” from 1748 to 1763. The third lecture of his last course, delivered on November 22, 1762, is a summary of FoL, which shows that he viewed FoL’s thematic as belonging to a broader concern with “explaining and illustrating the powers of the human mind” through an examination of “the several ways of communicating our thoughts by speech” and of “the principles of those literary compositions which contribute to persuasion or entertainment”.[28] This, and not the “taedious and unentertaining” (LRBL 8) details of traditional rhetorical treatises, is for him what rhetoric is all about, for “there is no art whatever that hath so close a connection with all the faculties and powers of the mind as eloquence, or the art of speaking”.[29]

            Smith’s rhetoric extends and deepens his theory of language in three significant ways. The first is a shift from a diachronic (even though conjectural) frame to an overtly synchronic one. Granted that English has “such great defects” due to “the very manner of its formation”, the question is now how English speakers and writers remedy them (LRBL 14). Phonetic contractions of all sorts help to overcome the inherent prolixity of the auxiliary verbs system; the compositional constraints of fixed order can be overcome by anteposition of “whatever is most interesting in the sentence” as well as by contractive stress (LRBL 18): only “ideots” and “a man who felt no passion” speak using “the most plain” syntactical order. Those who are affected by what they say will let the idea that affects them most “thrust itself forward” and will utter it “strongest” (LRBL 18, 23-24).

            Secondly, the range of mental operations related to language is extended well beyond the limited set of basic cognitive ones discussed in FoL, for “stile not only expresses thought but also the spirit and the mind of the author” (LRBL 19). The expression ‘spirit and mind’ covers, in addition to the author’s emotions, his character, ethical virtues, ability to argue, the knowledge he has of the theme treated, his sensitivity to the intended audience and circumstances, etc. The correspondence

between these mental properties and their suitable expression is ruled, according to Smith, by a general principle of propriety: “expression ought to be suited to the mind of the author” (LRBL 34). Given the variety of  ‘spirits and minds’, there is a corresponding variety of styles. The capital crime is to fashion to oneself a style (and a character) that does not fit one’s cast of mind, and therefore is artificial and sometimes ridiculous – as Smith claims referring to Shatesbury (LRBL 56, 59). Having extended as far as this the range of mental traits relevant for assessing the propriety of one’s linguistic behavior, Smith reaches an encompassing notion of ‘naturalness’ of which the ‘metaphysical’ and semiotic versions discussed in FoL are nothing but particular cases (LRBL 55).

            Finally, I think we can detect in Smith’s rhetoric a shift from language structure to language use as particularly relevant to the language-mind relationship. The above mentioned notions of naturalness and of what is of primary interest for the speaker (or writer) in each sentence he utters exemplify this shift. None of them can be accounted for in terms of language structure alone, for grammar is insufficient to express the ‘spirit and mind’ of the author (LRBL 19). The propriety of such an expression can only be judged by reference to the context where language is put to use to express a particular mental state. Writing that is unrelated to context  is unnatural (LRBL 58), and the writer must take into account his readers: the ease of understanding, care not to overburden the reader’s processing capabilities, an awareness for what one suggests (in addition to what one says).[30] This regard for the addressee is essential, along with clarity and true passion, for communication to proceed by sympathy (LRBL 25). Smith’s account of language use thus links up, unequivocally, with its ethical underpinnings.


Towards an ethics of communication


Leaving aside the more obvious connections between LRBL and TMS pointed out by the former’s editor (Introduction to LRBL, pp. 18-19), I would like to mention a couple of other significant relations between Smith’s theories of language and of morality.

            First, just as Smith’s ethics is a theory of the ‘propriety’ of action, his rhetoric is a theory of the propriety of linguistic action. The fundamental notion of Smithian ethics, sympathy, is in fact a principle of correspondence based on the instinctive analogy between what the other feels and what one would feel in similar circumstances (TMS We have seen the role of correspondence and analogy in the relationship between signifier and signified (FoL) and in the proper expression of a speaker’s or writer’s ‘spirit and mind’ (LRBL). In TMS, the emphasis is rather on the mutual correspondence that must obtain between interacting human beings in order to create and sustain social life. Accordingly, Smith stresses the human ability of mentally changing places with a sufferer (TMS,[31] Natural sympathy, cultivated and perfected by deliberate effort (TMS, yields a principle of ‘distribution of the effort’ each of the participants in the ‘communication of sentiments’ is required to perform:

[the effort] of the spectator to enter into the sentiments of the person principally concerned, and [the effort] of the person principally concerned, to bring down his emotions to what the spectator can go along with (TMS


To be sure, the “interval” between the participants’ sentiments (TMS cannot be entirely eliminated by this concerted effort, but its reduction is indispensable for ensuring the sharing of sentiments. Furthermore, it is essential that the effort be more or less equitably divided between the interactants. It would be morally improper, as well as inefficient, for any of them to lay upon the other the whole burden. The ‘spectator’ is required to make an effort to identify himself with the feeling of ‘the person principally concerned’, while the latter is required to make an effort to restrain the spontaneously strong expression of his emotion in order to let the spectator identify with his feeling, for, not being directly involved in what produced it, he cannot be expected to identify with its strongest expression. Replace ‘spectator’ by ‘hearer’, ‘person principally concerned’ by ‘speaker’, and ‘identify’ by ‘understand’ and the dialectics of emotion-sharing described by Smith becomes a formulation of the basic principle of communicative cooperation – the ‘division of communicative labor’. We encountered the rudiments of this principle in the rhetoric, and it comes as no surprise that, in TMS, he straightforwardly applies it to communicative acts such as excessive questioning motivated by unrestrained curiosity (TMS 7.4.28).

            The other parallel between Smith’s views on language and his ethics has to do with the notions of rule and system. Smith distinguishes between two kinds of rules in morality and in language (TMS 7.4.1), which I would call ‘algorithmic’ and ‘heuristic’. The former admit of no exceptions, whereas in the latter exceptions are the rule. Grammatical rules that admit of exceptions are simply bad rules that should be replaced by better ones. Smith criticizes the Greek and Roman grammarians for not having done so: “… when they came to find that many expressions could not be reduced to these rules, they were not candid enough to confess the grossness of their error and allow that these were exceptions to the generall they had laid down but stuck close to their old scheme” (LRBL 25). Rules of justice are of this kind. Morality rules for all the virtues except justice, on the other hand, cannot, by their very nature, be without exceptions (TMS 3.6.9). The reason for this multiplication of exceptions in this kind of rules is their context-dependence (TMS 3.6.9, 7.4.3-5).[32] Heuristic rules cannot provide,  therefore, a decision procedure for choosing a particular action in specific circumstances (TMS 3.6.11). Their function is not to decide mechanically for the agent, but to provide him with general guidelines for deciding by himself. Algorithmic rules, on the contrary, are perfect only when they give “infallible directions” that determine the correct choice. This is the ideal to be pursued by a correct grammar as well as by jurisprudence (TMS 3.6.10). In order to achieve this ideal, such rules are to be made as ‘context-free’ as possible, by incorporating all the relevant contextual specifications (ibid.), thus lending them infallibility, in grammar as well as in jurisprudence (TMS 3.6.11): ince they are absolutely dependable, they should be followed with “the most sacred, … reverential and religious regard” and often it is “a crime to violate them” (ibid.).

Algorithmic rules characterize the machine-like structure of what might be called the ‘basic institutional systems’ underlying communication and social life. Without the existence and unquestioned respect for such ‘structural laws’, neither could exist. In this sense they constitute an indispensable infrastructure, that should be valued as such, much like we appreciate the intricacy and cleverness of machines and contrivances independently of the ends for which they were contrived (TMS 4.1.3). But these rules must be complemented by another set of rules, the heuristic ones, which – so to speak – lift us above this structural ground. For we need also guidance in evaluating the different communicative and social ends we can pursue as well as the propriety of the linguistic and moral tools we use to that effect. Casuistry attempts to reduce the second type of moral rules to the first, and for this reason it is sharply criticized by Smith (TMS 7.4.16, 7.4.33); so too are the moralists, for their exclusive reliance on imprecise heuristic rules.  Smith’s message is crystal clear: moral philosophy requires the two different types of rule. Neither can be dispensed, nor can one of them be reduced to the other. Mutatis mutandis, this message should be applied to Smith’s philosophy of language. In addition to its grammatically and semantically codifiable part, a proper treatment of language should cover also the whole domain of language use (exemplified but not exhausted by rhetoric), a domain where the most appropriate rules are of the heuristic type (Dascal 1992).


4. Significance


Smith’s monograph on the formation of languages is extremely brief, if compared to the voluminous works by his contemporaries on the same subject. Although he addressed in it the main issue in the agenda of such studies – the relationship between language and mind – in a way that permits to locate him within Condillac’s ‘research programme’, his voice in the intensive dialogue his century sustained on this topic was a relatively minor one.[33] With characteristic moderation, he avoided taking a clear stance on the priority of language or thought, preferring simply to show how they are correlated in the various phases of language evolution. At times, as we have seen, he came close to suggesting a decisive influence of language on the development of mental operations, but on the whole he remained faithful to the traditional Cartesian and Lockean mind à language direction of explanation. He avoided the nominalist and materialist positions that authors like Tooke sought to develop from Locke’s observation that names of mental operations are derived from names for sensory and physical operations. Similarly, although he pointed out the complexity of the mental operations required for discerning qualities and relations, he avoided the ontological issue of whether they exist ‘out there’ or are simply creations of our mental (and linguistic) efforts, and – as far as substances are concerned – he assumed, contra Locke, the former.

            Smith’s thesis of the originality and epistemic-semiotic primacy of the impersonal verb, which corresponds to a whole event, has been hailed by Land (1974: 87) as a significant step in both logic and linguistics, for it emphasizes the proposition as the basic unit of form and meaning. Arnauld, in the Port Royal Grammar, had posited an invariable Subject-Copula-Attribute syntactic ‘deep structure’, from which all ‘surface structures’ are derived. On this view, shorter sentences would be the result of a process of abbreviation or ellipsis. Contrary to this, just as he had rejected the crypto-analyticity of morphologically represented relations, Smith (FoL 30, LRBL 217; EW 241) rejects Sanchez’s – an influential 16th century grammarian –  suggestion that even in impersonal verbs there is an underlying subject (so that pluit results by ellipsis from pluvia pluit). He also rejects the Port Royal formula on the grounds that it presupposes a logical sophistication the language creators could not have, including the ability to use the copula, “the most abstract and metaphysical of all verbs” (FoL 34, LRBL 221; EW 246). On epistemological grounds, Smith seems convinced that it is much more natural, both diachronically and synchronically, to take non-analyzed propositions that express whole events, rather than an abstract analytical formula, as the primitive linguistic structure from which the others are derived. However, Smith didn’t develop farther this insight,[34] so that it is a bit far fetched to see in it the seeds of propositional logic, not to say of the detachment of the study of linguistic form from the study of cognition (Land 1974: 93).

            However important FoL may have been, it seems to me that the broader significance of Smith’s ‘theory’ of language is to be gathered from what transpires in his other works, as I have tried to show.[35] As far as the mind-language issue is concerned, he went beyond the prevalent approach of considering mainly their cognitive and structural correspondence and interdependence. Thereby he extended considerably the range of ‘mental operations’ and linguistic phenomena to be considered within the tradition that put the mind-language relationship at the center of its concerns.

Unlike the dominant tendency among the rhetoricians of his time, who focused on the ‘ornamental’ role of language, Smith viewed style as a comprehensive mirror of the mind, and redefined beauty in terms of this mirroring relation (LRBL 34-35). This provided the guidelines for a kind of criticism that could appreciate Swift’s ‘plain’ style and be contemptuous towards Shaftesbury’s “dungeon of metaphorical obscurity” (LRBL 8). It also permitted to recover the forgotten Aristotelian rhetorical tradition that emphasizes persuasion as a main discursive function (HLM; EW 124) – a tradition that distinguishes between a ‘rhetorical’ discourse that “endeavours by all means to persuade” (LRBL 62; italics mine) and a ‘didactic’ one that “endeavours to persuade us only so far as the strength of the argument is convincing” (ibid.).[36] In terms of Smith’s functional approach, the ‘propriety’ of discourse had to be assessed in terms of the speaker’s ‘cast of mind’, his current interest and communicative intentions, the hearer’s capacity of understanding, and – in general – the context of use. He thus naturally moved from an exclusive interest in language structure and semantics to an awareness of the parameters relevant for language use or pragmatics.

            One of his disciples, Dugald Stewart, may have picked up his own sensitivity for the pragmatic aspects of language from the incipient theory of language use he detected in Smith’s writings.[37] Stewart observed that context-dependence or ‘ambiguity’, and consequently the need for a constant interpretive effort, is an ineradicable feature of language use in all the sciences, except mathematics. For this reason, whereas in mathematics “the solution of [a] problem may be reduced to something resembling the operation of a mill”, in the other sciences (especially the moral sciences) “the words about which our reasonings are conversant, admit, more or less, of different shades of meaning; and it is only by considering attentively the relation in which they stand to the immediate context, that the precise idea of the author in any particular instance is to be ascertained” (Stewart 1854/60: III, 106; in Land 1974: 113). This fact puts upon the mind an extra burden, possibly even a new set of pragmatically required mental operations: “the mind must necessarily carry on, along with the logical deduction expressed in words, another logical process of a far nicer and more difficult nature, -- that of fixing, with a rapidity which escapes our memory, the precise sense of every word which is ambiguous, by the relation in which it stands to the general scope of the argument” (ibid., p. 107; in Land 1974:121). These acute observations cannot but remind one of Smith’s distinction between the algorithmic and the heuristic types of linguistic (and moral) rules. It is a pity that Smith himself did not further develop his pragmatic insight along similar lines.

            He might also have elaborated upon other aspects of language use that his moral philosophy in general and his ethics of communication in particular imply. I have in mind especially the linguistic and mental operations aspects of the activity of moral criticism, embodied in the central Smithian figure of the ‘impartial spectator’. But I cannot develop this point here. Nor can I analyze, unfortunately, the methodological significance of his linguistic work. Let me only mention two points. First, the fact that FoL was hailed as a “very beautiful specimen” of  “theoretical or conjectural history”, which “may be traced in all his [Smith’s] different works” (Stewart 1793: 33, 36. 37). Like other theoretical systems, it is an “invention of the imagination, to connect together the otherwise disjointed and discordant phaenomena of nature” (HA, Section IV; EW 108). When such a connecting job succeeds, a scientific system – just like a linguistic system – can be compared to a machine (HA, Section IV; EW 66). Nevertheless, and this is my second point, this machine-like model is not the only model available in Smith’s repertoire, pace his Newtonianism. As we have seen, the language-machine analogy in FoL has a coda, which is absent from its counterpart in HA. Since languages are multi-functional, simplification and reduction in their case, unlike in machines, breeds imperfection, rather than perfection. Since language use is inherently context-dependent, the variability of the circumstances of use and of the characters and intentions of the users cannot be abstracted away nor accounted for in terms of a set of simple algorithmic rules. At most only a small part of language can be treated in this machine-like way, the rest requiring a systematization in terms of imprecise heuristic rules. And even for this small structural part, FoL didn’t succeed in accounting for the conjectural evolution of language structure in terms of a single criterion.

The expansion of his linguistic interests towards a more comprehensive treatment of language required him to account for an unexpected language-related ‘wealth of notions’.[38] As a result he had to depart from the esprit de système that had inspired his original plan for FoL.[39] This seems to have suggested to Smith a model of ‘system’ that could function as an alternative for the mechanical model, if not in the natural sciences, at least in the ‘moral’ ones. Although this model does not supersede the Cartesian-Newtonian one, it seeks “something more than mere tidiness and intellectual coherence” – as pointed out by the editors of LRBJ (Introduction, p. 35). It focuses not on the components of a whole but on the whole itself; it stresses the “variety of the parts” rather than their uniformity, as well as the harmonious whole that can result from such a variety (IA, Part II; EW 168); and it insists not on the sequential (axiomatic-like) mode of presentation but on the non-linear, parallel and simultaneous way of ‘connecting’ the parts to obtain a complete effect. In such a model, different approaches may complement each other and do a better descriptive and explanatory job than the reductionistic temptation to rely exclusively upon one set of assumptions. After all, in so far as different theories “are founded upon natural principles, they are all of them in some measure in the right” (TMS 7.1.1). Perhaps it is to this sort of critical-eclectic, non-reductionistic alternative that Dugald Stewart is alluding when he says that “when different theoretical histories are proposed by different writers, [they] are not always to be understood as standing in opposition to each other … for human affairs never exhibit, in any two instances, a perfect uniformity” (Stewart 1793: 37; my italics). At any rate, if the two different methodological models here sketched do indeed coexist in Smith, this might help at least to explain how authors who considered themselves to be followers of Smith, e.g. Malthus and Ricardo, could diverge so much not only in method but also in their conception of what the proper ‘style’ – i.e., the ‘spirit and mind’ – of a social science should be (see Cremaschi and Dascal 1996, 1998).




Aarsleff, Hans. 1967. The Study of Language in England, 1780-1860. Princeton: Princeton University Press.

Aarsleff, Hans. 1982. From Locke to Saussure: Essays on the Study of Language and Intellectual History. Minneapolis: University of Minnesota Press.

Arnauld, Antoine and Lancelot, Claude. 1676. Grammaire Générale et Raisonnée, 3rd ed. Paris: Pierre le Petit.

Arnauld, Antoine and Nicole, Pierre. 1683. La Logique ou l’Art de Penser, 5th ed. [L. Marin, ed., Paris: Flammarion, 1970].

Auroux, Sylvain. 1991. La tradition rationaliste dans la philosophie du language. In Dascal et al. (eds.), vol. 1, pp. 184-197.

Berkeley, George. 1710. A Treatise Concerning the Principles of Human Knowledge [Quoted from The English Philosophers from Bacon to Mill, ed. E.A. Burtt, New York, The Modern Library, 1939].

Berkeley, G. 1948/57. The Works of George Berkeley, ed. by A.A. Luce and T.E. Jessop.

Condillac, Etienne Bonnot de. 1746. Essai sur l’origine des connoissances humaines. In Oeuvres Philosophiques de Condillac, edited by G. le Roy. Paris: Presses Universitaires de France: 1947 (volume 2).

Cremaschi, Sergio. 2000. Les lumières écossaises et le roman philosophique de Descartes. In Y. Senderowicz and Y. Wahl (eds.), Descartes: Reception and Disenchantment. Tel Aviv: University Publishing Projects, pp. 65-88.

Cremaschi, Sergio and Dascal, Marcelo. 1996. Malthus and Ricardo on economic methodology. History of Political Economy 1996(3): 475-511.

Cremaschi, Sergio and Dascal, Marcelo. 1998. Malthus and Ricardo: Two styles for economic theory. Science in Context 11(2): 229-256.

Crosland, Maurice P. 1978. Historical Study in the Language of Chemistry. New York: Dover.

Dascal, Marcelo. 1978. Aporia and Theoria: Rousseau on language and thought. Revue Internationale de Philosophie 124/125: 214-237.

Dascal, Marcelo. 1983. Signs and cognitive processes: Notes for a chapter in the history of semiotics. In A. Eschbach and J. Trabant (eds.), History of Semiotics. Amsterdam: Benjamins, pp. 169-190.

Dascal, Marcelo. 1987. Leibniz: Language, Signs, and Thought. Amsterdam: John Benjamins.

Dascal, Marcelo. 1990. Leibniz on particles: Linguistic form and comparatism. In T. de Mauro and L. Formigari (eds.), Leibniz, Humboldt, and the Origins of Comparativism. Amsterdam: Benjamins, pp. 31-60.

Dascal, Marcelo. 1992. On the pragmatic structure of conversation. In J. Searle et al., (On) Searle on Conversation. Amsterdam: John Benjamins, pp. 35-56.

Dascal, Marcelo. 1993. One Adam and many cultures: The role of political pluralism in the best of possible worlds. In M. Dascal and E. Yakira (eds.), Leibniz and Adam. Tel Aviv: University Publishing Projects, pp. 387-409.

Dascal, Marcelo. 1994. Lenguaje y conocimiento en la filosofía moderna. In E. de Olaso (ed.), Del Renacimiento a la Ilustración I (= Enciclopedia Ibero Americana de Filosofía 6). Madrid: Trotta, pp. 15-51.

Dascal, Marcelo. 1995. Strategies of dispute and ethics: Du Tort and La Place d’Autruy. In Proceedings of the VI. Internationalen Leibniz-Kongress, pp. 108-116.

Dascal, Marcelo. 1996. The dispute on the primacy of thinking or speaking. In Dascal et al. (eds.), vol. 2, pp. 1024-1041.

Dascal, Marcelo. 1998. Language in the mind’s house. Leibniz Society Review 8: 1-24.

Dascal, Marcelo. Forthcoming. Hobbes’s challenge.

Dascal, Marcelo, Gerhardus, Dietrich, Lorenz, Kuno, and Meggle, Georg (eds.). 1992/1996. Philosophy of Language – An International Handbook of Contemporary Research, 2 volumes. Berlin: De Gruyter.

Dascal, Marcelo and Senderowicz, Yaron. 1992. How pure is pure reason? Language, empirical concepts, and empirical laws in Kant’s theory of knowledge. Histoire, Epistemologie, Langage 14: 129-152.

Formigari, Lia, 1988. Language and experience in Seventeenth-century British philosophy. Amsterdam: John Benjamins.

Formigari, Lia. 1992. The empiricist tradition in the philosophy of language. In Dascal et al. (eds.), vol. 1, pp. 175-184.

Formigari, Lia. 1993. Signs, Science and Politics: Philosophies of Language in Europe 1700-1830. Amsterdam: John Benjamins.

Foucault, Michel. 1966. Les Mots et les Choses: Une Archéologie des Sciences Humaines. Paris: Gallimard.

Girard, Gabriel. 1747. Les vrais principes de la langue françoise. Paris: Le Breton [reprinted, with an introduction by P. Swiggers, by Librairie Droz, Genève, 1982].

Goldschmidt, Victor. 1974. Anthropologie et politique. Les principes du système de Rousseau. Paris: Vrin.

Grice, Paul. 1989. Studies in the Way of Words. Cambridge, Mass.: Harvard University Press.

Haakonssen, K. 1998. “Divine/natural law theories in ethics”. In D. Garber and M. Ayers (eds.), The Cambridge History of Seventeenth-Century Philosophy. Cambridge: Cambridge University Press.

Harris, Roy (ed.). 1994. Comparativist Controversies. London: Routledge/Thoemmes Press.

Hewes, Gordon H. 1975. Language Origins: A Bibliography. The Hague: Mouton.

Hewes, Gordon H. 1996. Disputes on the origin of language. In Dascal et al. (eds.), vol. 2, pp. 929-943.

Joly, A. and Stéfanini, J. (eds.). 1977. La Grammaire Générale des Modistes aux Idéologues. Lille: Presses Universitaires de Lille.

Knowlson, James. 1975. Universal Language Schemes in England and France 1600-1800. Toronto: University of Toronto Press.

Land, Stephen K. 1974. From Signs to Propositions: The Concept of Form in Eighteenth-Century Semantic Theory. London: Longman.

Leibniz, Gottfried Wilhelm.1879 Die philosophischen Schriften von G.W. Leibniz, edited by C. I. Gerhardt, 7 volumes (Hildesheim: G. Olms, reprint, 1965). (Abbreviated by ‘GP’ followed by volume and page numbers).

Locke, John. 1690. An Essay Concerning Human Understanding [J.W. Yolton, editor. Londonand New York: Everyman’s Library, 1967].

Quine, William van Orman. 1969. Ontological Relativity and Other Essays. New York: Columbia University Press.

Reddy, M. 1979. “The conduit metaphor”. In A. Ortony (ed.), Metaphor and Thought. Cambridge: Cambridge University Press, pp. 284-324.

Robinson, Daniel N. 186. The Scottish Enlightenment and its mixed bequests. Journal of the History of the Behavioural Sciences ??: 171-177.

Rousseau, Jean-Jacques. 1755. Discours sur cette question proposée par l’académie de Dijon: Quelle est l’origine de l’inégalité parmi les hommes et si elle est autorisée par la loi naturelle [in J-J. Rousseau, Du Contrat Social; Paris, Garnier, 1954].

Ryle, Gilbert. 1932. Systematically misleading expressions. Proceedings of the Aristotelian Society 31: 139-170.

Smith, Adam

            Corr.                Correspondence

            EW                  Early Writings (ed. J. R. Lindgren)

            ES                    Of the External Senses (in EW)

            FoL                  Formation of Languages [in EW and LRBL]

            HA                   History of Astronomy [in EW]

            HLM                History of Ancient Logics and Metaphysics [in EW]

            IA                    The Imitative Arts [in EW]

            LJ                    Lectures on Jurisprudence (A and B)

            LRBL               Lectures on Rhetoric and Belles Lettres

            TMS                The Theory of Moral Sentiments

            WN                 The Wealth of Nations          

Stam, James H. 1976. Inquiries into the Origin of Language: The Fate of a Question. New York: Harper & Row.

Stewart, Dugald. 1793. Account of the Life and Writings of Adam Smith LL.D. In D. Stewart (1858).

Stewart, Dugald. 1854-1860. The Collected Works of Dugald Stewart, edited by W. Hamilton, 11 vols. Edinburgh.

Stewart, Dugald. 1858. Biographical Memoirs of Adam Smith, William Robertson and Thomas Reid (ed. by W. Hamilton), with a Memoir of Dugald Stewart by J. Veitch. Edinburgh: Thomas Constable.

Wilkins, John. 1668. An Essay towards a Real Character and a Philosophical Language. London: Gelibrand and John Martin.



[1] Stewart (1793: 74). In spite of Smith’s unwillingness to publish the lecture course, it occupies a central place in his reflections on language, whose broader scope it reveals – as we shall see.

[2] In contrast to TMS, Smith didn’t call his monograph a ‘theory’, preferring the modest ‘Considerations’.

[3] The 17th century displayed also other interests  in language, not directly connected with epistemic issues. But the latter gradually overshadowed and even influenced these other concerns. For a survey, see Dascal (1994, 1996) and the bibliography therein. For specific authors, issues, and currents, see Aarsleff  (1967, 1982), Auroux (1992), Dascal (1978, 1982, 1987, 1990, 1998, Forthcoming), Formigari (1988, 1992), Knowlson (1975).

[4] Notice the plural ‘languages’ (Dascal 1990, 1998).

[5] Among the inferences he considers helpful is the remark – often relied upon in the 18th century – that words for “notions quite removed from sense have their rise thence, and from obvious sensible ideas are … made to stand for ideas that come not under the cognizance of our senses” (3.1.5). His prime example are terms referring to mental operations such as ‘imagine’, ‘apprehend’, ‘comprehend’, ‘conceive’, as well as to the mind itself (‘spirit’).

[6] This program’s pivot was the proposition (rather than individual words, on which Locke’s semantics focused). Its ‘founding’ symbiotic works were Arnauld and Lancelot (1676) and Arnauld and Nicole (1683).

[7] In 1668 the Royal Society of London created a committee to examine the “philosophical language” it had commissioned one of its founding members, John Wilkins, to design. In spite of Wilkins’s (1668) effort and ingenuity, his ‘Real Character’ was never used. On universal languages in the 17th and 18th centuries, see Pombo (1987) and Knowlson (1975).

[8] For example, universal language and general grammar. The latter, after reaching its peak in mid-century, is replaced by a concern with the specificity of each language (Joly and Stefanini 1977).

[9] In this, as in other domains, continuity between the two centuries is the rule (Haakonssen 1998: 1350).

[10] For instance, Berkeley’s ‘nominalist’ theory of generality, where a general noun grants to a collection of particular ideas the stability and coherence required for its use in reasoning (Principles, Introduction #18; p. 519). Along this nominalistic path, Berkeley – comparing words to algebrical signs – points out that “they should not, every time they are used, excite the ideas they signify in our minds” (Alcyphron,  Works, III, 292; Principles, Introduction #19; p. 520).

[11] Which differs from the association of ideas (Aarsleff 1982: 29).

[12] Many important thinkers of the 18th century, in addition to Condillac and Smith, wrote on the ‘origin of language’ in this vein (e.g., Rousseau, Maupertuis, Turgot, Herder), and it has been suggested that “detailed attention to this discussion ought to be one of the best means to gain true understanding of the Enlightenment” (F. Venturi, La Jeunesse de Diderot, 1939, pp. 238-240; in Aarsleff 1982: 199). Smith was familiar with this discussion. In his letter to the Edinburgh Review (1755) he reviewed the French Encyclopédie, that devoted many articles to language (EW 18). He possessed copies of Condillac’s Essai and Traité des Sensations and, while in Paris, he held many conversations (and later also corresponded) with Turgot (Stewart 1793: 47-48).

[13] John Horne Tooke, Diversions of Purley (1786), II, 51b (quoted by Aarsleff (1967: 13).

[14] For example, by Hutcheson: “It is an easy matter for men to assert any thing in words; but our own hearts must decide the matter” (An Inquiry into the Origins of our Ideas of Beauty and Virtue, 1725); “Morality does not consist in significancy” (An Essay on the Nature and Conduct of the Passions and Affections, 1728). Both quoted in Formigari (1993: 16-17).

[15] The view, advanced by Condillac and developed by Rousseau in his Essai sur l’origine des langues, that language at its origin is ‘metaphorical’, with its implication that primitive men were ‘natural poets’, prompted the need to distinguish between such a natural poetry from ‘poetry proper’, which could not (e.g., for thinkers such as Shaftesbury) be reduced to its alleged natural linguistic origins.

[16] Letter to George Baird, 7 February 1763 (Corr.  87-88). William Ward’s An Essay on Grammar, as it may be applied to the English language was published in London in 1765.

[17] FoL was published in the first volume of  the London series Philological Miscellany (1761) and later on appended without significant revisions to the third edition (1767) of TMS. It is reprinted in both EW and LRBL, but not in the Glasgow edition of TMS, contrary to Smith’s express instructions.

[18]It is this application of the name of an individual to a great multitude of objects … that seems originally to have given occasion to the formation of those classes”; “what constitutes a species is merely a number of objects, bearing a certain degree of resemblance to one another, and on that account denominated by a single appellation” (FoL 2, LRBL 204-205; EW227).

[19] On ‘natural kind’ and ‘similarity space’, see Quine (1969: chapter 5).

[20] Rousseau’s problem is presented in a series of aporetic circles exploring the difficulties of explaining how man could acquire cognitive abilities beyond those of the animals. “The more one meditates on this subject – he says (1755: 50) – the bigger is the distance from pure sensations to the simple bits of knowledge”. Communication is introduced as a possible way out: “It is impossible to conceive how a man would be able, by his own forces, without the assistance of communication … to overcome such an enormous interval” (ibid.). But language, the tool of communication that alone would be able to overcome the gap, raises a “new and worse difficulty”, namely: “if men needed speech in order to learn to think, they were much more in need of thoughts in order to discover the art of speech” (p. 52). The reason is that “general ideas can penetrate the mind only with the help of words, and the understanding only grasps them through propositions” (p. 54). Assuming – as Smith does – that general ideas arise independently of language, so that green can be “from the very first start” a general word (FoL 7, LRBL 206; EW 229), or – as Condillac suggests (Essai 1.4.25-26) – that both evolve together, is hardly a solution to the problem (Dascal 1978).

[21] In the 18th century ‘metaphysics’ was a term generally used to refer to the theory of mind (cf. Goldschmidt 1974: 267-268). But Smith also uses ‘metaphysical’ in a more traditional sense, both in FoL (as when he says that “a relation is, in itself, a more metaphysical object than a quality” (FoL 12, LRBL 209; EW 232) and elsewhere (e.g. in HLM).

[22] “To express relation, therefore, by a variation in the name of the co-relative object, requiring neither abstraction, nor generalization, nor comparison of any kind, would, at first, be much more natural and easy, than to express it by those general words called prepositions, of which the first invention must have demanded some degree of all those operations” (FoL 17, LRBL 211; EW 235).

[23] This does not rule out the possibility that the two types of language evolve simultaneously. In fact, simultaneous evolution seems to be required by his later explanation of the rise of ‘mixed’ modern languages.

[24] Even the analysis of sensations that “seem capable of a certain composition and decomposition”, such as taste or sound, require the skill of a cook or an experienced ear (ES; EW 196). 

[25] This program culminated, via Condillac’s claim that science is nothing but un langage bien fait, in Lavoisier’s reform of chemical nomenclature (see Crosland 1978: Part II, Chap. 5).

[26] Letter to Baird, 7 February 1763 (Corr. 88).

[27] Smith does not favor the use of metaphor per se, but only when it is appropriate (LRBL 8, 26, 34). His positive valuation of iconicity and analogy in FoL can be compared, on the one hand,  to Rousseau’s thesis that language in its origin is metaphorical and, on the other, to Leibniz’s lex expressionum, according to which a relation of ‘expression’ obtains between two things “when there is a constant and ordered relation between what can be said of both” (Letter to Arnauld, September 1687; GP II 112), i.e. where no similitude between them is required other than structural analogy (“What is an idea”; GP VII 263). Neither Rousseau’s nor Leibniz’s relevant texts could have been known to Smith. Condillac’s langage d’action, on the other hand, was based on a metonymic relation of representation. Vico’s ideas on the role of metaphor and poetry in the formation and evolution of language had their counterparts in Britain at the beginning of the 18th century (Formigari 1993: 24ff.), and may have influenced Smith.

[28] From a report by John Millar, who had attended Smith’s lectures (Stewart 1793: 11).

[29] George Campbell’s opening sentence in The Philosophy of Rhetoric (1776). Quoted in the editor’s Introduction to LRBJ, p. 36.

[30] On the two first points, see LRBL 6-7, 24, 55. A good example of the third is Smith’s rule: “Your Sentence or Phrase [should] never drag a Tail” (LRBL 24). Hedging or qualification at the beginnings of a sentence is O.K. because it suggests accuracy, whereas at the end it suggests “a kind of Retractation and bears the appearance of confusion or disingenueity” (ibid.). Suggestions such as these belong to a family of non-semantic inferences whose study has been one of the mainstays of pragmatics ever since Grice’s groundbreaking articles on the ‘logic of conversation’ (collected in Grice 1989).

[31] For Leibniz, this principle was of fundamental ethical, social, and communicative importance. His notion of tolerance and his attempts to reconcile opposing views were largely based on it (Dascal 1993, 1995). Unlike Smith, who tended to focus on its ‘altruistic’ reading  (TMS,, Leibniz acknowledged also a selfish ‘strategical’ use of the principle.

[32] For instance, character and circumstances vary so much that virtually all the terms of the retribution rule (‘equal”, ‘superior’, ‘value’, ‘service’, ‘as soon as’, ‘you can’) can receive widely different values.As I write this, the radio reports that a lady legated in his will the sum of 60 thousand dollars to a group of chimpanzees in the local zoo. She explained her decision by mentioning the happiness the apes gave to her terminally ill granddaughter. This is a fine example of retribution, which illustrates the variability of the parameters of this rule well beyond what Smith would have imagined.

[33] Yet, in so far as strictly linguistic matters are concerned, Smith was “more directly influential than Condillac on later philologists: Monboddo draws heavily upon the Considerations and Smith anticipates A. W. Schlegel’s important distinction between analytic and synthetic structures” (Land 1974: 80).

[34] Elsewhere he even took for granted the universality of the Port Royal formula: “in every [phrase] there are generally three principall parts or terms because every Judgement of the humane mind must comprehend two Ideas between which we declare that relation subsists or does not subsist; concerning two of these we affirm something or other, and the third connects them together and expresses the affirmation” (LRBL 17).

[35] For lack of space I have not gone into his few remarks on language in WN nor into the many connections with language one can find in LJ.

[36] The urge to persuade others is so strong and basic, according to Smith, that he even suggests that it may be “the instinct upon which is founded the faculty of speech” (TMS 7.4.25). The connection of this urge with ambition, with the desire of “leading and directing other people”, however, suggests that the kind of persuasion a man motivated by this desire would undertake would be of the ‘rhetorical’, rather than of the ‘didactic’, kind.

[37] Stewart is customarily perceived as more of a follower of Reid than of Smith (Harris 1994: vii; Aarsleff 1967: 101-102), but I think some of his most original linguistic ideas can be traced back to Smith. Contrary to what Harris suggests (ibid.), Stewart did not detach himself entirely from the language-mind research programme. He only condemned its abuses, as exemplified mainly by Tooke, and redirected the program’s agenda to previously unexplored aspects of language and mind, as will be shown below.

[38] The phrase ‘wealth of notions’ is used by Stam (1976: 39ff.) to refer to Smith’s “capitalist ideology” as revealed in FoL by the “nominalization of all actions and reification of all relations” and by the contradictions of his analysis of the first person pronoun, which makes “the self” both “one out of an infinity of particular objects” and “the most abstract and metaphysical of all ideas”. It seems to me that his Marxist parti pris didn’t let Stam realize how wealthy Smith’s reflections on language are.

[39] Smith praises French talent for method and systematicity, but points out that this talent, most beautifully exemplified by Descartes, did not prevent Descartes’s physics from being “now … almost universally exploded” (EW 18). For a comparative study of the meaning of ‘Cartesian’ for several Scottish thinkers, including Smith, see Cremaschi (2000).