SOURCE: "Ockham on Mental Language" in Historical Foundations of Cognitive Science, edited by J. C. Smith, Kluwer Academic Publishers, 1990, pp. 53-71.

[In the following essay, Normore describes Ockham's concept of mental language and its purposes, highlighting some problems and difficulties associated with it.]

Thanks largely to the work of Noam Chomsky, we have witnessed over the last thirty years a revival of interest in two closely related ideas: that there is a universal grammar, a set of structural features common to every human language, and that the exploration of this grammar is, in part, an exploration of the structure of thought.

Fourteenth century grammarians and philosophers were also interested in this complex of questions, and debate about them raged as vigorously then as now. One tradition in this debate grew out of thirteenth century terminist logic and seems to have been given a distinctive shape by William Ockham. This tradition posited a fully-fledged language of thought common to all rational beings and prior to all linguistic convention. In this essay I will attempt to outline Ockham's account of this mental language, to consider some fourteenth century objections which lead to the refinement of the account by others in the fourteenth century, and finally to suggest that Ockham's approach has something to contribute to current debate about the relationship between the theory of meaning and any language of thought.

At the very beginning of his Summa Logicae Ockham claims that there are three distinct types of language: written, spoken, and mental. He insists that written and spoken language are distinct in kind and that there is a type of language whose terms are concepts and which exists only in the mind.¹

Ockham's mental language plays several distinct roles within his philosophy. On the one hand, mental language figures crucially in the semantics of spoken and written language. On the other hand, mental language is a fully articulated language which is suited to be spoken by natural telepaths and is spoken by the angels. These two kinds of role require very different features of mental language, features which, as we shall see, sometimes pull its structure in opposite directions.

1. Mental Language as Semantics

The key terms in Ockham's employment of mental language as a semantics are 'signification' and 'subordination'. 'Signification' is a primitive term within Ockham's semantics, though he explains its use by saying that a term signifies an object if it can be truly predicated of a pronoun picking out that object.² 'Human' signifies you because "That is a human" (pointing at you) is true. Ockham is one of the people later called "nominalistae," and his "nominalism" is reflected in his account of signification. On Ockham's view all signification is of particular concrete objects, concepts, and words. A singular term like 'Socrates' signifies a single thing, Socrates; a common noun like 'animal' signifies all the animals.

In the passage at the beginning of the Summa Logicae in which mental language is introduced Ockham claims that while terms of spoken language do not (usually) serve as signs of terms in mental language but rather serve as signs of things in the world ('human' signifies human beings, not the concept human being), nevertheless spoken terms signify what they do because they are subordinated to mental terms. If a mental term were to change its signification, the subordinated spoken term would as well. I will speak of the mental expression to which a spoken expression is subordinated as its corresponding mental expression.

The correspondence between spoken language and mental language induced by subordination is not simply a one-one mapping of term onto term. Ockham makes this clear in a crucial passage in his fifth Quod-libet:

I say that just as among spoken and written terms some are nouns, some verbs, some pronouns, some participles, some conjunctions, and some prepositions, so it is in the mind as is evident from this consideration that to every true or false spoken expression (oratio) there corresponds some mental expression made up of concepts. Therefore just as those parts of spoken sentences which are imposed to signify things because they are needed for sig-nification or expression… are distinct parts, so those parts of mental sentences corresponding to words are distinct [if needed] to make distinct sen-tences true or false.³

Here Ockham asserts that mental language has distinct terms corresponding to those terms of spoken language which cannot be substituted one for another salva veritate. This is amplified by his claim in Sum-ma Logicae I, c.13, that strictly speaking no term or concept of mental language can be equivocal. This, as Ockham explains, is because "a word is equivocal if, in signifying different things, it is a sign subordinated to several rather than to one concept or intention of the soul."⁴ On Ockham's account, then, to be equivocal requires being subordinated and so is confined to spoken and written speech. A mental term has, we might say, a single sense. Ockham amplifies this by remarking that "a term can be equivocal by the very fact that it can function as different parts of speech, i.e., as both a noun and a participle, a noun and an adverb, etc."⁵

These constraints on the structure of mental terms taken in conjunction with the subordination relation between spoken and mental terms suggest that mental language functions very much like a theory of sense for spoken language. Indeed, we can summarize many of the semantic roles mental language plays in four theses:

1. The signification of a spoken term is a function of that of the corresponding mental term.
2. The truth-value of a sentence is a function of the truth-value of the corresponding mental sentence.
3. If two expressions have the same corresponding mental expression then they have the same signification.
4. Expressions with the same corresponding mental expression are everywhere interchangeable salva veri-tate.

If we replace 'mental expression' by 'customary sense' and 'signification' by 'customary reference' in these claims, we obtain the theses of Frege's semantics often thought of as crucial for determining the relationship between sense and reference. Indeed, we can assert that the significations of spoken terms and the truth-values of spoken sentences are completely fixed by the corresponding mental terms and the subordination relation.

2. Mental Language and Epistemology

Mental language is fitted for its semantic roles not only by the subordination relation which connects it to spoken language but by its structure and, in particular, by the fact that it contains neither synonyms nor equivocal expressions. As John Trentman pointed out in his pioneering work on mental language, these structural features also make mental language look like the ideal languages of early twentieth century analytic philosophy.⁶ Both contain just those distinctions relevant to the truth-values of sentences. Moreover, just as the structure of an ideal language was thought to be the proper study of the logician, so for Ockham, as we shall see, the proper study of the logician is the structure of mental language.

It must be emphasized, however, that mental language (Mental) was understood by Ockham to be a medium of communication. If we were telepaths we would communicate in mental language, and the angels, who are telepaths, do so.⁷ To see how it could function as a medium of communication, we must take a closer look both at its structure and at its relations to the world.⁸

Spoken and written terms signify things in the world, but they do so in subordination to mental terms. Ockham distinguishes the signification of mental terms from that of spoken and written terms. He calls the former natural signification and the latter conventional signification. There is a clear sense in which, for Ockham, natural signification is primary. Indeed, as Paul Spade has pointed out, we can think of conventional signification as,

… the relative product of the subordination relation into the natural signification relation. A term conven-tionally signifies a thing x if there is some concept to which the term is subordinated and which in turn naturally signifies x.⁹

It is because the relation of conventional signification can be so defined that mental language can serve as a semantics for spoken language. Spoken language is connected to the world by its relations to the connections which mental language has to the world.

Just what are those connections? The problem of the nature of the natural signification relation is very nearly the analogue within Ockham's philosophy of the general problem of the nature of representation. It should come as no surprise, then, that it is a controversial subject whose full treatment would take us very far afield. What does seem clear is that for Ockham natural signification involves two natural relations between the mental term and what it signifies. One is a relation of similarity, while the other is a causal relation: mental terms are the natural effects in the soul of encounter with the objects which they signify. There is no doubt that Ockham thinks that both causality and similarity are involved in determining what a mental term signifies, but there are two reasons for thinking that causality is the more fundamental. First, the process of concept formation, that is, the process whereby simple mental terms are acquired, is a causal process. As Ockham explains:

This is the process. First a human is apprehended by some particular sense, then that same human is apprehended by the intellect, and from this having been apprehended there is had a general notion common to all humans. This apprehension (cognitio) is called a concept, intention, or passion, and from it existing in the intellect, the intellect at once knows that a human is something without any process of reasoning (sine discursu). Then another animal or other animals distinct from a human having been apprehended, there is elicited a general notion of all animals, and that general notion of all animals is called a passion or intention of the soul or concept common to all animals.¹⁰

Second, Ockham considers the possibility that a singular concept may be as similar to one thing as to another and explicitly claims that it is the causal connection which determines which thing the concept signifies:

I say that we have an apprehension proper to one singular thing not on account of a greater likeness to one than to another but because this intuitive apprehension is naturally caused only by the one and not by the other, and cannot be caused by the other.¹¹

So causality plays a crucial role in the acquisition of concepts, and it figures in the tying of singular concepts to the singulars of which they are concepts. Nevertheless, as Marilyn Adams has emphasized, similarity does figure prominently in Ockham's account of why concepts signify as they do. Indeed, it seems clear that a sufficient degree of natural similarity between the mental term and what it signifies is a necessary condition for signification on Ockham's view.¹²

The process of concept acquisition which Ockham mentions is central to his epistemology. Following a tradition which goes back to Scotus, Ockham posits two kinds of intellectual apprehension: intuitive cognition and abstractive cognition. An intuitive cognition is an apprehension which supports an evident judgment that what is apprehended exists. An abstractive cognition is an apprehension which does not support such a judgment.¹³ Intuitive cognition can only be of singulars, and it and the associated process of "eliciting" concepts give rise to what he calls absolute concepts. Absolute concepts are those which constitute the basic significative elements of mental language.

The fundamental distinction among the terms of any language, for Ockham, is that between categorematic and syncategorematic terms. Syncategorematic terms are those like 'and' and 'all' which do not by themselves signify; categorematic terms are those which do. Syncategorematic terms create special epistemological problems for Ockham, and I shall return to them below. First, however, there is a distinction among cate-gorematic terms, namely their division into the absolute and the connotative, which must be clarified. Ockham distinguishes absolute and connotative terms by the ways they signify:

Purely absolute names are those which do not signify one thing principally and another or even the same thing secondarily. Rather, everything which is sig-nified by the same absolute name is signified pri-marily [and]… properly speaking, such names have no definition expressing the meaning of the term (quid nominis). For, strictly speaking, a name that has a definition expressing the meaning of the term has only one such definition, and consequently no two sentences which express the meaning of such terms are so different in their parts that some part of the first sentence signifies something that is not signified by any corresponding part in the second. The meaning of an absolute term may, however, be explained in some manner by several sentences, whose respective parts do not signify the same things. Therefore, properly speaking none of these is a definition giving the meaning of the term.

… A connotative name, however, is one which signifies something primarily and something else secondarily. Such a name has, properly speaking, a definition expressing its meaning. In such a defi-nition expressing the meaning of the term it is often necessary to put one of its terms in the nominative case and something else in an oblique case.¹⁴

In this passage Ockham distinguishes primary and secondary signification. Paul Spade has suggested that we can understand this distinction if we take the basic relation of signification to be given by the claim that for a simple term signification is as explained above while a complex expression signifies exactly what is signified by its constituent non-complex categorematic terms.¹⁵ Here 'signifies' is taken more broadly than in my earlier discussion, and a complex expression will, in this broader sense, signify objects of which it cannot be truly predicated. We can distinguish primary and secondary signification within this broader sense by combining two criteria each of which has initial plausibility but each of which is insufficient by itself. The resulting distinction is that a term primarily signifies that of which it can be correctly predicated, and if it is complex then it secondarily signifies whatever is signified by those of its noncomplex constituents which are in an oblique case and whatever it signifies but does not primarily signify.

According to the text quoted above, absolute terms have only primary signification. Ockham claims that from this it follows that they have no nominal definition. His argument seems to be that any term which has a nominal definition is synonymous with its definition. A definition, being a complex expression, typically signifies some things which it does not primarily signify. (For instance a definition by genus and differentia of some species always signifies the members of the genus outside the species but does not primarily signify them.)

Since an absolute term does not signify anything which it does not primarily signify, then, if it had a nominal definition, the signification of the defining expression would not be exactly that of the expression defined. But, Ockham claims, two expressions are synonyms only if they signify the same thing in every way.¹⁶ Hence the term and its putative definition would not be synonyms after all.

Absolute terms do not have nominal definitions; connotative terms do. If mental language has no synonyms, then since a term and its nominal definition must be synonyms, there is no place in mental language for both. The terms which appear in the definition of a connotative term typically have other uses and so will usually be required for the expressive adequacy of mental language. It seems, then, though Ockham appears never to draw the conclusion explicitly, that if mental language has no synonyms it has no simple connotative terms. Each spoken connotative term and its defining expression will be subordinated to the same complex expression of the mental, and that expression will have at least as much structure as the definition. For example, in discussions of analyticity one sometimes finds the claim that 'bachelor' and 'unmarried man' are synonyms. Suppose this to be so. Then 'bachelor' and 'unmarried man' will be subordinated to the same mental terms. An unmarried man is something which is both a man and unmarried; hence if there is a mental term corresponding to 'man', one corresponding to 'unmarried', and one doing the work of conjunction, we could form a mental term corresponding to the written expression 'unmarried man'. Since mental language is a language with a grammar, it will have such a term if it has all of its components. Of course, 'unmarried' is itself logically complex; hence if there is a mental term for 'married' and one doing the work of term negation, there will be a complex term corresponding to 'unmarried'. Thus, the process of definition in spoken language has as analogue the process of composition in mental language.¹⁷

Absolute terms are the effect in the mind of causal contact with objects.¹⁸ Connotative terms can be built out of absolute terms and syncategoremata. But where do the syncategorematic terms come from? It is certainly not plausible to suppose that they are obtained by intuitive and abstractive cognition in the way absolute terms are. The world doesn't contain 'ifs' and 'cans' in the way it does pots and pans.

This is a subject on which Ockham seems to have changed his mind as he changed his mind about the nature of concepts themselves. Ockham's earlier view of concepts as we find it reported in his Reportatio, for example, had it that concepts were intentional objects, ficta in Ockham's terminology. During the period in which he held this theory of concepts in general, Ock-ham seems to have emphasized a distinction between two kinds of mental language. One was the mental language with which this paper has so far been concerned; the other was a language whose terms were the mental representations of the spoken expressions of natural language itself. Terms of the first kind of mental language (what other writers sometimes called 'proper' mental language) were, as we have seen, prior to spoken language. Terms of the second kind of mental language ('improper' in some terminologies) were pos-terior to spoken language. While he held the 'fictum' theory, Ockham seems to have held the astonishing view that all syncategorematic concepts were terms of improper mental language. He writes that:

… matters are similar with this word 'not' and with these 'per se', 'insofar as', 'if, and syncategoremata of this sort. From these words, so signifying, the intellect abstracts common concepts pre-dicable of them and imposes these concepts to signify those same things which those words signify outside [the intellect], and in the same way and from such as these it forms similar sentences having properties like those which the spoken sentences have.¹⁹

This would make hash of the claim put forward in the Summa Logicae that spoken language depends on mental language for its sense. Indeed since spoken language is arbitrary (ad placitum), it would make hash of the claim that mental language is natural. Hester Gelber has pointed out that Ockham's contemporary Walter Chatton criticized Ockham's early approach along just these lines.²⁰ Perhaps in response to Chatton's criticisms of it, Ockham abandoned the 'fictum' theory in favor of an 'adverbial' view which identified the concept with the act of conceiving itself or with a quality in the mind and seems also to have abandoned his early view of the origin of syncategoremata. But it is less clear to what view of the origin of synca-tegoremata he then turned.

One suggestion is that they are innate. According to Ockham the intellect prior to all experience is a tabula nuda. Thus it has no intellectual habits and no concepts. It does, however, have the power to perform certain mental acts. It can, for example, assent to the proposition that a whole is greater than a part if this proposition is presented to it. But syncategorematic terms just are certain operations which the mind performs on categorematic terms. They are not so much acts of thought as acts of will. Hence the mind has innately the power to perform them.

This innatist position may well have been Ockham's. But there are other possibilities. Although syncategore-mata could not be acquired by intuitive cognition and abstraction in the way categorematic terms are (because there are no syncategorematic objects), this does not establish that they cannot be acquired at all during intuitive cognition. If the objects of intuitive cognition are sufficiently complex, if, for example, one intuits not just Socrates's paleness or Socrates himself but, say, pale Socrates sitting on a fence, then it may be that the syncategorematic elements of the complex are presented in intuition.

But it seems unlikely that Ockham would allow that we have such complex intuitive cognitions. T. K. Scott has argued that he does at least sometimes hold that "what is intuited is a substance as subject of an accident, not merely an accident."²¹ Since Ockham holds that certain accidents, such as color, are real things which could exist apart from any substances, the claim that one could intuit Socrates as white would be the claim that one can intuit the inherence of whiteness in Socrates. Such an 'inherence in' would be a synca-tegorematic element of the kind in question. Ockham does sometimes speak in this vein. For example in II Sentences, q. 12-13, Ockham writes that "if someone experiences something to be white he sees the whiteness to be in it." But it is not obvious how we are to take such claims. What is clear is that Ockham thinks that our intuitive cognitions ground evident judgments that contingent objects are related as they contingently are—that Socrates is white, for example. What is not clear is whether we, strictly speaking, ever intuit that Socrates is white. Indeed, as Marilyn Adams has argued, it is not even clear that Ockham thinks we intuit substances at all.²²

Even if we cannot determine the source of syncategorematic mental terms within Ockham's philosophy, the picture which emerges so far is a remarkably simple and attractive one. On this picture causal contact with particular objects gives rise to concepts which serve simply as signs of things sufficiently like the original cause. These are the absolute mental terms. Which absolute terms there are in someone's mental vocabulary will depend on the person's causal history, and so there are differences among mental languages, but only in the lexicon. Mental expressions consist of absolute terms combined by mental syncategoremata. If the combination is itself something which could be the subject or predicate of a sentence, then it is a mental connotative term. All simple terms in mental language are absolute. All complex terms are connotative. Mental sentences, we may suppose, are composed of mental terms joined by a mental copula. Mental language has no equivocal expressions and no synonyms. Spoken expressions are subordinated to mental expressions and, though they signify directly, signify as they do because of the expressions to which they are subordinated. Since subordination is conventional, so is spoken language. Mental expressions function in many ways like Fregean senses for the expressions subordinated to them. In particular, the process of definition corresponds to the process of decomposition in the mental.

This picture can be made to do work elsewhere in Ockham's philosophy, as well. One might suggest, for example, that only those items should be admitted into ontology which are signified by some absolute term, a familiar theme for Ockham.

3. Mental Language as a Medium of Communication

The picture painted above also unifies much of Ockham's epistemology and semantics, but it is almost certainly too good to be true. Some of the difficulties begin to arise when we remember that Ockham intends mental language not only as a semantics for spoken language but further as a medium of communication in its own right. If we were telepaths we would communicate in Mental, and the angels, who supposedly are telepaths, are held in fact to do so. What must the mental be like in order to play such a role?

One thing it must be is in some sense concrete. Ockham thinks that the mind is an immaterial substantial form and that we must be prepared for the possibility that an implementation of a language in such a medium will be rather different from implementations of spoken or written language. Yet there will be questions about Mental as a language which would not arise about Mental as a semantics. One of these is a question raised by Gregory of Rimini and revived recently by Paul Spade. It is the question of whether mental sentences are made up of their parts.²³

This is the problem. Subordination is a relation between terms of spoken language and terms of mental language. Hence to each significant element of a spoken sentence there corresponds a significant mental element. But some spoken sentences contain exactly the same significant elements and yet have different truth-values. That is how it is with:

"Every human is an animal" and
"Every animal is a human"

for example. How do the mental sentences corresponding to these differ?

The written sentences above and the corresponding spoken sentences differ in word order. In a written sentence spatial configuration is significant, and in a spoken sentence temporal configuration is significant. But neither spatial nor temporal arrangement is essential to a mental sentence: not spatial arrangement simply because the mind is not a spatial object, and not temporal arrangement because Ockham believes that mental sentences can be produced all at once. Indeed, Ockham believes that the practiced logician grasps long chains of argument at once.²⁴

One solution to this problem would be to suppose mental language to be so highly inflected that no two mental sentences ever would contain exactly the same significant elements. Ockham is already committed to markers for case and tense, so why not commit him to markers for subject, predicate, etc.? This is, after all, a popular solution in the tradition which follows him.²⁵

There are two reasons why we should not. The first is that it would not solve a deeper and related problem. What, on such an account, would be the difference between simply conceiving the parts of a mental sentence and perceiving the sentence itself? How are the parts of a mental sentence bound into a whole?

Ockham is well aware of this problem. He discusses it in several places, notably in Book II of his Commentary on the Sentences and in the Introduction to his Commentary on the Perihermenias. The first discussion sets the problem within Ockham's earlier theory of concepts as ficta or intentional objects; the discussion in the Commentary on the Perihermenias sets it within his mature view that concepts are intellectiones or acts of understanding, themselves.

In II Sent., q.1 and q. 12-13, Ockham tells us how the mind understands a mental proposition composed of ficta and how such comprehension is distinguished from the understanding of the mental terms and the copula. The immediate problem under discussion is whether there has to be a separate item of knowledge (ratio cognoscendi) for each part of the sentence known. Ockham claims not, suggesting that when we have habitual knowledge of a sentence we know the whole sentence and its terms by the same habit. Moreover, even when our grasping of the sentence is for the first time we do not need to grasp the subject, predicate, and copula in separate acts.²⁶

Within the 'fictum' theory we can distinguish between the ficta, which are the subject, predicate, and copula, and the acts of 'uttering' and 'grasping' the sentence, whose objects are those ficta. So within this theory the problem of distinguishing the mere collection of ficta (the list of 'words' in the sentence) from the sentence as an ordered whole is particularly acute. Ockham's solution is to treat the act of uttering or grasping the copula as a very special act. As he says:

… the act terminating at the copula or at the concept of the copula does not terminate there absolutely but at the same time [terminates] at the subject and the predicate. Therefore the acts which terminate only at the subject and the predicate are incomplex, but the act which terminates at the copula is complex inasmuch as it terminates immediately at the whole complex; and this is called composition and divi-sion.²⁷

Within Ockham's later 'intellectio' theory of the nature of concepts there no longer is a distinction between the act of grasping something by a concept and the concept itself. There are just the acts themselves. On Ockham's version of the 'intellectio' theory the subject and predicate of a sentence are acts, presumably the acts analogous to those which on the 'fictum' theory grasped the subject and predicate. If this is so, then what on the 'fictum' theory is the act terminating at the copula is, on the 'intellectio' theory, the copula of the mental proposition. Thus the copula itself is a complex act of compounding or dividing, one which somehow makes essential use of the two "incomplex" acts which are the extremes. This makes it possible to think of a sentence (or even an argument) as a single complex mental act. As Ockham says:

… a sentence could be an act of understanding equivalent to the one whole sentence made up of distinct parts [which there would be] if [the parts] had the order they have in speech, and then there will be distinct sentences whenever there are distinct corresponding sentences in speech.²⁸

The problem with this approach is that it threatens the recursive character of mental language and so also the possibility that it could be a medium of communication. If each sentence is a distinct primitive mental act, then decoding each sentence will be a completely novel experience in which my knowledge of other acts will be of no help. Ockham tries to evade this problem by treating some acts, such as the copula on the 'intellectio' view, as complex acts in which the simpler parts are somehow represented, but exactly how this works remains a mystery.

A second difficulty with attributing to Ockham the view that mental language is so highly inflected that problems like word order never appear is that there are cases in which he could appeal to such a device but does not do so, even though the cost of not doing so is making Mental in one way equivocal. For instance, Ockham's account of supposition requires that a term as it appears in a sentence have personal supposition when it is used to stand for what it signifies, simple supposition when it is used to stand for the concept to which it is subordinated, and material supposition when it is used to stand for itself. The sentences,

1. Humans have been on the moon.
2. Humans is not a genus.
3. 'Humans' has six letters.

illustrate the three types of supposition for the term 'humans'.

In his discussion of equivocation Ockham expressly claims that, while mental terms cannot be equivocal in the sense of having two different ranges of signification, they are equivocal in the sense that they can have personal supposition in one sentence and material supposition in another.

For spoken sentences this raises no special problem, because one can presume that the intention of the speaker determines the supposition of the subject term even if it is difficult for a hearer to discover it. But for mental sentences there is a real problem because, on Ockham's view, once the concepts which make up the sentence have been specified the supposition of the subject is still underdetermined. But what could determine it?

Ockham seems never to have faced this problem. By insisting that the concepts which make up a mental sentence do underdetermine the supposition of the subject term, he rules out the solution taken by Pierre d'Ailly which maintains that the subject term of a mental sentence is always in personal supposition. So it cannot be assumed that he would rely on inflection to solve problems about the structure of Mental. And Ockham's willingness to leave equivocity about supposition in Mental raises a question about his commitment to the view that it is a thoroughly univocal language.

4. Logic and Mental Language

Besides its role as semantics for spoken language and as a medium of angelic communication, mental language serves Ockham as a medium of logical analysis. It is here, I think, that we find some of the most striking successes and the most disturbing difficulties of Ockham's approach. I will consider one of each.

One of the tasks any logician must face, and most especially a logician who works as much with natural languages as medieval logicians did, is that of providing analyses of the logical structure of arguments and their component sentences. If one holds that spoken expressions are subordinated to mental expressions and that mental language is the appropriate medium for logic, then one can formulate a rather powerful constraint on acceptable analysis: an analysis will be acceptable only if the replacement of the analysandum by the analysans preserves grammaticality in mental language. This need not imply that such replacement will preserve grammaticality in the spoken language, however.

The idea can be illustrated by seeing how an analysis along these lines treats a problem which Norman Kretz-mann has discussed.²⁹ Kretzmann draws attention to the first sophism in Richard Kilvington's Sophismata:

Socrates is whiter than Plato begins to be white.

Kilvington analyzes 'Plato begins to be white' as "Plato is not now white and immediately after now he will be white." As Kretzmann points out, Albert of Saxony in his own Sophismata objects that either the analysis is wrong or the original sophism is ill-formed. He supports this claim by pointing out that if we replace the embedded occurrence of 'Plato begins to be white' by its analysis, we get:

Socrates is whiter than Plato is now white and immediately after now he will be white.

And this is simply ungrammatical. Kretzmann suggests that Albert has the wrong side of this dispute but that it is not easy to see exactly why. My proposal is that Albert is mistaken in thinking that the correct analysis of an expression has to preserve the grammatical category of the expression. This amounts, I think, to the claim that grammatical form mirrors logical form in natural language and is refuted by theories, like Russell's theory of descriptions, in which expressions of one grammatical category are analyzed as being of another or even, as with Russellian descriptions, of no grammatical category at all. Of course, an analysis has to preserve meaning, but that, within a framework like Ockham's mental language, just amounts to the claim that analysandum and analysis be subordinated to the same mental expression. If we insist that analysis preserve grammaticality (indeed, identity) in Mental but not n spoken language, Albert's argument fails to be sound.

The suggestion just made is that the mental sentence to which a spoken sentence is subordinated wears its logical form on its sleeve, so to speak, whereas spoken sentences themselves may conceal their logical form to some degree. In mental language all logical relations are to be explicit. Mental just is the language of logic.

This view is tested by Ockham's account of the fallacies. As Hester Gelber has argued, Ockham refurbished the classification of fallacies, treating the traditional Aristotelian distinction between fallacies 'in speech' and 'outside speech' as a distinction between fallacies which occur because someone has mistaken how spoken or written expressions are related to mental expressions and those which occur because someone has mistaken the structure of the mental argument itself.³⁰ But if the logical structure of mental expressions is explicit, how could a speaker be mistaken about it?

Ockham's paradigm of a fallacy extra dictione is the fallacy of accident. As Gelber has pointed out, Ockham rejects the diagnosis of this fallacy as arising from variation in the middle term and argues that it arises from mistaking the structure of the mental argument itself. How is this possible?

Ockham argues in several places that we cannot be mistaken about certain kinds of sentences whose subject and predicate are both absolute terms.³¹ It is tempting to generalize from this to the conclusion that any mental sentence whose only categorematic terms are absolute terms would, since its logical structure is explicit, be transparent to the thinker. If, as I have argued above, all mental sentences contain as simple components only absolute terms and synca-tegoremata, then on this view all mental sentences would be transparent and there would be no fallacies extra dictione.

The key to this puzzle is the assumption that the structure which is explicit in a sentence is manifest to the thinker. Ockham claims evidence only for certain categorical sentences involving only absolute terms, and there is no reason to suppose that he thought bolder claims would be true. Moreover, Ockham does not in general hold that our mental acts are evident to us. It is perfectly plausible, then, to suppose that fallacies arise because thinkers fail to notice such things as the scope of negation operators in their thoughts and so make inferences which are not warranted: what is happening in such cases is that one mental expression is being treated as if it were of a different structure than it is. Complicated mental acts are complicated, and the human mind, being the finite instrument that it is, may not react correctly to this complexity. Ockham does not explicitly endorse such a view, but it is, I think, very much in the spirit of his work.

5. Conclusion

It is always tempting and always dangerous to suggest that a long dead thinker's views have a close contemporary analogue. Nevertheless, this is precisely what I wish to suggest for Ockham on the basis of the foregoing. It seems to me that Ockham's approach is being rediscovered at this time, and that many of the problems and theoretical devices of the fourteenth century should be and are being considered again.

Perhaps the contemporary thinker whose work seems in many ways closest to Ockham's is Jerry Fodor. In a series of works going back at least to The Language of Thought, Fodor argues that the explanation of an organism's behavior requires determination of how external stimuli are represented internally.³² Representation requires a medium of representation, and such a medium must be symbolic in character. Thus there is no internal representation without an internal language. Moreover, this internal language cannot be a natural language. Natural languages cannot be the medium of thought because non-humans and pre-verbal children exhibit the behavior we need to postulate internal representation in order to explain. In particular, they learn. Fodor argues that learning, especially language learning, presupposes a language within which hypotheses about the material to be learned can be represented. Such a language, he argues, must be expressively as powerful as the natural language to be learned.

Thus Fodor believes that he has shown that no language can be learned unless one already has a language within which to formulate hypotheses about the meaning or use of predicates in the new language. But, as he readily admits, this does not mean that no language could be acquired without another language. On the contrary, a first language could be acquired ex nihilo provided its acquisition did not involve hypothesis formation.

This is exactly how Ockham thinks mental language is in fact acquired. As we have seen, Ockham holds that absolute categorematic concepts are either the effects in the mind of acquaintance with the world or the result of performing an apparently innate operation of comparison and abstraction on such effects. In any case they involve no hypothesis formation.

Although the situation with syncategoremata is less clear, it is, as we have also seen, plausible to attribute to Ockham the view that these are operations of the mind which arise simply from the exercise of its innate powers. If this is so then a mental language of the sort Ockham describes seems very attractive. It offers the avid psychosemanticist all the internal representational and computational character that could be wanted, while preserving for empiricism the spirit and most of the letter of the dictum that there is nothing in the mind which was not previously in the senses.

Nor is this all. Ockham stands near the beginning of a long tradition.³³ He and his successors make a number of substantive hypotheses about the structure of mental language. The mature Ockham, for example, claims that it contains neither synonyms nor equivocal terms and that it will be inflected only to the point needed for expressive adequacy. Buridan claims that it does contain simple connotative terms. Albert of Saxony claims that every categorical mental sentence contains a subject term, a predicate term, and a copula. The tradition contains discussions about the very nature of mental representation, as well. Chomsky has traced the history of some elements in his approach to language back to Descartes. I would like to suggest that it is in Ockham and his successors that we find other and perhaps more significant elements, embedded in a research program which spans much of the later Middle Ages and which may well turn out to have more than merely historical significance.

Notes

¹ Cf. W. Ockham, Summa Logicae (hereafter S.L.) I, c.l. in P. Boehner, G. Gal, S. Brown (eds.), Opera Philosophica (hereafter O.P.) I (St. Bonaventure, NY: The Franciscan Institute, 1974).

² Cf. S.L. I, c.33, in O.P.I, p. 95 ff.

³ W. Ockham, Quodlibeta V, Q.8, Art. 1, in Joseph C. Wey (ed.). Opera Theologica (hereafter O.T.) IX (St. Bonaventure, NY: The Franciscan Institute, 1980), p. 509.

⁴ W. Ockham, S.L. I, c.13, in O.P. I, p. 45.

⁵ Ibid., p. 46.

⁶ John Trentman, "Ockham on Mental," Mind LXXIX (1970), pp. 586-590.

⁷ That they do creates some interesting problems. Is it possible to lie in mental language, for example? For discussion of this question, cf. Hester Gelber, "I Cannot Tell A Lie: Hugh of Lawton's Critique of William of Ockham on Mental Language," forthcoming in Franciscan Studies 1984.

⁸ A full study of mental language as a medium of communication would require some account of what it is to 'speak' and 'hear' it. There will be some indirect remarks about 'speaking' it below, but a full treatment of the subject will not be attempted here. For a beginning consideration, cf. Hester Gelber, op cit., and Joan Gibson, The Role of Mental Language in the Philosophy of William of Ockham. Ph.D. dissertation, University of Toronto, 1976.

⁹ Paul Spade, "Ockham's Distinctions between Absolute and Connotative Terms," Vivarium 13 (1975), pp. 55-76.

¹⁰ Ockham, S.L. 111-2, cap.29, in O.P. I, p. 557, 1.14-22.

¹¹ Ockham, Quodlibeta 1, q.13, in O.T. IX, p. 74. Ockham holds the view that the causes of each thing are naturally proper to it, so that a particular thing, in this case an apprehension, could be produced naturally by only one set of causes.

¹² Cf. Marilyn M. Adams, William Ockham (Notre Dame Press, 1987), ch. 4.

¹³ Cf. Marilyn M. Adams, op. cit., ch. 13.2, for a full discussion of this and for texts. There are interesting similarities between Ockham's distinction between intuitive and abstractive cognition and B. Russell's distinction between knowledge by acquaintance and knowledge by description.

¹⁴ Ockham, S.L. I, cap. 10 (O.P. 1, pp. 35-36).

¹⁵ Paul Spade, loc. cit.

¹⁶ For example, Ockham writes in S.L. I, c.6 (O.P. I, p. 19, 1.8-16):

Loosely speaking those are called synonyms which simply signify the same in every way with the result that nothing is in any way signified through the one but that it is in the same way signified through the others even if not those using it believe them to signify the same. … It is in this second way that I intend to use this noun 'synonym' in this chapter and in many others.

¹⁷ Some of Ockham's successors do seem to have admitted simple connotative terms into Mental. For evidence that Buridan does, cf. the introduction to P. King, John Buridan's Logic: The Treatise on Supposition; The Treatise on Consequences, Synthese Historical Library, Vol. 27 (Boston: D. Reidel Publishing Co., 1985).

¹⁸ Cf. Ockham, S.L. 111-II, c.29. in O.P. I, p. 559.

¹⁹ Ockham, I Sent., d.2, q.8, in O.T. II, p. 285.

²⁰ Cf. Chatton, Reportatio I, d.3, q.2, quoted in Gelber, op, cit.

²¹ T. K. Scott, "Ockham on Evidence, Necessity, and Intuition," Journal of the History of Philosophy, Vol. 7 (1969), pp. 27-49, note 22.

²² Ockham, Sentences, Prol. q.1, in O.T. I, pp. 6, 31.

²³ Paul Spade has drawn attention to medieval discussions of this problem. Cf., for example, P. V. Spade, Peter of Ailly: Concepts and Insolubles, an annotated translation (Dordrecht: D. Reidel Publishing Co., 1980). Attend especially to his note 313.

²⁴ Ockham II Sent., q. 12-13, in O.T. V, G. Gal and R. Wood (eds.) (St. Bonaventure, NY: The Franciscan Institute, 1981), p. 279.

²⁵ For information on this tradition, cf. P. Spade, Peter of Ailly, note 217 and the references therein.

²⁶ Cf. the citation in note 24.

²⁷ Ockham, II Sent., q. 12-13, in O.T. V, p. 280.

²⁸ Ockham, Proemium to Expositio in Librum Perihermenias Aristotelis in O.P. II, E. A. Moody, G. Gal, A. Gambatese, S. Brown, P. Boehner (eds.) (St. Bona-venture, NY: The Franciscan Institute, 1978), p. 356.

²⁹ Norman Kretzmann, "Socrates is Whiter than Plato begins to be White," Nous XI (1977), pp. 3-15.

³⁰ H. Gelber, "The Fallacy of Accident and the Dictum de Omni: Late Medieval Controversy over a Reciprocal Pair," Vivarium 25 (1987).

³¹ Cf., for example, S.L. III-II, c.29, in O.P. I, p. 558.

³² J. A. Fodor, The Language of Thought (New York: T.Y. Crowell, 1975). Cf. also Psychosemantics (Cambridge: The M.I.T. Press, 1987).

³³ Among the significant figures in this tradition one may mention Jean Buridan, Buridan's student Pierre d'Ailly, and John Maior. For some discussion of Buridan's view, cf. P. King, John Buridan's Logic. For work on Pierre d'Ailly, cf. P. Spade, Peter of Ailly. This tradition was controversial, and Hester Gelber has discovered at least one opposing voice in Hugh of Lawton. Cf. H. Gelber, "I Cannot Tell A Lie." For an introduction to the later tradition, cf. E. J. Ashworth, "The Structure of Mental Language: Some Problems Discussed by Early Sixteenth Century Logicians," Vivarium 20, 1 (1982).

³⁴ This paper has benefited from many kind people over many years. Paul V. Spade helped enormously when we were both graduate students at the University of Toronto. The earliest version of the paper itself was read at Princeton University in 1977. The most recent version has benefited from ongoing argument with Hester Gelber and Peter King about whether there are simple mental connotative terms and from the very concrete help of Amelie Rorty.