UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Primacy of linguistic units Aklujkar, Vidyullekha 1973

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-UBC_1973_A1 A35_5.pdf [ 8.17MB ]
Metadata
JSON: 831-1.0100888.json
JSON-LD: 831-1.0100888-ld.json
RDF/XML (Pretty): 831-1.0100888-rdf.xml
RDF/JSON: 831-1.0100888-rdf.json
Turtle: 831-1.0100888-turtle.txt
N-Triples: 831-1.0100888-rdf-ntriples.txt
Original Record: 831-1.0100888-source.json
Full Text
831-1.0100888-fulltext.txt
Citation
831-1.0100888.ris

Full Text

PRIMACY OF LINGUISTIC UNITS by VIDYULLEKHA AKLUJKAR B.A. University of Poona, 1965 M.A. University of Poona, 1967 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in the Department of Philosophy We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA May 1 9 7 3 In p resen t i ng t h i s t h e s i s in p a r t i a l f u l f i l m e n t o f the requi rements f o r an advanced degree at the U n i v e r s i t y of B r i t i s h Columbia, I agree that the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r re ference and s tudy . I f u r t h e r agree tha t permiss ion fo r e x t e n s i v e copying of t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the Head of my Department or by h i s r e p r e s e n t a t i v e s . It i s understood that copying or p u b l i c a t i o n o f t h i s t h e s i s f o r f i n a n c i a l ga in s h a l l not be a l lowed wi thout my w r i t t e n p e r m i s s i o n . The U n i v e r s i t y o f ' B r i t i s h Columbia Vancouver 8, Canada Department o f ABSTRACT Philosophers of language engaged in an analysis of the notions of meaning, synonymy and the like regard different linguistic units as basic units of language, and they believe that their views con-tradict each other. Thus, for example, Russell and Ziff regard words as the primary meaningful units of language, Alston and Searle hold speech acts to be the fundamental units of linguistic behavior and Frege and Quine believe that sentence is the basic meaningful unit of language. In this thesis I examine different theories of linguistic primacy, discuss different relevant senses of primacy and try to find a viable criterion of linguistic primacy that is important in philoso-phical discussions. I point out that philosophers treat the question of primacy of linguistic units either learning-theoretically as Russell does, or meaning-theoretically as Alston does; and that sometimes, as in Quine's work, these two approaches coincide. The criterion of linguistic primacy important in philosophical discussions is shown to be a 'functional' criterion, namely, the essential function of the so-called primary unit of language in the analysis of meaningfulness and/ or synonymy of other linguistic units. After discussing each primacy theory individually, I show, nega-tively, why neither words, nor speech acts nor propositions can be regarded as the basic units of language in any of the senses proposed, and positively, why sentences can be so regarded. With regard to words i t is argued that they are not learning-theoretically basic to language because the first learned units of an i i individual's language are not words but sentences. Meaning-theore-tically words are shown to have meaning always and only in sentential context. With regard to speech acts, I point out that they cannot be meaning-theoretically basic, i.e., cannot be helpful in analyzing the notions of sentence-meaning or sentence-synonymy, because, the full description of a speech act necessarily includes a;specification of the meaning of the sentence uttered in the speech act. In refuting Searle's version of the speech act primacy theory I discuss the notions of speech and language and point out that language is not exhausted in speech, that language encompasses boith speech and thought, and that speaking a language has a facility aspect as well as an activity aspect. I argue that the common element shared by the speech acts and by the propositional attitudes is the meaningful sentence, and that therefore, sentence rather than speech act can be regarded as basic to both the domains of speech and thought, i.e., to language. Propositions are shown to be dependent on meaningful sentences for their own individuation, and therefore are ruled out as unnecessary in the theory of language. An attempt to regard propositions basic to language by believing that they are 'objects' of both illocutionary acts (the domain of speech) and the propositional attitudes (the domain of thought) is argued to be misguided by showing that i t hinges on a wrong analogy of the cloth and the clothed or the carried and the carrier. Finally, on the strength of the above considerations I conclude i i i that sentences are learning-theoretically primary since they are the first learned units of an individual's language, and that they are also meaning-theoretically primary in that they supply founda-tions to the meaningfulness of other linguistic units. iv CONTENTS Pages Abstract i Acknowledgement v Chapter 1. Words 9 Word-primacy Thesis: Russell's version 9 Word-primacy Thesis: Ziff's version 24 Ziff's Theory of Meaning 24 Evaluation of Ziff's Claim 38 Chapter 2. Speech Acts 53 What is a Speech Act 53 Intentions and Securing Uptake 71 Speech Act-primacy Thesis: Alston's version* 99 Speech and Language 106 Speech Act-primacy Thesis: Searle's version 114 Chapter 3. Sentences 131 Sentence-primacy Thesis: Quine's version 131 Indeterminacy: its implications for meaning 136 Sentences versus Propositions 140 Final remarks 147 Bibliography 158 V ACKNOWLEDGEMENT I am grateful to my thesis supervisor, Prof. Howard H. Jackson, for his guidance and encouragement at every stage of my research; discussions with him have often helped me see my topic in clearer perspective. I also wish to thank Prof. Thomas E. Patton and Prof. Peter Remnant for their comments and criticisms which helped me in preparing the final draft of my dissertation. I thank the above three members of my supervisory committee for suggesting many stylistic improvements in my writing. In my years at this university, I have learned much of philo-sophical worth from Prof. Jonathan F. Bennett's teaching and discus-sions. Although he was not directly involved in the present project, I take this opportunity to acknowledge my indebtedness to him. Introduction Divergent and incompatible views are expressed with regard to the nature of the basic meaningful units of language. To note only a few: Words are generally said to have meaning or to have a meaning, not utterances and not sentences. The new freedom that paraphrasis confers is our reward for recognizing that the unit of communication is the sentence and not the word....We can allow the sentences a monopoly of full 'meaning', in some sense, without denying that the meaning must be worked out.^ The unit of linguistic communication is not...the symbol, word or sentence, but rather the production or issuance of the symbol or word or sentence in the performance of the speech act....More precisely, the production or issuance of a sentence token under certain conditions is a speech act, and speech acts (of certain kinds...) are the basic or minimal units of linguistic communication.3 In the following chapters I propose to examine these and other related views e.g., those of Austin, Alston, Russell, etc. This involves roughly to set and try to answer questions such as: What is the cri-terion of primacy involved in this notion (these notions?) of the basic meaningful unit of language? Is i t the smallest acoustic measure The page references are to the reprint or edition used, and not necessarily to the first publication. *Paul Ziff , Semantic Analysis (Ithaca, New York: Cornell University Press, 1960), rept. 1967, 149 2 W.V.O. Quine, "Russell's Ontological Development," The Journal of  Philosophy, Vol. LXIII, No. 21, Nov. 1966, rept. in D.E. Pears, (Ed.) Bertrand Russell: A Collection of Critical Essays, (New York: Double-day, 1972) 292 John Searle, Speech Acts: An Essay in the Philosophy of Language (Cambridge: Cambridge University Press, 1969), rept. 1970, 16 2 of significance? More generally, is such a criterion expressible in terms of non-analyzability into further linguistic units or should i t be couched in terms of the dependence of other units on i t in accounting for their meaning? Another question is that of determining the scope of 'language' in any claim about the primacy of linguistic units. Is 'language' restricted to the 'normal* adult communication through speech or should i t include both speech and thought? Is language merely a performance of various speech acts or does i t also have a competence aspect? The study will make sense of the talk about 'the basic meaningful unit of language'. Hopefully i t should also provide some insights into the larger issues such as the relationship between linguistic competence and performance, meaning, individuation of 'speech acts', and in general, the nature of language. I start out with words. Popularly as well as philosophically words have enjoyed a certain privileged position. Popularly words are regarded as the basic units of language perhaps because they are relati-vely manageable. Words of a given language are finite in number, where-as other units such as sentences are indefinitely large (potentially infinite) in number. So words appear to be more basic in the sense that their totality is more comprehensible than that of (some) other units. Words are also likened to the building blocks of language, as more complex structures may be decomposed into them. In a dictionary, which is helpful in learning a new language, one finds a (partial, perhaps,) l i s t of the words of a language, not of other units 3 (sentences, phrases, phonemes, etc.). When we observe a child learning his first language, we say, "Today he learned his first word" or "Now he knows about half a dozen words". It is only after he has uttered his first multi-word compound, which sufficiently approximates a sig-nificant sequence, that we say, "He just said a sentence" or "Now he can compose sentences". Words appear to be the primary units of lan-guage also in the sense that we often talk of dogs or parrots as knowing certain words. But we do not attribute to them a knowledge of more complex structured units of language. So a knowledge of words, which we seem to share with some lesser animals, makes us wonder whether words are at the very bottom of language, as they are thus obviously more easily comprehended. Philosophically, the question of the primacy of words can be treated from at least two points of view. One, epistemologically, as Russell does. Two, from the point of view of a theory of meaning, as Ziff does. A third viewpoint, namely, that of a theory of the process of signification is adopted by some philosophers of India. However, since the third point of view deserves a separate and extensive treat-ment, I shall confine myself to discussing only the first two views. Russell's exposition is so clear, that i t is easy to see where he goes off the track. However, when one construes his account in the right spir i t , and straightens out the inconsistencies, one gathers some important insights with respect to the nature of words, their relation with sentences, and the respective functions of words and sentences with regard to our learning a language. Z i f f , on the other hand, does not 4 always give sufficient reasons for his assertions, and at times one has to try to construe his account so as to understand i t . This, one is forced to do in spite of his passionate protest: Virtually no such exegesis, virtually no such interpretation, virtually no such construal, is called for here....I almost certainly meant what I said and i f i t was stupid then I was being stu-pid at the time whether I meant what I said or not. (Ziff, 1967, vii) The chapter on words will help us to see in what ways words are unique, and in what sense they are riot basic to language. Of course, these points can be fully developed only after we have gone through the accounts in favor of other units, namely, speech acts and sentences. Next, we turn to speech acts. With words, their appeal for l i n -guistic primacy was obvious. With speech acts, i t is not only not so obvious, but the very notion of the primacy of a speech act is compara-tively recent in the philosophy of language. Words and sentences were acknowledged units of language from the days of Aristotle. And there were elaborate and systematic debates on the question of word-meaning primacy versus sentence-meaning primacy in the circles of Indian philosophers, as early as the 4th century A.D. But the notion of the speech acts is a latecomer in this area; for i t was Austin who first brought the notion into debate. However, since then i t has been (and continues to be) so widely discussed by philosophers that its claims to primacy must be examined despite their relative recency. Since a speech act, by definition, consists (among other things) of the utterance of other linguistic units such as words and sentences, 5 which are themselves meaningful, i t is obvious that speech acts cannot be regarded basic in the sense of 'the smallest significant unit'i They can, nevertheless be regarded basic from the meaning-theoretic point of view, as Alston tries to do. Alston regards them as basic in that he believes that their study will supply foundations to the meaningfulness of other linguistic units such as words and sentences. We shall see an inherent circularity in this position, in that one cannot hope to give a full description of what speech act is performed in the utterance of a sentence unless one specifies the meaning of the sentence. Thus, speech acts cannot supply foundations to the meaning of a sentence, because the very identity of a speech act is dependent upon or a function of the meaning of a sentence. This will lead us to drop the attempt to regard the speech act as basic from the meaning-theoretic point of view. There is one more approach (represented by Searle's account) with regard to the question of the linguistic primacy of speech acts. This approach regards language itself as performances of various speech acts. Then the theory of language becomes part of the theory of action, and speech acts become not only the basic units of language, but the units of language or — what is the same in this theory — of linguistic communication. Other units, such as words and sentences are reduced to the status of mere counters that become meaningful only insofar as they are used in the performance of various speech acts. This viewpoint also makes words and sentences dependent for their share of meaning on the speech acts, but more radically so than does Alston's account. 6 Alston is consistent in his views when he says that (1) the meaning of a sentence is dependent upon or a function of the illocu-tionary act potential attached to the sentence, and that (2) speech acts are thus the basic units of language. Once his first claim is shown to be false, his second claim falls with i t . But Searle claims on the one hand that (1) the speech act performed in the utterance of a sentence is a function of the meaning of the sentence, and yet main-tains that (2) speech acts are the basic units of language. While refuting Alston's first claim I conclude that Searle's first claim holds, but that that does not supply any evidence for his second claim. Searle's second claim can be refuted by showing that his theory of language is a faulty theory that disregards the competence aspect of lan-guage and only deals with its performance aspect. In showing this I discuss and endorse Vendler's account of language as encompassing the domains of both speech and thought. His analysis in terms of the logical grammar of the verbs of thinking and the verbs of saying helps to bring out the fact that speech and thought both are manifestations of language, each with:somewhat different verbs and different logical grammars, but at the same time, these are overlapping domains with a floating popula-tion of utterances that might belong to both the domains at once. This strengthens my notion that speech acts with a focus mainly on the aspect of linguistic communication leave out an analysis of the verbs of propo-sitional attitudes and achievements, and that therefore they cannot be the basic units of language, nor can the theory of language be reduced 7 to simply a theory of action. I also discuss Vendler's suggestion of 'mental acts' and the possibility of saving the present analysis of speech acts as an analysis of communication by trying to supple-ment i t with a parallel analysis of 'mental acts'; I show that there is more to language than mere performance of 'acts — mental or in speech — and that therefore any analysis that regards only acts as basic misses something of vital importance to the nature of language. This brings me to one last alternative, so far unconsidered in my discussion, but often suggested by the insights from Russell's work, and by my conclusions from examining Alston's and Searle's accounts: the sentence. By way of exploring this possibility Quine*s views in Word and Object are discussed and an attempt is made to show in what sense sentences can be regarded as basic. Following Quine, i t is argued that sentences are most basic to language, both epistemologi-cally and meaning-theoretically. In the chapters on words and speech acts I have tried to make way for a theory of sentence-primacy by pointing out flaws in the proposed theories regarding word-primacy and speech act-primacy. In the chapter on sentences I discuss and answer the objections raised against the sentence-primacy thesis. The ob-jections are shown to be based mainly on uncritical readings of the sentence-primacy thesis. It is also shown that in order to regard sentences as basic, meaning-theoretically, i t is essential to consider the function of sentences learning-theoretically. Discussion of the sentence-primacy thesis naturally brings in the old philosophical controversy over propositions. After discussing the 8 pros and cons of positing propositions as meanings of sentences or as objects of the propositional attitudes I come to the following conclu-sions: (1) that the existence of 'unformulated thoughts' is a myth, (2) that the units of speech and the units of thought belong to the same category, that of sentences, (3) that language — the totality of speech, and thought, of competence and performance — is best viewed as a net-work of sentences. 9 CHAPTER 1 WORDS Word-Primacy Theory; Russell's Version The question of words and sentences, how they mean what they mean and their relationship with or dependence on one another is discussed to some extent by Bertrand Russell in his An Inquiry into Meaning and 4 Truth. This book is based on the text of the William James Lectures that Russell gave at Harvard University in 1940. As is characteristic of Russell, he may have later modified, criticized, changed or completely discarded the views that he advocated in these lectures. Whether or not he did this is immaterial to my purposes, for I am concerned with ex-pounding and explaining the view as i t is stated in his book. If some-body at some time held this view, that gives me sufficient reason to see what its philosophical implications are and whether i t is tenable as expressed. Russell's work in question was intended (as he makes clear in his introduction) as an investigation of certain problems concerning empi-rical knowledge. One such problem in relation to which he proposes to consider language is stated as follows: What is meant by 'empirical evidence for the truth of a proposition'? Russell defines 'proposition' in the above sentence as (the set of) " a l l the sentences which have the same meaning as some given sentence"; He then says that a sentence may be a single word or, more usually, a number of words put together ac-cording to the laws of syntax; and that a sentence expresses something of the nature of an assertion, a denial, an imperative, a desire or a 4 Bertrand Russell, An Inquiry into Meaning and Truth (London: Allen & Unwin, 1940) rept. Baltimore: Penguin, 1969 10 question. He notes that what is remarkable about sentences from his point of view is "that we can understand what i t expresses i f we know the meaning of its several words and the rules of syntax". And so he concludes, "our investigation must therefore begin with an examination first of words, then of syntax". (Russell, 1969, 10) This gives us a gist of his argument for treating words as the basic units of his own investigation. Since he starts with an episte-mological question about 'propositions', and since propositions are explained in terms of sentences and sentences are composed of one or more words, he starts his inquiry with considerations of the 'meanings' of words. This argument in itself would not have been particularly relevant to my present inquiry. For each philosopher may choose to consider some units as basic to his analysis of language, without necessarily implying that his choice reflects the nature of language itself. However, there is one aspect of this argument which suggests that epistemologically, knowing the meaning of its constituent words is prior and necessary to understanding what the sentence expresses. This epistemological priority of (some) word-meanings is what seems to transcend the restricted realm of just one philosophical analysis and suggest something about language as such. This suggestion makes Russell's theory relevant to my present inquiry. In the first few chapters of his work, Russell starts with con-sidering words as 'occurrences in the sensible world!^ that are spoken, heard, written and read. He says that the distinctive feature of 'words' as opposed to other classes of shapes or noises is that 'words' have 11 'meaning'. Before examining 'meaning' in detail, he notices that there are some words which only have meaning in a suitable verbal context; such words as 'than', 'or', 'however' cannot stand alone. They presuppose other words. But he assumes that there are words — including a l l those that a child learns first — that can be used in isolation. These are the words he calls 'object-words' and they compose 'object-language' or the language that is epistemologically a 'primary language'. These words are said to have the following peculiarities: First, their meaning is learnt (or can be learnt) by confrontation with objects which are what they mean, or instances of what they mean. Second, they do not presuppose other words. Third, each of them, by itself can express a whole proposition; you can exclaim 'Fire!', but i t would be pointless to ex-claim 'than!*. It is obviously with such words that any explanation of 'meaning* must begin, for 'meaning', like 'truth* and 'falsehood* has a hierarchy of meanings, corresponding to the hierar-chy of languages. (Russell, 1969, 23) It should be clear from the above passage that Russell considers some words as fundamental to any inquiry of meaning, not only his inquiry, because he believes that " i t is with them that language begins". (Russell, 1969, 24) In his view, epistemology involves both logical and psychologi-cal elements (Russell, 1969, 15) and he believes that what he calls 'object-words' are both 'logically' and 'psychologically* more basic so far as meaning is concerned than either other words or sentences. Object-words are defined logically as words having meaning in isolation, and, psychologically, as words which have been learnt without its being 12 necessary to have previously learnt any other words. (Russell, 1969, 62) Russell admits that these two definitions are not strictly equivalent, and says that wherever they conflict, the logical definition is to be preferred. He arrives at the logical definition by way of accepting Tarsky's suggestion as regards language hierarchy. He starts with Tarsky's thesis that the words 'true' and 'false* as applied to the sentences of a given language must require a language of higher order for their adequate definition. This hierarchy of languages can extend upwards indefinitely since i t is possible to talk about words that talk about other words that in turn talk about...and so on. But Russell argues that this hierarchy must have a lower limit, since, i f i t didn't, "language could never get started". Therefore, there must be a language of the lowest type, and i f there is a primary language, then its words must not be such as presuppose the existence of a language. Thus, he comes to the logical definition of 'object-words' as those that belong to such a primary language, as the words which do not presuppose any other words to account for their meaning, but in terms of which the meaning of other words and of empirical sentences that contain them can be explained. The above argument uses some concepts which require more discussion. One such is the concept of the language of the lowest type. The existence of such a language is predicted on the strength of the argument that i f there were no such language, then language could not begin. What is the threat of this argument? What does Russell mean when he declares that 13 otherwise language could never get started, or in connection with 'object-words* that " i t is with them that language begins"? If Russell is talking about the historical beginnings of human language then his logical definition cannot be interpreted in terms of his psychological definition. For i f he is talking about the point or rather, the level of the historical beginning of human language, then there is not much sense in distinguishing between the words whose meanings are learnt through direct confrontation with the objects and the words whose meanings are not so learnt. Making such a distinction would involve postulating the existence of a class of 'words-to-be-learnt ' in the first place and would go against the hypothesis that language is yet to begin. In order to avoid this difficulty, let us try to interpret Russell's psychological definition in a more liberal manner. Let us not take the word 'learnt' in Russell's psychological definition of 'object-words' in the strict sense in which one learns something which is already existing, e.g., as in 'John learns German.'. Let us instead take 'learn' to mean 'arrive at' or 'think of' so as to avoid the sug-gestion of a pre-existing realm of 'words—to-be-learnt'. Interpreted thus, the object-words are to be defined psychologically as the words whose meanings are arrived at or thought of without its having been necessary to have previously thought of other words. But then, this definition does not t e l l us much about why ' i t is with them that lan-guage begins'. All we know from this i s , that i f there were not the words whose meanings were thought of prior to the meanings of others, 14 then there would have been no beginning of human language. This is as informative as saying, 'if the numeral with which the number series starts has not been thought of, then the number series could not begin'. This appears to me to be a truism. If, however, Russell is talking about (as I think he is,) the beginning of language-learning in the case of ah individual, then his logical definition can be understood in terms of his psychological definition, without straining the usual sense of 'learning'. One can say, that for a child, the words of the primary language are those that he learns f i r s t . In this form, the thesis of the primary language is rather innocuous. Instead of saying, (a) 'there are some words a child learns f i r s t , without its being necessary to have learnt any other words or sentences', one should simply say (b) 'a child learns some words before others'. Whereas (b) is acceptable, (a) is much stronger. I object to the specific condition in (a), namely, 'without its being necessary to have learnt any other words or sentences.' I shall argue below that this condition does not apply to the first few sentences of an individual's language. That a child learns certain words before some other words is an empirical truth which can be tested. In this form, however, the thesis says nothing about how the child knows what these words mean. It sug-gests that since, by hypothesis, the child does not possess, at this stage, any language, he does not learn the meanings of his 'object-words' with the help of any verbal paraphrases. This might be the negative aspect of the claim that the 'object-words' have 'meaning in isolation'. 15 The positive aspect can be summed up by using Russell's description, "their meaning is learnt by confrontation with objects which are what they mean or instances of what they mean.". It is worth noticing that here Russell uses 'means' as synonymous with 'denotes', a move which has unhappy consequences for his general theory. In his l i s t of 'object-words' are not only proper names of people we know, class names such as 'yellow', 'hard', 'sweet', but also names of action such as 'walk', 'run', 'eat*, drink', and also such words as 'up', 'down', 'in', 'out', or even 'quick' and 'slow'. All these words, as 'object-words', are said to have meaning in isolation, i.e., are said to be learnt by direct confrontation with 'the objects that they mean'. Russell uses a simple behaviorism to describe this process of learning a word's meaning in isolation, and also says that a child learning the 'object-language' applies Mill's Canons of Induction and thus gradually corrects his mistakes. In short, he says, By the usual pleasure-pain mechanism which is employed in training performing animals, children learn, in time, to utter noises appropriate to objects that are sensibly present, and then, almost immediately, they learn to use the same noises when they desire the objects. As soon as this has happened, they possess an object-language: objects suggest their names, their names suggest them, and their names may be suggested, not only by the presence of the objects, but by the thought of them. (Russell, 1969, 66) I shall not, at this point, go into a discussion of Russell's supposed equivalence between 'denotation' and 'meaning' or of his behavioristic account of what i t is to learn the object-language, or to learn the meanings of the object-words in isolation of any other 16 words. However, one must admit that certain questions remain un-answered in an account of the above type. Questions of the following form: What is i t for a child to be confronted with an 'object' which is 'what is meant by the 'object-word" such as 'up* or 'down' or 'quick* or 'slow'? If the same sensible 'object' is presented in the presence of words such as 'baby', ' g i r l ' , 'cry' and 'Lisa', are we to say that al l these words denote and mean the same object? Whether or not we accept Russell's behaviorist account of how children learn language, certain things emerge out of the above discus-sion. Russell treats the question of the basic meaningful unit of lan-guage mainly from an epistemological point of view. That i s , the 'object-words' which are said to be the most fundamental to any theory of meaning, are the words that can and do occur in a child's first voca-bulary. What is most important, even from this point of view, the decision about the basic units is not simply between words and sentences, but between those words or sentences which must be learnt f i r s t , and those words and sentences which must be learnt later. I say 'words or sentences' because in spite of Russell's several statements to the effect that i t is with 'object-words' that language begins, i t is clear that his 'object-words* though structurally word-like, are functionally sentences. In other words, i f Russell's theory is accepted, then the basic units of language are not words in contrast to sentences, but words which are sentences; i.e., simply, one-word sentences. Let us see why 'object-words* are to be regarded as sentences. Russell says that a sentence may be a single word, or, more usually, a 17 number of words put together according to the laws of syntax; and that i t expresses something of the nature of an assertion, a denial, an imperative, a desire or a question. (Russell, 1969, 10) Sentences but not words are capable of (1) indicating the state of the world and (2) expressing the state or belief of the speaker. Sentences (one-word or multi-word) are capable of being true or false but single words are neither. (Russell, 1969, 71) And lastly, sentences are composed of words and have a meaning derivative from that of the words which they contain. (Russell, 1969, 18) Russell holds that every 'object-word' is_ a sentence. This is so because at the lowest level of speech every 'object-word' is used in an exclamatory manner and so used, i t always expresses an assertion about a sensibly present object. Thus when a child learning his first words says or hears 'dog' in the presence of a dog, that noise for him is equivalent to an assertion of the form 'That is a dog' or 'There is a dog' or 'A dog i s ! ' etc. Therefore at this stage, his one-word sen-tences share the property of multi-word sentences in that they are capable of being true or false assertions, (Russell, 1969, 26, 71-2, 180, 253, 255 etc.) The most explicit statement of the view that the so-called 'object-words * are actually sentences is to be found in the following quotation: If I exclaim 'Fire!' I express my own state and indicate an occurrence different from my state. The single word is a complete sentence. This is a prerogative of object-words. Other words can only be parts of sentences. I maintain that the use of an object-word as a complete exclamatory sentence is its primary use from which its use as part of a longer sentence is derivative. It is qua sentence 18 that an object-word has the two aspects of expression and indication. (Russell, 1969, 253) If the above view is granted, then the epistemologically basic units of a language for an individual turn out to be one-word sentences. And since Russell believes that the meaning or what he later calls 'sig-nificance' of a sentence is always propositional, i.e., consists of an assertion, a denial, a question or an order, even the significance of these one-word sentences must be propositional. This is what Russell at times admits. But this view comes into conflict with some of his remarks about 'object-words',e.g., "their meaning is learnt (or can be learnt) by confrontation with the objects which are what they.mean, or instances of what they mean." (Russell, 1969, 23) It is clear that i f , on his account, the first learnt linguistic units — the so-called 'object-words' — are in fact one-word sentences, then their meaning cannot just be 'the object'; i t must be propositional. This is just another crack in Russell's supposed equivalence between what the words denote and what the words mean. Confusing statements result from Russell's tendency to call the first learnt linguistic units, i.e., the one-word sentences, 'object-words'. For example, consider the following remarks: Only sentences have intended effects, whereas meaning is not confined to sentences. Object-words have a meaning which does not depend upon their occurring in sentences. (Russell, 1969, 26) It should be observed that meaning of words and sig-nificance of sentences are intertwined except as regards object-words. Other words are defined by means of the simplest sentences in which they occur. (Russell, 1969, 159) 19 ...there is a difference between propositions and single words. Single words, at least such as are object-words, have a meaning which is external to language but propositions, since they can be false, must, except when they express perceptions, have some less direct relation to objects. (Russell, 1969, 253) ...in the object-language, upon which a l l others are based, every single word is an assertion (Russell, 1969, 72) In the above remarks we see a strenuous effort to force 'object-words' into the category of words as well as of sentences. On the one hand, Russell admits that they have a l l the characteristics of sentences, and on the other hand, wants to maintain their 'independence' from sentences. Actually, i f Russell holds that there are one-word sentences and that their distinctive feature is to have a propositional significance in the form of say, an assertion, then on both these counts, i.e., structural and functional, the entities which he calls 'object-words* deserve to be treated as sentences. Russell's urge to treat them also as words stems from his other belief that " a l l necessary words...have ostensive definitions and are thus dependent on experience for their meaning", whereas our understanding a sentence composed correctly out of words depends on our understanding the words, and not on directly 'ex-periencing' the meaning of a sentence as a whole. (Russell, 1969, 290) It is this belief that a l l necessary words are directly related to the experiencial 'meanings* through ostensive definition, coupled with his argument that the units of the lowest language cannot be dependent on anything verbal, that forces him to regard the basic units as 'words' which 'mean' the objects. But as we have seen, this creates confusion. Russell either has to retain the 'object-words' as first learnt 20 words having a direct (?) relationship with the objects that they 'denote' or 'mean', and let go the claim that as 'object-words' they are also sentences, i.e., express assertions, can be true or false, etc.; or he has to hold that the epistemologically basic units of language are one-word sentences, capable of expressing and indica-ting, being true or false, and having-propositional significance which is not confined to any single 'object' as their 'meaning'. He cannot both have his cake and eat i t too. An attempt can be made, however, to straighten out this problem, at the same time bringing out Russell's insights about word-meanings and sentence-significance. In order to do that, two notions have to be discarded. First, the belief that denotation and meaning are one and the same thing, and second, that the first learnt units of an in-dividual's language are words rather than sentences. This is how the attempt goes: Granted that there should be a lowest level of language for an individual, what are the epistemolo-gically basic, i.e., the first learnt units of this language? As i t turns out from Russell's own analysis, there are, both structurally and functionally, one-word sentences. The linguistic expressions that a child learns first are one-word expressions such as 'Dog!' or 'Mama!'. These are truly sentences since at this stage, the one-word expression is used in an exclamatory manner, and serves as an assertion about a sensibly present object. Thus its meaning is propositional, and i t is capable of being true or false. "In this language, when you say 'dog', your statement is false i f i t is a wolf that you are looking at." (Russell, 1969, 71) 21 Once we are clear about the identity of the first-learnt linguistic units, then we can turn to words and say, that among the stock of words of an individual's language, there are some words that he learns before he has learnt any other words. (Note that I say wordst and not 'words or sentences'.) Obviously, these first learnt words would be those that featured in the basic one-word sentences of his primary language. The exact meaning of even these words would have to be worked out by selec-tion and rejection of nuances through various occasions on which the primary sentences are uttered and heard. As Russell says, a child learning the primary language applies Mill's Canons of Induction and gradually corrects his mistakes. If he knows a dog called 'Caesar', he may think this word applies to a l l dogs. On the other hand, i f he knows a dog whom he calls 'dog', he may not apply this word to any other dog. But by and by these mistakes of commission and omission are cor-rected and the meaning of a word is sharpened through its many occurrences in sentences. This account serves to bring out the following insights about lan-guage: A word always has a meaning in a sentential context. Even the words that must be learnt before other words must be learnt through their sentential context. These words occur in the basic one-word sentences, although one may say that this is a limiting case of occurrence, as these words do not occur 'properly' (to use a set-theoretic term,) in the one-word sentences. Since by Russell's own standards the first learnt units of an in-dividual's language have a l l the characteristics of sentences, calling 22 them sentences rather than 'object-words' is certainly less confusing and a lot tidier. It would spare him from making an exceptional case for the notorious 'object-words' every time he talks about either words or sentences. Thus he would not have to say, Single words other than object-words only express and do not indicate. That is why, unlike object-words, they cannot be complete sentences. (Russell, 1969, 255) Nor would he have to say, ...meaning of words and significance of sentences are intertwined except as regards object-words. Other words are defined by means of the significance of the simplest sentences in which they occur. (Russell, 1969, 159) He could do away with the treatment of 'object-words' as exceptional words only i f he separates them from the basic one-word sentences in which they occur. Let me make it clear, however, that I am objecting here not so much to Russell's style as to the gist of his account. By calling the basic sentences 'words' Russell confuses a member of one class with the member of another class. Sometimes, such a way of speaking is entirely unobjectionable. For example, instead of saying, "The birds are quiet", one may very well say, "The chickadees are quiet." i f the birds in ques-tion are in fact chickadees. Although the class of birds and the class of a chickadees are different, there are no chickadees that are not birds. And therefore, when something is called a chickadee, one need not add "...and this chickadee is a bird.". But the case is different with respect to words and sentences. Not every single word is a sentence. Nor is every sentence a one-word sentence. Therefore, when one says of 23 a word, "...and this word is a sentence", one has significantly changed the information content about that linguistic unit. In fact, one has transferred the linguistic unit from one syntactical role (as a sub-sentential unit) to another quite different role (as a self-sufficient sentence). As Russell himself points out, the syntactic and semantic functions of a word of a language are essentially different from those of a sentence. Hence to confuse a sentence with a word is a serious mistake resulting in philosophical muddles, and not just a question of theoretical elegance. Once the mistake is corrected, however, the account will be better and theoretical elegance will also be gained as a bonus. Keeping the basic sentences distinct from the words that occur in them would, as we have seen above, dispel the myth of some words having 'meaning in isolation' or there being words that can 'stand by them-selves'. What Russell really means would be better expressed i f he said, that that which is 'used in isolation1 or 'stands by i t s e l f at the lowest level of language is (not an object-word, not a word at all,) just a one-word sentence. Since, by hypothesis, its meaning is proposi-tional, the meaning of the word occurring in i t , namely, the object-word, is s t i l l a case (however unobvious,) of word-meaning in a sentential con-text. But then one may ask, what makes 'object-words' so very different from the other words in a language? I suggest the following possible answer: The only prerogative of the 'object-words' is not that they have 'meaning in isolation', but that they wear their denotations on their sleeves, as i t were. One would have to distinguish in the case of object-words'their meaning — which is worked out from their sentential 24 context, like a l l other words, and their denotation — which is the object in whose presence the basic one-word sentences are mostly ut-tered and heard. The characteristic of having an obvious object as their denotation would suggest an answer as to why these words are easier to grasp than some other. It would explain why they are learned at a very elementary level of language acquisition, or why the most basic sentences contain these words with obvious denotations rather than words such as 'mind', or 'symbolism' or 'anthropological'. Word-primacy Theory : Ziff's Version In his Semantic Analysis Ziff puts forward the thesis of word-meaning primacy and defends i t against the rival theses such as of the sentence-meaning primacy or the speech act primacy. On p. 149 one finds a statement of Ziff's views on the primacy of meaning in case of l i n -guistic elements. Words are generally said to have meaning or to have a meaning; not utterances and not sentences. We speak of understanding what is said and of knowing the meaning of words. We do not generally speak of "understanding a word" or of "knowing the meaning of what is said". Or more clearly, ...neither utterances nor sentences can in general be said to have or not to have meaning or to have or not to have a meaning. One cannot show or even try to show that two sentences differ in significance in the way one can show or try to show that two words differ in meaning. Ziff's Theory of Meaning Before I start to examine these claims, a brief sketch of Ziff's theory of meaning is in order. I shall only emphasize its salient 25 features and avoid dealing with the details. Early in his book (p. 4) Ziff remarks that a language is defined only by the utterances that can be uttered (by its native speakers) without deviation from certain regularities. He then proceeds to analyze what the word 'good' means in English on the major assumption that meaning is essentially a matter of the non-syntactic semantic regularities. (p. 42, 146) This assumption, spelled out into a state-ment of procedure reads, ...an element's having meaning in the language can be explicated primarily in terms of the non-syntactic semantic regularities to be found pertaining (directly or indirectly) to the element. (Ziff, 1967, 42) Ziff gives a general procedure to decide an element's meaning in English as follows. [There is no suggestion here of a mechanical proce-dure such that i f one followed i t in an algorithmic way, one would get the meaning of an element. These are more or less like guidelines on a hunt.] In determining precisely what meaning a morphological element nu has in J5 (English) 1. the first step is to determine as precisely as possible the membership of its distributive and contrastive sets. [The distributive set for an element i i , of E is a set of syntactically non-deviant whole utterances in which the element occurs where the element is a word or a phrase. (Ziff, 1967, 90, 91). Its contrastive set is the set of those syntactically non-deviant whole utterances which are similar to the utterances of its distributive set except for the fact that a different word m. occurs in these utterances in the place -3 of the element in question in the utterances of the distributive set. 26 (Ziff, 1967, 147Y] 2. The second step consists in determining the relevant nonsyntactic semantic differences between the relevant utte-rances of the distributive set and the relevant utterances of the con-trastive set for the element in question. This step is referred to as 'disambiguation'. (Ziff, 1967, 189) 3. The last step is consoli-dating the results of the second step, which means formulating the rele-vant nonsyntactic semantic differences between utterances of the distri-butive set and the comparable utterances of the contrastive set for nu in E_. This includes forming and testing a hypothesis about what meaning nu has in 12 and preparing something approximate to a dictionary entry for in E. (Ziff, 1967, 193) The first two steps present the most difficult problems. How do we single out the distributive and the contrastive sets for m. in E? - i — How do we separate the relevant and the nondeviant utterances in which m, occurs from the nonrelevant and deviant ones? Ziff denies himself — i the use of the terminology 'rule' and 'exception'. He only deals with regularities, and faces the problem of deciding between the relevant and irrelevant regularities. The principle of conventionality and the  principle of composition are the two principles he accepts as checks on the relevance of regularities. The principle of conventionality requires that a relevant semantic regularity is such that the speakers of the language can deviate from i t at w i l l . Thus this principle rules out the personal emotional reactions of the speakers associated with some words as nonconventional in a special sense and therefore as semantically irrelevant. Ziff gives an 27 example of a woman who generally experiences abdominal discomforts whenever she hears the word 'date'. Noticing this regularity will not help one to find out anything about her language or dialect or idiolect. According to Zif f , the only regularities relevant in a semantic analysis of a corpus in question are those semantic regularities that bear on questions about what words mean, about learning and knowing what words mean, about using or misusing words, about understanding what is said, and the like. Therefore this regularity is ruled out as irrelevant in a semantic analysis of the idiolect in question. It is not clear whether this principle would as well rule out the regular and conventional emotional reactions of a group of-speakers associated with some four-letter words but not x^ith their scientific synonyms. The principle of composition as stated is very vague and gives rise to some problems of interpretation. In a relatively precise but hardly explicit form the principle requires that the structural similarity between w^  and W j be a reflection of the structural similarity between uj. and iy. (Ziff, 1967, 62) Here u. and u. are two distinct utterances and w. and w. are their —3 — i —] 'exclusive' semantic correlates or the metalinguistic expressions that can be 'paired' with them in accordance with the relevant nonsyntactic semantic regularities found in E. This device of pairing an utterance with a metalinguistic statement about its significance rather than pairing i t directly with a state of affairs is employed by Ziff in order to avoid any specific ontological commitments. Anscombe^ in her review of Ziff's Semantic Analysis criticizes this move as unsuccessful. 5G.E.M. Anscombe, "Critical Notice: Ziff's Semantic "Analysis" Mind Vol. LXXII, 1963 28 However, I shall not pursue the matter further since whether or not he succeeds in this move is not of much consequence to my present purposes. The more serious issue here is of specifying the exact nature of the 'structural' similarity and dissimilarity, and deciding on the bases of that which utterances can and which cannot be paired with the ut-terances in question. In order to decide about the structural similarity of two utterances in question the notion of structure must be understood. But there is good reason to think that an adequate understanding of this notion does not now exist. For example, possible u^, U j pairs as the following: 'He is easy to please.' / 'He is eager to please.' or 'She was to go tonight.' / 'She has to go tonight.' present a problem for Ziff. He is well aware of this difficulty and admits that in default of a complete and adequate deep structure grammar one is forced to talk in a more or less vague way. Other difficulties arise as we probe deeper into his theory. His analysis is based on two important assumptions about meaning. (Ziff, 1967, 147) The first i s : i f we consider the syntactically riorideviant  whole utterances of a spoken language in which a morphological element — say a word — occurs, i t is reasonable to suppose that that element will  not differ in its meaning from one utterance to another. The second as-sumption i s : the significance of what is said depends on what is not  said. Both these assumptions are intuitively appealing but Ziff's elaboration on them seems to give rise to certain difficulties. Ziff's first assumption is that a morphological element of E that 29 has meaning in English will not differ in meaning in each and every utterance of E. This assumption helps him to determine whether or not an element of E_ has meaning in _E, and i t also helps him to test his hypothesis "about what meaning nu has in E_ by considering the relevant conditions associated with the utterances of the distributive set of m.. In other words, his thesis is —a. to have a meaning in English, m^  must have meaning in various utterances of its distributive set and i t must not appreciably differ in meaning in these utterances. (Ziff, 1967, 171) He discusses the following case to bring this point home. Consider the pair of "tigers' 'That is a tiger.' and 'I want a tiger.'. In order to see that the word 'tiger' in the first utterance does not differ in its meaning from 'tiger' in the second utterance, consider the sets of conditions associated with these two utterances and those associated with other comparable utterances of the contrastive set for 'tiger'. The sets of conditions are indicated by the brace '{' and '}'. There is a difference between {'that is a tiger.'} and {'I want a tiger.'}. But there is also a difference between {'That is a tiger.'} and {'That is a lion.'}, and there is a difference between {'I want a tiger.'} and {'I want a lion.'}. And more to the point, there does not seem to be any difference between the differ-ence between {'That is a tiger.'} and {'That is a lion.'} and the difference between {'I want a tiger.'} and {'I want a lion.'h (Ziff, 1967, 155) This lack of difference is seen as a result of the fact that 'tiger' does not change its meaning in the two utterances in which i t occurs. Anscombe in her review criticizes Ziff on this point as follows: He says that there does not seem to be any differ-ence between the first difference and the second. It is difficult to see how he can t e l l . His judgement 30 is surely merely based on the fact that one account can be given of the differences between the sentences in the two cases, namely that 'lion' occurs in one sentence where 'tiger' occurs in the other, without any switch of meaning between the pairs. He cannot professedly rest his case on this because he is trying to supply foundations for i t . (Anscombe, 1963, 290) Anscombe is not right here because Ziff's judgement need not be merely based on the account of the difference between the sentences, as she believes. Ziff's concern here is with measuring the relevant dif-ference in terms of conditions associated with the utterances rather than the sentences themselves. So the charge of circularity cannot be made against him. At least in the above case he can compare the conditions associated with both the pairs of utterances and say that the only dif-ference between the first pair of conditions is in terms of lion(s) and tiger(s) and not between 'lion' and 'tiger'; and that the only dif-ference between the second pair of conditions is also in terms of (a desire for) a lion/tiger. He can thus attribute the difference to the occurrence of 'lion' in one case and 'tiger' in the other. However, this procedure is threatened with serious difficulties in some other cases, such as the following one. Suppose that one wants to see, following Z i f f , whether or not 'plan' changes its meaning in the utterances 'I want a plan' and 'I carried out a plan'. He will start to compare the sets of conditions associated with these two ut-terances with those associated with parallel utterances of the contras-tive set for 'plan'. Consider the following pairs: (la) {'I want a plan.'} (lb) {'I want a division.*} (2a) {'I carried out a plan.'} (2b) {'I carried out a division.'} 31 Would he want to maintain that there is no difference between the dif-ference between (la) and (lb) on the one hand and the difference between (2a) and (2b) on the other? The real question i s , how can he decide that there is or that there is not a difference? He cannot appeal to the ambiguity of 'division' to support a claim that there is_ a difference in the associated sets of conditions. If he did, then he would be talking about words, not conditions, and Anscombe's charge of circularity would apply. Further, he cannot simply rule out (la) and (lb) as an irrelevant comparison, on the grounds that the associated conditions of these are different, because he has no way of telling that they are different other than by appeal to the ambiguity, and, hence, meaning of 'division'. And, again, he cannot bank on the ambiguity of an element, because he is in the process of addressing the question of what i t is for an element's meaning to remain constant from one utterance to another. There is a certain uneasiness about the role of 'conditions' in Ziff's analysis of meaning. Relevant conditions associated with the ut-terances of the distributive set of word are supposed to yield certain nonsyntactic semantic regularities, the regularities in turn forming crystals of the semantic information about the word in question. So determining the exact nature of conditions is of utmost importance. Conditions are the extralinguistic circumstances and the relevant con-ditions with respect to the meaning of a word are not merely truth conditions of an utterance but other conditions as well. Ziff c r i t i -cizes as a serious confusion the belief that truth conditions and not conditions simpliciter are of fundamental importance in the speaking 32 and understanding of a language. Consequently he says, There is no reason why the declarative utterance must be, as i t were, the standard meter rod of a language. (Ziff, 1967, 139) And yet at one place we find Ziff entertaining a similar confusion between truth conditions associated with declarative utterances and other conditions associated with nondeclarative utterances. Consider his remarks on p. 154. . . . i t might seem that there is a simple criterion for whether or not elements mj[ and m^  differ in meaning in English, viz., m^  and mj differ in meaning in£English i f and only i f {d-j(m^)T is not identical with {d^(n^)/mj}. But this won't do at a l l . There are cases in which {d^(mi)} is identical with {d-f (m-j)/m-j} and yet m^  and mj differ in meaning in English. Thus {'The glass is exactly half (full).'} is identical (or virtually identical) with {'The glass is exactly half (full) .'/'empty'}. (i.e., the condi-tions associated with .'The glass is exactly half full.',} are identical (or virtually identical) with the condi-tions associated with 'The glass is exactly half empty.1) and yet ' f u l l ' and 'empty' differ in meaning in English. Thus one says, ' F i l l i t half full.* but not ' F i l l i t half empty.', and consider the difference between the two imperatives 'Keep on pouring until i t is half empty!' and 'Keep on pouring until i t is half f u l l ! ' . This counterexample is peculiar. In order for i t to be counted as a counterexample, Ziff has to show that 'empty' and ' f u l l ' differ in meaning while the conditions associated with the comparable utterances of the relevant sets containing these words do not differ. But on his own word, he cannot show that. For, granted that the pair of the declarative sentences in question will have truth-conditions, Ziff must admit that the {d^O^)} is not identical with {d^m^/mj} so far as the imperatives are concerned, contrary to his own assumption. Only by insisting on mere truth-conditions of the declarative utterances in 33 question as relevant for deciding the sameness or difference of meaning of m. and m. can Ziff treat this case as a serious counterexample to -a- ~3 what he calls 'a simple criterion'. But he would not want to do that. He insists elsewhere (Ziff, 1967, 139) on regarding conditions other than truth-conditions as just as relevant in semantic analysis. If so, i t is obvious that the so-called counterexample fails to invalidate the criterion, because while the words in question differ in meaning in English, the conditions associated with the imperatives in which they occur also differ accordingly. Even so, in the light of Ziff's remarks- on p. 139, we might try to reconstruct his argument (in the above passage) against the exclusive importance attached to mere truth-conditions associated with the dis-tributive and contrastive sets for an element. For this, we introduce the notion '{d.(m.)} ' to stand for the set of truth-conditions asso-— i — i 1 ciated with the distributive set of the element nu ; and the corresponding notion '{d. (m.)/%•}' to stand for the truth-conditions associated with the — L -a j T contrastive set of that element. Then the 'simple criterion'which he is trying to denounce would be rewritten in terms of truth-conditions as follows: nu and differ in meaning in English i f and only i f (d^(nu)}^ is not identical with (d^(m^) /',Ej}T.. Then his counterexample might suggest, that whereas (f^On^)}^ = {^Q2L[)/Ej \ hence, by hypothesis, the meaning of nu = the meaning of nu ; this is not correct, and i t shows up in the cases of imperative utterances when we see that {d.(m.)} ? {d.(m.)/m.}. Thus, understanding — l — l -I-I —] ' 6 34 1 conditions' to be broader than truth-conditions agrees with our intuitive conviction about the meanings of 'empty' and ' f u l l ' . However, i f this is the sort of criticism Ziff is trying to advance in the passage in question, then the following two objections can be raised. First, supposing for a moment that his line of attack is correct, producing only one context which has the same truth-condi-tions when ' f u l l ' is replaced by 'empty' hardly confirms that {d.(m.)}m = {d,, (m,) /m. },„, i.e., that the set of truth conditions as-—1 —a. T — i — i — j I sociated with the distributive set of m. is identical with the set of truth conditions associated with the constrastive set of that element. For even i f we confine ourselves to truth-conditions, what about cases like "The glass is empty.' and 'The glass is full.'? This shows that in order to find a real counterexample, Ziff has to find out cases where the set of truth-conditions associated with the distributive set of an element is identical with the set of truth-conditions associated with the contrastive set of that element, and yet the elements in question differ in meaning, which difference shows up in the set of conditions associated with the distributive and contrastive sets of that element. This leads to my second objection. It seems to me that although Ziff's insistence on considering other conditions along with truth-con-ditions is intuitively appealing, it would be very hard to find con-vincing cases of two elements differing in meaning where the difference was reflected only in conditions associated with imperatives or questions and not so reflected in the conditions associated with the declaratives as well. If there is any difference between the meanings of the two 35 elements in question, there is no reason why i t should be obvious only in utterances other than the declaratives. Since i t is possible to find at least one corresponding declarative for every imperative in question, i t is possible also to find corresponding similarities or differences in the meanings of some elements occurring in the declara-tives. Thus, for example, i f the difference in the associated conditions is obvious when one considers the two imperatives, 'Keep on pouring until i t is half empty.' and 'Keep on pouring until i t is half f u l l . ' ; then I think i t is equally obvious in the corresponding declaratives, such as 'She kept on pouring until i t was half empty.' and 'She kept on pouring until it was half f u l l . ' . In short, I do not think that Ziff has given us sufficient reason to believe that an analysis of truth-conditions is inherently incapable of revealing the similarities or the differences in the meanings of words, which are revealed only when other conditions are considered. Next, there is the question of inferring the relevant semantic regularities from the conditions associated with an utterance, which in turn determine the semantic contribution of a word in a given utterance. Mien he wants to determine the semantic contribution of a word in a given utterance, Ziff is concerned with actual regularities, and not with projections on the bases of these regularities. This means that the con-ditions which yield actual regularities relevant for his purposes are not of the form, 'the circumstances under which one would normally say, "I saw a tiger.", and would not say, "I saw a lion."'. Rather, the condi-tions will be of the form, 'the circumstances that have generally held 36 when "I saw a tiger." has been uttered, and have not held when "I saw a lion." has been uttered.' This restriction confines Ziff to an ex-tremely limited 'four-dimentional world', as i t were, and i t is rather an awkward position with regard to regularities. Anscombe rightly criticizes this position as leading to shipwreck when she says, If there has mostly been mustard about when 'Pass the salt' was said, that there has mostly been salt about is not going to distinguish 'Pass the salt'. Not merely must there be a large number of stock utterances in stock contexts — which is surely untrue: only a small part of language is like 'Hello', an example on which Ziff heavily leans — but these utterances must very often have been made in given contexts which obligingly avoided also being each of other (indefinitely numerous) given contexts....For Ziff cannot rely on single occasions to show anything about meaning:" i t takes a lot of occasions to have a regularity. (Anscombe, 1963, 291)' Let us now turn to the second assumption. The significance of what is said depends on what is not said. The utterance actually uttered stands in contrast with and takes its shape from what is not but could without deviation be uttered. (Ziff, 1967, 147) At the least, this principle tells us that language forces an option of elements from which some element is selected in an utterance, and that there is no such option with respect to 'to' in 'I want to go through Istanbul.' indicates that 'to' in that utterance does not have meaning. It is on the basis of this principle that Ziff considers the contrastive set for an element in question. As I said before, i t is not entirely clear which of the elements of the contrastive set are going to be relevant for testing a hypothesis about a word's meaning in an utterance in accordance with the above principle. The only time Ziff actually 37 compares 'good' with the elements of its contrastive set is when he contrasts i t in 'That is a good painting.* with the available alter-nates such as 'beautiful', 'magnificient', 'splendid', and the like. He uses these contrasts to suggest that 'good' has a relatively dis-passionate feeling associated with i t , and that the other words express more passionate feelings. He accounts for this fact about the feeling of 'good' on the bases of his hypothesis about the meaning of 'good' in English — its connection with interests. But looking for only such elements of the contrastive set clearly isn't going to be enough. For there are other elements belonging to the contrastive set of utterances, elements such as 'horrible', 'cheap', 'ugly' etc., which could without deviation be employed, but were not actually employed. It would be interesting to see just in what way these other elements were supposed to shape the morphological identity of the element in the given utterance. One thing Ziff would not do at this point is to suggest that somehow these other elements of the contrastive set were not employed because the specific intended speech act in the given utterance was that of commending or the like. For Ziff criticizes any such attempt to relate the contribution of a word in an utterance directly with the intended speech act as misguided. He could, however, try to relate the whole ut-terances of the contrastive set such as 'That is a horrible painting.' etc., with the conditions of different speech acts and associate them with the semantic contribution of the contrastive elements. This is a line he could have taken, but does not take. Therefore i t remains to be seen in 38 what way the utterance actually uttered takes its shape from all those possible but unactualized contrastive utterances. Consider another environment for 'good', e.g., 'What good is that?'. Here 'good' supposedly contrasts with 'use' to give 'What use is that?'. But does i t also contrast with 'bird' or 'book'? So far as I can see, it does, at least on the surface of i t , and does present a problem to anyone who maintains that 'the significance of what is said depends on what is not said.'. In order to dismiss these as nonrelevant contrasts he would have to discuss the structural dissimilarities between the con-trastive pairs and the question of deciding what that is is left open. Ziff dismisses 'use' as a contrastive element for 'good' in some environ-ments e.g., 'That is a good book.' by showing that the substitution of 'use' for 'good* in this frame produces a syntactically deviant utterance. (Ziff, 1967, 152) But since this move would not work in 'What good is that?' for 'use', the question of fixing the ultimate set of contrastive elements for 'good' remains unsettled. Evaluation of Ziff's Word-meaning Primacy Claims The statements of the word-meaning primacy view occur in Ziff's Semantic Analysis on two different (although not wholly unrelated) levels. Sometimes Ziff gives an impression that he is simply stating i t as a fact about how people talk of language. Thus we find him saying, Words are generally said to have meaning or to have a meaning; not utterances and not sentences. We speak of understanding what is said and of 39 knowing the meaning of words. We do not generally speak of "understanding a word" or of "knowing the meaning of what is said". (Ziff, 1967, 149) If his claim here is an empirical one, that i s , i f he is simply observing and describing the way people talk about words and sentences, then i t is easy to point out that this is not the case, that this is not how people talk and that therefore his claim is empirically falsified. Linguists as well as philosophers have expressed diverse views on which linguistic units have meaning and which ones don't. To quote just a sample, A phonetic form which has a meaning is a l i n - guistic form. Thus, any English sentence, phrase s or word is a linguistic form, and so is a meaning-ful syllable.... L. Bloomfield, Language, (New York: Holt, 1933, 138) Many linguistic elements other than words may be said to have 'meaning' of some kind; all morphemes are by definition significant...and so are the combinations into which they enter, and al l their various meanings play their part in the total meaning of the utterance. Stephan Ullman, Semantics, (New York: Barnes and Noble, 1964, 54) This is the principle of the theory of denoting I wish to advocate: that denoting phrases never have any meaning in themselves, but that every proposition in whose verbal expression they occur has a meaning. Bertrand Russell, Logic and Knowledge, (Ed.) Robert C. Marsh (New York: Capricorn, 1971, 42) - However, i t does clearly make sense to speak of different kinds of meaning for sentences on the one hand and sentence-components like words on the other, for the account of what i t is to have 40 a certain meaning is different in the two cases. William P. Alston, Philosophy of Language, (New Jersey: Prentice-Hall, 1964, 74) And Max Black even speaks of "understanding a word". We might say, in general: a person understands vermilion when he can recognize the presence and the absence of the colour in question... A necessary condition for understanding a word is the ability to use i t in the correct setting. Understanding involves, also the ability to apply and withhold the word in novel and complex situa-tions , and hence the ability to combine i t cor-rectly, and in the correct order, with other words.... r-^Briefly, then, words are understood, not in isolation, but in interrelationship with other words. We understand systems of symbols, rather than single words; a word is no more to be understood without understanding its linguistic associates than a hand can be shaken without touching a body. Max Black, The Labyrinth of Language, (New York: The New American Library, 1969, 51) Examples such as these can be multiplied. All they show is that i f Ziff is making a general statement about the way people talk of the meaningful units of language, then his statement meets with serious ex-ceptions from various corners and that therefore i t cannot be of much value in his semantic analysis unless considerably modified. Of course, Ziff can always take refuge in his own idiolect and say that although many others may have said things different than he does, that is not the way he_ uses his language. In that case, the issue boils down to terminological differences between Ziff's writings and those of the others, and Ziff cannot claim the support of common usage for his semantic theory any more (or any less) than the others can. Terminological differences do not cause any problem so long as they are regarded as being just that. Thus, one can see that whereas 41 Carnap uses 'meaning' with respect to sentences, and Strawson uses indiscriminately 'meaning' and 'significance' with respect to sentences, Ziff reserves 'meaning' for words and 'significance' for sentences. J But Ziff frowns at metalinguistic sentences such as 'What does that sentence mean?' and calls them odd. He seems to score a point for his choice of terminology on the bases of the common usage. Russell once made a distinction similar to the one made by Z i f f , but he claimed fib support for.it in common usage. "significance" and "significant" are words that I apply to sentences, whereas "meaning" is a word that I apply to single words. This distinction has no basis in usage, but i t is convenient. (Russell, 1959, 158) We see now that either Ziff's statements, taken as observations about peoples' talk turn out to be evidently false, or they can be reduced to his terminological peculiarities. In either case, they alone do not serve as reasons or evidence for Ziff's word-meaning "If we knew what i t would be for a given sentence to be found true then we would know what its meaning i s . And i f for two sentences the conditions under which we would have to take them as true are the same, they they have the same meaning." Carnap, "Testability and Meaning" Philosophy of Science, vol. i i i and i v , 1936, 1937. 7 "To talk about the meaning of an expression or sentence is not to talk about its use on a particular occasion, but about the rules, habits, con-ventions governing its correct use, on a l l occasions, to refer or to as-sert. So the question of whether a sentence or expression is significant  or not has nothing whatever to do with the question of whether the sen-tence, uttered on a particular occasion, is on that occasion, being used to make a true or false assertion or not or whether the expression i s , on that occasion, being used to refer to, or mention anything at a l l . " Strawson, "On Referring" reprinted in Contemporary Readings in Logical  Theory, Ed. Copi and Gould, p. 112. 42 primacy theory. As a third possible alternative we can take Ziff to be talking not about philosophers' specialized talk of meaning nor about his own idiolect, but about the way "common men" talk. Even then, Ziff has not provided any evidence in his favor. So far as I can see, we do talk of both words and sentences as having meaning. There-fore, in the lack of any pressing evidence, Ziff's claims stand un-substantiated. However, sometimes Ziff appears not to be just describing the common usage, but explaining why people talk the way (he thinks) they talk. These explanations are worth considering in their own right, because they supply reasons for Ziff's own version of the word-meaning primacy theory. These explanations have a prescriptive flavor to them, as Ziff is concerned here not so much with pointing out the 'fact' of common usage as with justifying what he takes to be the point of such a usage. If Ziff is right about these explanations, then regardless of whether or not i t is in accordance with common usage, his semantic theory would strongly prescribe that words be rightfully treated as the units which can properly be said to have meaning or have a meaning. One such explanation is given by him in terms of what he calls a matter of memory limitations. He notes i t as a fact, yet not as an inevita-ble fact about language that words primarily, but not utterances or sen-tences are said to have meaning. Thus he says, Phonemes, morphemes, words, phrases, sentences, ut-terances, are a l l relatively arbitrary units singled out and employed in the analysis of language. If we were not the sort of creatures we are, i f we could remember i t or hold i t in mind, we might tend to focus 43 on some unit larger than an utterance, say a speech, or conversation, and then talk of the or a meaning of an utterance with respect to this larger unit. We might have been the sort of creatures that would be apt to talk in this way. But we are not and do not. That words primarily, but not utterances, are generally said to have meaning in English should not be surprising: that a word has a distributive set in _E is obvious; that utterances have a distributive set is not obvious. It is a matter of what we can remember, of what we can hold in mind. (Ziff, 1967, 151) This account brings in a factor of human memory. It is not very clear how one is supposed to take i t . If the point about memory is sup-posed to be a determining factor in the decision about the basic meaning-ful unit, then the fact that other people have decided in favor of units other than words will prove (as I shall argue below,) that Ziff's obser-vation about our memory limitations must be wrong. If, on the other hand, the observation about memory is just another cause contributing to Ziff's own choice of the unit of analysis, leading to some unit as having a dis-tributive set, and therefore as having meaning, then the argument is no longer a decisive one for regarding only words as having meaning. Let us discuss this in detail. First of a l l , there are difficul-ties about interpreting the exact sense of the word 'focus'. What does focusing on a linguistic unit amount to? Does i t mean starting our analysis with? Or is Ziff talking about a l l people in general and not just about philosophers of language? If he i s , then what does i t mean to say that people in general 'focus' on whole utterances? Does i t mean that they are normally aware of the utterances any more-.than of some other units? Or that they pick out utterances when they are thinking about language? These questions are unanswered and the word 44 'focus' in his account remains unexplained. But let us proceed.. The fact (if i t is a fact,) that we can only focus on an utterance and not on any other unit, how does this serve to point out that only words can be regarded as the units that have meaning? At the most, i t can indicate that an utterance can be the most convenient unit for an analysis of speech. How does Ziff connect this result regarding the useful unit of analysis with the result about the primary linguistic unit of meaning? He seems to connect the two with his notion of 'having a distributive set1. The main line of his argument seems to be as fol-lows: (1) Memory limitations cause us to focus on whole utterances rather than any other units. (2) Whole utterances are the members of the distributive sets for words, and since we focus on utterances, that words have distributive sets becomes obvious. (2a) Since we do not focus on any other units, that other units have distributive sets is not obvious. (3) By hypothesis, the meaning of an element is a function of the relevant regularities revealed by the distributive sets of that element. (4) The meaning of words is thus obtained as a function of the relevant regularities revealed by the distributive sets of words. (4a) Since i t is not obvious that other units have distributive sets, i t is not obvious that they have meaning either. (5) Therefore, only words and not any other units are said to have meaning in E. It seems to me that Ziff's appeal here to our natural memory limitations and the resulting tendency to focus on the utterances is not very convincing. If memory restrictions are/given to us as human beings, then the resulting disability to focus on a unit other than an 45 utterance must also be universal. But i t turns out to be not so uni-versal after a l l . Austin and philosophers after him such as Searle, Alston have focused on a unit other than an utterance, namely, a speech act, and at least Alston, in a sense, has tried to determine the meaning of a sentence with regard to the illocutionary acts which are performed in the utterance of that given sentence. Although they do not use Ziff's terminology, one may interpret their claims in such a way as to result in showing that sentences or utterances have distri-butive sets whose members are speech acts. Thus one may point out that certain sentences have only certain illocutionary acts as the members of their distributive sets and not others, and then analyze how this may reflect on the meaning of the sentences in question. Such a line of thought is not altogether implausible. In fact, when Ziff talks about speech acts, (1967, 78-9, 87, 88) this is what he appears to be doing. Moreover, Alston, in his review — "Ziff's Semantic Analysis", (Che Journal of Philosophy, Vol. LIX, No. l j Jan. 1962; 15) thinks that Ziff's account of semantic regularities and deviations is oversimpli-fied because instead of concentrating on a speech act, Ziff merely con-centrated on an utterance as his unit of analysis. Alston's reasons for this criticism might be wrong, but the fact that he finds fault with Ziff's choice of utterance as a unit of analysis at least casts a doubt on Ziff's observations regarding our natural tendency to focus on an utterance, and the unobviousness of the existence of distributive sets for units other than words. Producing a rival analysis which starts with a unit of analysis 46 quite different from one's favored unit of analysis may not in itself be a threat to any theory about the basic meaningful unit of language. But in Ziff's case, i t i s . This is so because Ziff seems to establish a vital connection between our choice of a unit of analysis and the emerging meaningful unit of language through the notion of having a distributive set. So focusing on any other unit of analysis might result in the emergence of a different unit of language with dif-ferent distributive sets, and thus ultimately prove to be a rival for Ziff's theory about the primary unit of meaning. However, i f he is not counting on memory limitations or our natural tendency to focus on utterances as a decisive reason for re-garding words as the primarily meaningful units of language, then i t is hard to see why he insists time and again that only words can be said to have meaning, not utterances, nor sentences. There is a possibility that in the above passages he is saying something as follows: It is theoretically possible for us to adopt the approach of finding distri-butive and contrastive sets for units other than words. However, one finds that we simply do not adopt i t with respect to any other unit than words. Therefore, only words, have meaning in that only they seem to have distributive sets which reveal relevant semantic regularities. If this is. what Ziff is trying to say, then not only is this version much weaker, but it is susceptible to more objections. If the word is to be regarded meaningful on the ground of its merely contributing to the significance of a sentence, or to the total meaning of an ut-terance, then why cannot the same be said of a phoneme? And more 47 seriously, i f , as Ziff himself has said, words, phonemes, and sentences a l l are linguistic elements which are yielded by different analyses, and are normally said to be meaningful or significant, then why should the fact that one explains word-synonymy in a different manner than one would explain sentence-synonymy count as a reason for restricting the term 'meaning' to words rather than to sentences? At the most, they can explain Ziff's own motives for using 'meaning' only with words rather than with sentences, but they are not decisive reasons to prove that only words should be regarded as the units that have meaning. Ziff is mainly concerned with the meaning of a word. But in order to fix that, he has to start with the correlation between units larger than words such as sentences or utterances and consider the elements of meaning reflected in the metalinguistic statements of the projections from these larger units. That he is obliged to do so suggests something important about the dependence of word-meanings on the meanings of larger units. Zif f , however, fails to acknowledge the significance of his own procedure for the questions of the meaning-theoretic primacy of certain linguistic units. A good case of this, oversight is his criticism of Frege's oft-quoted principle. I shall quote Ziff in f u l l : Furthermore, one may know what meaning m^  has in English and yet not be able to specify the member-ship of either the distributive or the contrastive sets for mi in E.. This is to say that one may know what meaning a word has in English and yet not know how and when to use i t . Frege said: "Only in the context of a sentence do words mean something.". But as against this one can say 'Famulary' means of or belonging to servants.': here no reference 48 either is or need be made to any utterance or sen-tence in which the word occurs. I know what 'famulary1 means but I do not know how or when to use the word. Again i f someone tells me that 'ultus', a word of Latin, means revenge, do I know how to use the word? Only i f I speak Latin. These are sur-prising facts only when seen from the vantage point of a misguided theory. (Ziff, 1967, 189) I think that Ziff's criticism of Frege's principle here is unfair. First of a l l , because Ziff takes the principle out of its context, and secondly, because he does not allow Frege the privilege of certain refined distinctions he allows himself. My first point is as follows: Frege's principle quoted in translation by Ziff appears in Frege's Grundlagen der Arithmatik. For Frege, i t is one of the three funda-mental principles of his inquiry into the concept of number. The principle in its first occurrence in his work reads as "never to ask for the meaning of a word in isolation, but only in the context of a proposition." (1950;Xe) Frege's reason for adopting this rule as a principle of inquiry is as follows: "If the ...rule is not observed, one is almost forced to take the meanings of words as mental pictures...". Frege does not relish the notion of equating the meaning of a word with a mental picture or an idea, because 'ideas' are strictly sub-jective entities for him, and he regards the concept of the meaning of a word as objective. The fact that one cannot form an idea of the content of a word is not, according to Frege, a good reason to My references on this and the next page are to Frege's.Grundlagen der Arithmatik translated by J.L. Austin as Foundations of "Arithmatic. (Oxford: Blackwell, 1950) ~~ 49 say that the word does not have any meaning. In order to avoid this 'psychologism' Frege adopts the above-mentioned rule as one of the principles that guide his inquiry. He believes that by allowing for the possibility of the meaning of a word to be determined by the para-phrastic analysis, rather than presenting a direct equivalent to i t , he can explain reference to fictitious objects in meaningful sentences, without being committed to the existence of the objects. Frege's prin-ciple when taken in its proper context appears to be a good principle of semantic inquiry. Ziff's effort of transplanting i t puts i t in a dim light. It creates a wrong impression that Frege and Ziff are talking about the same thing and that Ziff's cases are serious counterexamples to Frege's 'misguided theory'. In fact, this is not the case. Ziff's cases are not a real threat to Frege's semantic insights. They simply miss the point of his principle. This brings me to my second point. I said that Ziff is not as generous to Frege as he is to himself. Let me make this clear. Earlier in his work, Ziff (1967; 95) distinguishes between what is meant by an element, what is connoted by an element and what the ele-ment means. To talk about the first and the second in the trio, he says, is to talk from the speaker's and from the hearer's point of view respectively. Only to talk about what an element means is to be, "in between speakers and hearers". If this is the case, then on the same grounds Ziff must allow for the fact that Frege's principle is indeed a rule of thumb for starting a fruitful semantic inquiry and not an account of when a person can say that he knows what a word 50 means without committing himself to knowing how to use i t . Obviously, for Frege, what a word means is an objective concept f i t to be a sub-ject of a semantic analysis and a person's being able to say that he knows the meaning of a word is at best a subjective condition not necessarily related to an objective analysis. One might say that these two are just as different for him as are for Ziff what an element means and what is connoted by an element. Therefore the cases that Ziff comes up with need not be surprising for Frege. They need not even be relevant to his purposes. He would simply point out that knowing what a word means is not to be confused with being able to make certain metalinguistic statements about what a word means. Both Ziff's cases are about persons who apparently are in a position to make such statements but are not in a position to do anything else with that piece of information. Apart from what Frege might have said on this point, i t appears to me that Ziff's examples hinge on ah ill-applied distinction between knowing that and knowing how. The distinction is ill-applied in this case because knowing what a word means is indeed more intimately connected with knowing in what kind of sentences i t might appear, rather than merely being able to say that i t means this or that. Coming across a statement such as ' 'Ultus' means revenge in Latin' may be a beginning of my learning Latin; but only after I have seen how that word is used in connection with other words to produce sentences in Latin will I be able to claim reasonably that I know what the word means. It may only be a matter of degrees of knowledge, but possessing an isolated metalinguistic 51 statement about the meaning of a word puts one on a much lower level (of knowing what the word means) than being able to operate with i t . In order to see why, let us consider the following. Even i f I know what a word means in English, and can show that I know what i t means by using i t correctly in my utterances, by detecting its misuses on occasion, and perhaps even by making some cor-rect remarks about its syntactic role (i.e., by saying that i t is a noun or an adjective and so on,) I may not necessarily know how to summarize my know-how in a neat metalinguistic statement of the type 1 'ultus' means revenge in Latin'. This disability on my part will not be counted as a sign of my ignorance _of the meaning of a word. It may only point towards a fact that some words, even some very simple, preliminary words in my vocabulary defy neat statements _of their meaning. [Try stating what words like 'than' or 'am' mean in as simple a statement as the one about 'ultus'.] On the other hand, when someone informs me that 'ultus'means revenge in Latin, the extent to which I may nontrivially say that now I know what 'ultus' means depends on my already knowing what 'revenge* means in English. This latter, in order to avoid a regress, must be grounded in my knowledge of how to use 'revenge' properly in sentences of English, rather than a similar metalinguistic statement such as ' 'revenge' means....in English.'. In short, knowing what a word means is very similar to knowing what a king in chess is or to knowing what the sign '+' in arithmatic i s . Full knowledge of any one of these amounts to knowing their proper contexts and their relations with the other members of the group in 52 a particular context, also knowing what would amount to their misuses in a certain context, and not only coming across an isolated statement about their meaning. Ziff's two cases are cases of meaning-i'n-isolation and they can only be understood on the basis of some previous knowledge of meaning-in-context in general. We have seen that Frege's principle i s , as a matter of fact, a guideline to be followed in a semantic analysis. Ironically, even though Ziff appears to object to its soundness on p. 189, the procedure of his own semantic analysis is more or less an application of this very prin-ciple. In order to-:see what the word 'good' means in English, he goes through several sentential contexts in which i t occurs and only at the end of that inquiry summarizes his results in a metalinguistic state-ment about what that word means. Frege warns us against asking for the meaning of a word in isolation. Ziff is careful not to take the isolated meaning statements fresh from a dictionary as Holy Writ, (p. 168, 225, etc.) We find Ziff always checking such statements against actual evidence of the words occurring in different sentential contexts; and weighing their relative worth. All this shows quite clearly, that in spite of Ziff's unfair criticism of Frege':s principle, in his actual practice he does follox* Frege's advice in its proper spirit. One may say that he credits Frege's insight after a l l , i f not in words, certainly in deeds. 53 CHAPTER 2 SPEECH ACTS What is a Speech Act Speech acts or some kinds of speech acts have been regarded as 'the appropriate unit of linguistic behavior' (Alston), or as 'the basic or minimal units of linguistic communication' (Searle). What is a speech act? This question is answered differently by different philosophers. 9 As far as I know, i t was Austin who first used the term 'speech acts' in its more or less technical sense and put forward a tentative theory regarding different sorts of speech acts, mainly, the locutionary, i l — locutionary and perlocutionary acts. Austin does not give a definition of speech act but what we can gather from his account and examples i s , in short, as follows: He considers 'the cases and senses in which to  say something is to do domething; or in which by_ saying or in saying something we are doing something.' (Austin, 1967, 12) Thus rather than concentrating on words, phrases, or sentences as such he focuses on the acts of uttering these, 'the total situation in which the utte-rance was issued — the total sppech act.' (Austin, 1967, 52) Speech acts fa l l roughly in three main groups: locutionary, i l -locutionary and perlocutionary. Out of these three Austin is mainly concerned with the illocutionary speech act. Austinians such as Alston and Searle also enthrone the illocutionary act as the basic unit of linguistic communication. So I shall be mostly concerned with the i l -locutionary act, but f i r s t , a brief sketch of Austin's general theory. The locutionary act, following Austin, is the act of 'saying something in the full normal sense' which includes the utterance of certain noises, ('the phonetic act'), the utterance of certain words 9 John Austin, How to Do Things with Words (1962, rept. New York: Oxford University Press, 1967) 54 in a certain grammatical construction, ('the phatic act'), and the ut-terance of them with a certain 'meaning' ('the rhetic act') which Austin equates with a certain sense and a certain reference. 'To perform a locutionary act is in general, and 'eo ipso1 to perform an illocutionary act.' (Austin, 1967, 98) Austin is at pains to make clear what he means by an illocutionary act, and more often he relies on giving examples rather than giving a definition of illocutionary  act. In the locutionary act of uttering a meaningful utterance, e.g., 'There is a bull in the field,', one may perform one or more of the illocutionary acts of stating, reporting, warning, reminding, etc. The act of doing some such thing in the act of saying something is called the illocutionary act. Austin contrasts this group of acts with the locutionary as well as with the perlocutionary act. Roughly, a locutionary act is the act of saying something meaningful, an i l l o -cutionary act is the act of doing something in the act of saying some-thing meaningful, and a perlocutionary act is the act of doing some-thing by_ saying something meaningful. The category of the perlocutionary acts consists of the achieve-ment of the intentional as well as the unintentional consequences of the acts of saying something meaningful. For example, by saying 'Ce"zanne is greater than Picasso.', I may please some, insult some, or astound some of my hearers. These acts of pleasing, insulting, and astounding, insofar as they were performed by_ saying something, are to be regarded as perlocutionary acts. The purported distinction between illocutionary and perlocutionary 55 acts is brought out by Austin with the help of the following: (1) The formula 'doing v_ in saying x_' as opposed to the formula 'doing v_ by saying x_'. (2) The perlocutionary act is in essence the achieving of certain effects or consequences, by saying something while the i l l o -cutionary act is not, except for 'securing illocutionary uptake' i.e., the audience hearing and understanding the intended illocutionary force of the speaker's speech act. (3) The illocutionary rather than the perlocutionary acts are said to be 'conventional' in the special sense that at least the illocutionary act could be made explicit by the per-formative formula, 'I promise that...', 'I warn you that...' etc. but we cannot say e.g., 'I persuade you that...' or 'I bore you that...'. All these tests for distinguishing the illocutionary from the perlocutionary act have certain exceptions and Austin notes some of them. Thus, the formula, 'in saying x* versus 'by_ saying x* might yield cases like the following: "In saying 'She is beautiful' I was joking, or reciting a line of verse, orsteaching grammar." But Austin would not want to call these illocutionary acts, for they are, accor-ding to him, the non-literal or not serious ways of using language. Or again, the illocutionary act of judging a suspect guilty may nec-essarily have the effect of his being required to serve the sentence, or my act of promising you ten dollars may necessarily have the effect of my being under obligation to pay you ten dollars. But perhaps Austin can cover these as 'conventional' rather than spontaneous consequences and. s t i l l try to maintain the distinction between the illocutionary and the perlocutionary acts. Thirdly, the formula which marks the 56 illocutionary acts as conventional in his special sense is not appli-cable to some of the illocutionary acts, e.g., we cannot say, 'I side with that...' or 'I welcome you that...' and yet 'side with' and 'welcome' are listed as illocutionary verbs. In passing, I may mention that the perlocutionary acts, since they include both the intended as well as the unintended consequences (Austin, 1967, 106) of the utterance on the audience, are indeed, un-limited and theoretically difficult to handle. One may say, for ex-ample, that bj_ giving them a long sermon, the priest bored the people, put them to sleep, made them get up with a sense of relief, start their cars, go home, lose their interest in religion, and so on. The ques-tion i s , why call a l l such 'effects' or 'consequences' speech acts of the speaker at all? In fact, the unintended consequences such as the above ones tend to dissociate themselves from the speech of the speaker, and also become more and more the acts (if at all,) of the audience rather than of the speaker. However, the perlocutionary acts do not claim to be the basic units of language in the works of Austinians, and so are not important for my present purposes. Austin (1967, 132) maintains that every act of speaking in the normal and serious sense, except perhaps a mere exclamation like 'ouch', is both a locutionary as well as an illocutionary act. This gives rise to the question of the distinction between the locutionary and illocu-tionary acts. It may well be that a sentence such as 'It is raining' may have associated with its utterance a battery of different illocu-tionary forces such as staging, warning, asking a question, protesting 57 and so on. But in the case of some utterances x^ hich consist of only the explicit performative formula such as 'I protest.' or 'I thatk you.', it is hard to say how one can characterize their locutionary act without bringing in the description of the illocutionary act as well. That i s , i f one wants to separate the meaning (brought out in the locutionary act) of an utterance from its illocutionary force (brought out in its illocutionary act,) then in the case of at least some explicitely per-formative utterances such as the above, one finds that the account of the meaning of these utterances necessarily exhausts that of their i l l o -cutionary force. If this is so, then insofar as meaning and illocutio-nary force are overlapping categories, Austin's attempt to keep them apart suffers a loss. L.J. Cohen10 has argued on the strength of this and other such considerations that what Austin calls the illocutionary force is not anything different in kind from the meaning of an utterance, but rather that i t is just another aspect of the meaning of an utterance. Besides Cohen, at least two other philosophers, namely Hare 1 1 and 12 Searle have criticized Austin's distinction between locutionary and illocutionary acts. The main difference between Cohen and Searle is that Cohen rejects the notion of illocutionary force as empty whereas Searle tries to preserve that notion. Searle claims that illocutionary 1 0L.J. Cohen, "Do Illocutionary Forces Exist" Philosophical Quarterly 14, 1964, 118-137. **R.M. Hare, Practical Inferences (London: Macmillan, 1971). 12 John R. Searle, "Austin on Locutionary and Illocutionary Acts" Philosophical Review Vol. 77, 1968, 405-424. 58 speech acts are basic to language, but rejects the notion of locutio-nary acts as untenable. Hare's position is similar to that of Searle. I argue below that both Hare and Searle claim to have proved more than they have shown. That i s , both Hare and Searle claim that the dis-tinction between locutionary and illocutionary acts breaks down while at the most they can show that illocutionary force is an aspect of the meaning of a locution. I shall mainly discuss Searle?s treatment of the issue but my conclusion would also apply to Hare's criticism. Searle criticizes (1968; 405-424) Austin's original distinction between locutionary and illocutionary acts as not completely general. He tries to show that Austin's distinction cannot be maintained as a distinction between two mutually exclusive classes of acts. He uses two arguments. First, he argues from the so-called explicit performa*-tive utterances where there is an indication of the illocutionary force of the utterance right in the meaning of the sentence. Thus, for example, he says, that a 'serious and l i t e r a l ' utterance of 'I hereby promise that I am going to do the job' must be a promise; as-suming that the act of promising is successful, i.e,, that the condi-tions of successful utterance are satisfied. Since the meaning of the sentence determines uniquely the illocutionary force of its utterance, the description of the act as a happily performed locutionary act is already a description of the illocutionary act. Therefore, he concludes that at least for the cases involving a performative use of the illocu-tionary verbs the distinction between locutionary and illocutionary acts does not exist. That i s , the class of illocutionary acts will contain 59 at least some members of the class of locutionary acts. Now the above argument gets off the ground by assuming that the conditions of successful utterance of a performative is satisfied, i.e., by blurring the distinction between an utterance of a sentence with meaning m including illocutionary force I? and a successful performance of the illocutionary act of the type F_. In a footnote Searle suggests that although there might have been a possible distinction between the two, he treats them as equivalent throughout his paper,, because he says that that distinction is immaterial for his purposes. It is not obvious that this distinction is of no consequence to his arguments. In fact, i t appears that i t is crucial for his argument against Austin. It is on the strength of the equivalence between an utterance with a certain force and a performance of an illocutionary act that he can pass from statements of the form: 'for utterances of the performative use of the illocutionary verb, the illocutionary force is included in the meaning of the sentence' to statements of the form: 'for utterances of the per-formative use of the illocutionary verbs, there is no way of abstracting the locutionary act which does not bring in the illocutionary act with itL' (emphasis mine). One may grant Searle that for some utterances their illocutionary force is included in the meaning, is a part of the meaning, or is uni-quely determined by the meaning of the sentence. But this is not to grant that the successful performance of the locutionary act is already a successful performance of the illocutionary act. The account in terms of meaning and force is not necessarily the same as the account in terms 60 of acts; and one who accepts the former may s t i l l object to the latter account. This is so because the account in terms of acts requires as a necessary condition the condition of the successful performance of the illocutionary act, i.e., the avoidance of various 'infelicities', securing illocutionary uptake etc.; but the account in terms of meaning and force has nothing to do with this condition at a l l . The account in terms of meaning and force is a general one and is somehow prior to con-siderations regarding the 'felicity' of a particular illocutionary act. This priority is in a sense epistemological in that one may know upon hearing the utterance "I hereby order you to leave the room" the propo-sitional content of the utterance and also know that the illocutionary force 'uniquely determined' by the meaning of the sentence is that of an order. And yet one might not know anything as to whether i t was a successfully performed illocutionary act of ordering, for one may not kncn-; whether the several necessary conditions for its successful per-formance were fulfilled. In other words, knowing the force of an utteran-ce as determined by its meaning (in Austin's sense) is quite different from and also independent of knowing what illocutionary act was actually performed on an occasion. Therefore, i t is advisable to leave the 'act' terminology and reconstruct Searle's argument in more neutral terms. In fact, Searle says in footnote 5 that nothing in his argument hinges upon his interpretation of the relation between a meaningful utterance with an illocutionary force J_ and an illocutionary act of type F. He also claims that his arguments are statable in essentially the same form on either interpretation. Let us see how one can do this. 61 Let us say that a locution is defined as an utterance with a certain meaning (in Austin's sense) and that an illocution is an ut-terance with a certain force. To start from the explicit performa-tives, one may argue that in the utterance of "I congratulate you" its illocutionary force, namely that of congratulating, is uniquely determined by the meaning of the word 'congratulate'. Therefore in the case of the above utterance, to know its meaning is to know what force it has. Thus one may claim that the class of locutions will contain at least some members of the class of illocutions. But now let us pass from this to Searle's more general argument which is adduced to support a much stronger claim that the class of locutionary acts contains no member which is not also a member of the class of illocutionary acts. He argues that no sentence is completely force-neutral. Every sentence has some illocutionary force potential, i f only of a very broad kind, built into its meaning. He notes that the examples Austin gives of the locutionary verb phrases— 't e l l someone to do something', 'ask whether' are just as much illocutionary verb phrases as are 'promise someone', 'order someone', etc. He thinks that this is no..accident;- He takes Austin's inadvertent characterization of the locutionary acts as symptomatic of a general epidemic where no sentence is found to be completely force neutral, and hence no locution comple-tely free of a germ of an illocutionary force. He says, The concepts locutionary act and illocutionary act are different. But the conceptual difference is not suffi-cient to establish a distinction between separate classes of acts, beeause just as every terrier is a dog, so 62 every locutionary act is an illocutionary act. (Searle, 1968, 413) So the conclusion i s , "there is no specification of a locutionary act performed in the utterance of a complete sentence which will not de-termine the specification of an illocutionary act as well." Trans-lating this into the meaning-force terminology, the conclusion becomes, "there is no specification of a meaningful utterance as locution which will not determine the specification of ah illocutionary force." Let us test this against the following case: "You grow sweet potatoes". This is an utterance of a meaningful sentence with a pro-positional content JJ and supposedly with some force J _ . But unlike the utterance of the explicit performative type, one cannot say that its force whatever i t i s , is uniquely determined by the meaning or the pro-positional content of the utterance. It seems less likely that one would claim that to know the meaning of this utterance is_ to know its illocutionary force. It could be a command, a statement, a question, an accusation, and so on. It is hard to see why one is obliged to specify any of its several force potentials while specifying the pro-positional content of the utterance. One could describe the utterance as follows: the speaker S ut-tered a string of phonemes _r understood to have the propositional con-tent J J , where _r stands for the actual string of phonemes 'You grow sweet potatoes' and p_ stands for the proposition that 'the hearer (addressed in the second person singular/plural) raise a crop of large sweet f a r i -naceous, tuberous root, eaten as a vegetable'. It seems to me that this is s t i l l a description on the levels of the phonetic, phatic and 63 rhetic acts (i.e. , on the level of locutionary act) and that one need not bring in an illocutionary verb even of a very general kind such as ' t e l l ' or 'ask' in order to so describe i t . To insist that the above is not the usual way to give an indirect speech report is of no use. If Searle's challenge is that there is no way to specify the proposi-tional content of a meaningful utterance without determining its i l l o -cutionary force, then the challenge is met by inventing a forced or artif i c i a l yet intelligible description of the utterance which remains neutral to its illocutionary force potential. To find such a des-cription is not to deny that the utterance may have some illocutionary force attached to i t . It is only to deny that one is forced to specify the illocutionary force while describing the propositional content of an utterance. Someone may s t i l l wish to object that the above descrip-tion is incomplete insofar as i t lacks information on stress, intonation contour, etc. But to insist that these things should be.included in the account of the propositional content is already to show a theoretical bias as regards the scope of the propositional content. To include factors such as stress, and intonation contour, may help to isolate the intended illocutionary force (although they are perhaps not decisive); but that these must be included is yet to be shown, and it-seems that i t cannot be shown without question-begging. What we can conclude from the above discussion is as follows: One, locutionary and illocutionary acts need to be separated from mean-ingful utterances and illocutionary forces. The condition of 'success-ful' performance is crucial to the first category, i.e., that of the 64 acts. It need not play any role while considering questions of meaning and force relations. Two, one may grant that no sentence is completely force-neutral and that therefore to know (or to describe) the proposi-tional content of a meaningful utterance is to some extent, to specify (a range of) its Illocutionary force, which, in the case of utterances involving explicitly performative verbs, is 'uniquely' determined by the meaning, while in other cases, only limited by the meaning of the sentence. This, i f true, will show only that illocutionary force is a part of the meaning (in the case of performatives), or a function of the meaning (in the case of the other utterances) of the sentence. This is an important consideration in the decision regarding the question of being the 'basic meaningful unit of communication', but more about that below. What Searle cannot prove is the claim that every locutionary act (where the locutionary act is defined as the act of uttering a meaning-ful sentence) is_ an illocutionary act. He cannot prove this because there is an asymmetry between the concepts of the locutionary act and the illocutionary act. Unlike the illocutionary act, there are no f e l i -city conditions required (by either Austin or Searle) for the performance of a locutionary act. Every time one utters a meaningful sentence, one performs a locutionary act. But one does not perform an illocutionary act every time one utters a meaningful sentence with a certain illocu-tionary force. Several other things, such as having the right authority the right occasion, securing uptake have to go right before one can say that i t was a successful performance of the particular illocutionary act. 65 In the absence of these factors, one may have performed a locutionary act, but not necessarily an illocutionary act, of however general type. In fact, there are many locutionary acts which are not illocutionary acts of any type because one does not know what to call them and how. These are infelicitous acts involving explicit performatives, or acts involved in stage-acting, or practicing sarcasm, etc. One cannot deny that they are locutionary acts for they do involve meaningful utterances. If one wants to maintain that even these locutionary acts are i l -locutionary acts of some sort, one has to pay the price of giving up the conditions of 'successful' performance such as securing uptake. This would then allow us to count in acts of speakers with proper inten-tions, say, to issue a warning as illocutionary acts of 'issuing a warning' (as opposed to 'warning somebody') even i f the warning was not heard or understood by anyone. This may perhaps help one to maintain that every locutionary act is an 'illocutionary' act of some kind or other, but i t would do so at the cost of stretching painfully the notion of illocutionary force. For i f the conditions of proper speaker, proper occasion, securing uptake are removed, then there is no way to keep the babbling of fools off limits. If a blessed soul believes himself to be Genghis Khan and commands someone to slay his harem, even this locu-tionary act of his would be an illocutionary act of some (?) sort. On the other hand, i f someone insists on successful performance as a necessary condition on the concept of an illocutionary act, then he has to admit that there are some locutionary acts that are not i l l o -cutionary acts. This is to admit further that even though there may be 66 some cases of utterances holding a dual citizenship, that is not suffi-cient to conclude that the class of locutionary acts is no different than the class of illocutionary acts. The concept of the locutionary act is_ different from the concept of the illocutionary act, and what is more important, even their extensions are not the same, so the analogies of terrier and dog or of unmarried man and bachelor are of no help. Keeping the above criticism in mind, let us see i f we get a clear notion of what an illocutionary act i s . We saw that Austin wants to keep distinct the meaning of an utterance and the illocutionary force attached to i t . Cohen maintains that there is virtually no distinction of kind involved in these two, and that the force, along with other con-textual aspects, is only an aspect of the total meaning of an utterance. Searle, as we saw, maintains a thesis stronger than Cohen's in saying that there are no locutionary acts that are abstractably distinct from illocutionary acts. We saw too, why he cannot hold this stronger thesis without giving up the notion of felicity conditions of an i l l o -cutionary act. There is another curious consequence of Searle*s posi-tion. By claiming that Austin's examples of the locutionary verb phrases, e.g., 't e l l someone to do something','say that*, 'ask whether' are as much illocutionary verb phrases as 'state that', 'promise some-one that', Searle considerably widens the class of the illocutionary acts. What is not so clear i s , do we by the same token include in this class acts of speaking such as 'talk to', 'shout at', 'whisper', 'lecture', 'interrupt', 'interview',;'converse with', etc.? No specific reason is given to exclude these; for they a l l suit the description of 67 speech acts: 'to say something meaningful'. Austin's criterion for the so-called conventional nature of the illocutionary acts, namely, that they can be made explicit by the use of the performative formula, e.g., 'I promise that', will not be of any use to rule out the above acts, for the criterion fails with respect to many of the accepted i l -locutionary acts as well, (see p. 56 above)t On the other hand, i f we make place for verbs such as 'lecture' or 'interrupt', in the illocu-tionary category, then there are some obvious differences, say of the 'verbal form' (see below) of these acts that become problematic i f i l -locutionary acts are to be regarded as the most basic meaningful unit of linguistic communication. Take the form, for example. Searle (1970, 25) says that the char-acteristic grammatical form of the illocutionary act is the complete sentence, although i t can be a one-word sentence. But i f the linguistic acts such as lecturing, giving a sermon or interviewing are to be in-cluded in the illocutionary acts' category as well, (and there is no reason given why they should not be,) then some unit larger than a sen-tence seems to be needed. One way out of this difficulty would be to say that the above acts consist of different illocutionary acts which are characterized by the use of a complete sentence, so the above acts are activities rather than individual illocutionary acts as such. This is not very satisfactory. For similar description can be given of some other illocutionary acts in -'Austin's original l i s t , and their claims as genuine illocutionary acts can be questioned. In order to rule out acts such as 'lecturing' or 'interviewing', i t has to be proved in the 68 first place that the standard grammatical form used is a complete sentence, but this meets with exceptions even in the case of accepted, so-called genuine illocutionary acts, for example, acts such as 'analyze', 'recapitulate', 'describe', 'answer' and so on. They seem to take some unit larger than a sentence as their form too. And as for the speech acts such as 'interrupt', one does not even know what unit can characteristically be allotted to them. So far as the content is concerned, speech acts vary enormously. One simply does not understand why only certain acts are then to be in-cluded in the illocutionary category while keeping certain others out. "Using the simple test (with caution) of the first person singular present indicative active form, and going through the dictionary in a liberal spirit, "Austin got a l i s t of the illocutionary verbs in English of the order of the third power of 10! He classified them in five groups called 1) verdictives, 2) exercitives, 3) commissives, 4) behabitives, and 5) expositives. Searle does not attempt any such classification, but analyzes the act of promising and says that the analysis can be ex-tended with some modifications to other illocutionary acts as well. Both Austin and Searle exclude from the class of illocutionary acts acts such as 'joking', 'acting a part', or 'writing poetry'. They both call these uses of language parasitic, non-serious or non-literal, and there-fore exclude them from the class of the illocutionary acts, which class, they claim, is restricted to 'the full normal use of language' alone. There are differences of opinion on this point among philosophers, Ziff (1967, 77), for example, classes any act as a speech act insofar as the performance of the act necessitates the, uttering of an utterance, and in accordance with this characterization, he cites speech acts such 69 as making a statement, giving an explanation, quoting, reciting a poem, or telling a story. Alston expresses a similar view in his review of Ziff's Semantic Analysis when he says, (1962, 15) that the utterance of the sentence 'Pass the salt.' in the course of reciting a poem or telling a story or in jest, a l l constitute genuine and distinct speech acts. What is more important, Alston also notices that "there is an important sense in which these uses of the sentences are derivative from the more primary use of asking for salt, but that does not give them the status of deviations". I think that this is an important in-sight because i f one is engaged in a synchronic study of language, (as I believe both Austin and Searle are,) then i t is not quite justified to rule out the acts in question here as not fully normal or not genuine speech acts before one has analyzed what a normal speech act i s . Insofar as speech acts are said to be the acts of the speakers in day-to-day communication, or the acts of saying something meaning-ful, the above acts of telling a story or joking are just as normal and regular as promising or ordering are. Austin's reasons for ruling out these acts as not serious are that in these acts the normal requi-rements of reference are suspended and that no attempt is made at a standard perlocutionary act. However, these reasons are somewhat cir-cular, since what the normal requirements of reference are or the standard attempt at the perlocutionary act i s , becomes clear only after the analysis of the 'normal' illocutionary acts is given, and in order to begin this enterprise, one is required first to rule out those acts \tfhich are not normal. 70 In order to avoid this circularity, we must be given an indepen-dent criterion which would first single out only those speech acts which coincide with what is popularly called the normal or serious use of language. Then we can perhaps see why the so-called non-serious or fictive uses of language are not normal. What is needed and is not to be found so easily is an intelligible criterion in order to separate the poetical use of language in everyday conversation from the use of language in what is specifically called poetry or fiction. This is a rather slippery area and much detailed work is needed before we can even begin to distinguish normal and non-normal uses of language. Our every-day language is studded with frozen or half—frozen metaphors. We talk of the legs of a table, arms of a chair, ears of a corn-stalk, eye of a needle and so on. We 'jump to conclusions' , "stand by an opinion' , 'hold a b e l i e f , 'back up a statement', 'evade an issue', 'attack an opponent', 'stress a point', and 'hang on to or leave a subject'. Examples can be endlessly multiplied. The point is that unless one specifies what i t is that makes the above phrases and verbs acceptable in our 'normal', 'standard' and 'serious' conversations while barring acts such as telling a story, or joking, one cannot simply brush aside the latter as non-serious or non-literal. Then there is the whole lot of T.V. commercials, political campaigns and international peace talks — a l l quite respectable, normal-and supposedly non-fictive uses of lan-guage, where one is hard put to say whether to regard this use of lan-guage as serious or not, as suspending the normal attempt at the perlo-cutionary acts or not. 71 On the other hand, is i t really true that when one is writing poetry or acting a part, one is necessarily saying something in the non-serious mode? Can we say every time a sentence occurs in the course of a poem that the normal requirements of reference are not fulfilled? Consider the following case: 'A thing of beauty is a joy for ever.' This is a line of poetry. What unfulfilled reference requirement or unattempted standard perlocutionary act makes us entitled to say that therein is a non-serious use of language? To begin with, i t is not at al l clear what perlocutionary acts are to be regarded as 'normally* associated with certain illocutionary acts. Consider: 'Children are always younger than their parents.' What perlocutionary acts are normally associated with the above sen-tence and how do we decide that they are inhibited when the sentence appears as a line of a poem? If, as on Austin's original analysis, even unintended effects of the speaker's utterance on the audience are to be counted as the speaker's perlocutionary acts, then there is simply no line of demarcation between the normal and the not normal acts of the perlocutionary kind. Even i f one sets out on a Quixotic enter-prise after listening to Walt Whitman's 'Go catch a falling star'', that can s t i l l be counted as a perlocutionary act, whether intended or not. Once unintended consequences are in, i t is awfully hard to keep normalcy in as well. Intentions and Securing Uptake: The concept of what i t is to perform a successful or complete or happy or felicitous speech act depends on two notions: the notion of 72 intentions and the notion of securing uptake. Let us first consider the notion of securing uptake. 'Securing uptake' is Austin's term for bringing about the intended illocutionary effect upon the audience. From some stray remarks of Austin (1967, 22, 115), this effect seems to consist in the audience understanding (the meaning and) the illocu-tionary force of the speaker's locution. Searle, (1970, 47) calls this the 'illocutionary effect' and says that the effect on the hearer is not a belief or response, but that it consists simply in the hearer understanding the utterance of the speaker. Both Austin and Searle think that this element-is essential for the felicity of an illocutio-nary act. Austin talks of the "doubt about whether I stated something i f i t was not heard or understood" (1967, 136), and calls such a case a case of the infelicity arising out of misunderstanding (Ibid., 22). Searle (1970:47) makes a similar claim when he says, "unless he recognizes that I am trying to t e l l him something and what I am trying to t e l l him I do not fully succeed in telling.it to him." I want to discuss the condition of securing uptake as it appears in the writings of Austin and Searle. But before that I would like to mention one more interpretation of the condition in order to keep i t 13 aside for the moment. P.F. Strawson considers the following objec-tion to Austin's doctrine of the necessity of securing uptake for the performance of an illocutionary act. P.F. Strawson, "Intention and Convention in Speech Acts" Philoso- phical Review Vol. 73, 1967, 439-460. 73 a man may, for example, actually have made such and such a bequest, or gift, even i f no one ever reads his will or instrument of gift. To cover this he modifies Austin's notion such that at least the aim, i f not the achievement of securing uptake is an essential element in the performance of an illocutionary act. To this modified version he considers another objection, a case where a person makes a gift without any expectation of i t being known. He replies by amending the original account so as to make the aim of securing uptake a standard, i f not an invariable element in the performance of an illocutionary act. Thus he maintains that the analysis of the notion of securing uptake remains an essential element in the analysis of the notion of the illocutionary act. On this interpretation, however, the condition of felicity of an illocutionary act becomes much weaker than either Austin or Searle are willing to make i t , and 'the aim to secure uptake' turns out to be just another name for the speaker's intention to let the hearer recognize what he is saying. I shall therefore reserve the discussion of this as-pect of intention t i l l we come to the discussion of intentions in general. Coming back to the condition of securing uptake, I would like to argue that this condition is not very helpful in the matters of the felicity of an illocutionary act. For one thing, we do not always in everyday conversation find the simple one-to-one correspondance between the speaker and the hearer. One speaker may be confronted with many hearers and may address them as a group. If one goes by what Austin and Searle say, then depending on the different responses or lack of responses 74 of the members of the audience, the same speech act of a speaker would have to be described as both felicitous and nonfelicitous. Think, e.g., of a teacher announcing the date of an examination to his class. If some unattentive students did not hear him or understand what he said, are we entitled to say that he_ did not succeed in announcing it? Perhaps one can try to avoid this sort of difficulty by trying to distinguish between the speech acts of making an announcement and 14 announcing. Siegler suggests the following as a logical principle when he distinguishes between issuing a warning and warning somebody. He says, In order to complete the act of issuing a warning it is not necessary that the person to whom i t was directed hear and understand i t . But, to complete the act of warning i t is necessary that the person to whom i t was directed hear and understand i t . This suggestion seems promising, but in fact, i t is rather lame. First of a l l , i t is not clear that we have the corresponding verbal conjunct (with an auxiliary and a nominalized verb) for several verbs in Austin's original l i s t of illocutionary verbs in English. For ex-ample, the following do not seem to have such a corresponding form: 'read i t as', 'reckon', 'put it at', 'declare closed', 'side with', 'bind myself, ' t e l l ' . Even for verbs which may have a corresponding verbal conjunct with an auxiliary and a nominalized verb, i t does not seem that the condition of securing uptake is necessary in one case and Frederick A. Siegler, "Lying" American Philosophical Quarterly Vol. 3 No. 2, 1966, 128-136. 75 not necessary in the other. For instance, take the following verbs from Austin's original l i s t and compare their corresponding verbal conj uncts: 'describe' 'give a description' * remark' 'make a remark' 'revise' 'make a revision' 'correct' 'make a correction' 'define' 'give a definition' ' understand'- ' come to an understanding' 'interpret* 'give an interpretation* 1 diagnose' 'make a diagnosis' 'calculate' 'make a calculation' It is clear that one could not perform the act denoted by the verbs in the left hand column i f there was no understanding hearer any more than the acts denoted by the phrases in the right hand column? It ap-pears to me that the condition of the hearer's understanding is equally unnecessary (or necessary) for verbs in either column. Nevertheless, granting for a moment that we do distinguish between 'warn somebody' and 'issue a warning1, let us go back to the verbs which can be distinguished on this model. Tine question then i s : which of the groups deserves the title 'illocutionary act verbs'? And how are we to regard the rest? How we answer these questions will determine the verbs denoting those acts that are the basic units of linguistic communication. The decision is not an easy one. Since we are discussing conditions on the hearer's understanding, that understanding itself could hardly serve 76 as the deciding factor for bestowing the title on one rather than the other group. On a l l other counts such as being that which we do in the act of saying something, or being an act that is conventional, both single verbs and their nominalized forms with some auxiliaries seem to hold equal chance of being called the illocutionary act verbs. Another difficulty involved in Searle's insistence on having the hearer's recognition as an essential element in the notion of success of a speech act can be seen in the following case. Suppose A's new servant is a Chinese boy who does not understand English. A does not know this and thinks that the boy knows English. The boy makes some inexcusable mistake in his work one day. A gets infuriated, calls the boy a good-for-nothing fool and orders him to leave his sight immediately. The boy does not follow a word, but seeing that his master is in a bad mood, leaves immediately. It seems that A's speech act of ordering has not secured uptake in either Austin's ...or. Searle's sense. They would say that the. speaker's act of ordering the boy to leave was unsuccessful or infelicitous. But what happens when A's son who knows both English and Chinese and who is also present there later tells the boy what i t was that A had said, and the boy recognizes exactly what i t was that A was trying to te l l him? Do we then say, following Searle, that since the boy has now understood what A was trying to tell.him, A's illocutionary act is now successful or felicitous? It seems that what we need here is an optimum modulus of any speechpperformance, so that we can talk about felicity as one of the functions of the modulus. That we can fix an unarbitrary optimum modulus is not obvious. And in the absence of 77 any given modulus, questions like the following arise: How long can the characterization of the felicity of a speech act wait before the hearer (or the hearers) understand the import"of the act? Is i t per-missible to have someone other than the original speaker convey what the original speaker said? Or does the notion of securing uptake stand on its own and apply only to direct, one-to-one communication? If the latter is the case, then how do we account for political communications with the help of interpreters? And where do we draw a line between such organized activities of interpreters and an occasional interpreter's activity as in the Chinese boy case? To rule out every such case as non-standard seems arbitrary. There seems to be something wrong in characterizing an utterance as act x. depending upon the condition that the hearer understand i t as act x. I think that the fault lies somewhere in the process of gene-ralizing from an analysis of some explicitly performative utterance such as promising to a general analysis of a l l utterances belonging to the class of illocutionary acts. Having the hearer recognize you as promising him something may be essential to the success of the act of promising, especially since there are some legal considerations in-volved in its public aspect. But to apply this to a l l other illocu-tionary acts gives rise to certain problems. We sat/ above with respect to some verbs in Austin's l i s t that the condition of securing uptake does not seem to be necessary either for their illocutionary use, or for the illocutionary use of their nominalized forms with some auxi-liaries. The condition is especially troublesome when one considers 78 the propositional acts or attitudes involved in the locutionary act. For example, i f I say out loud "llwonder whether the new teacher's name is 'Jones'", must a hearer recognize what I said as having the illocu-tionary force of a certain type? (Must there be someone who can be spoken of as 'the hearer1?) Must I be able to specify my intention in the utterance as that of performing this particular illocutionary act rather than that one? In turn, must he also recognize that this utter-ance was not to be taken as that type of an act? Suppose he misunderstands the force of my remark in such a way that he responds to i t thinking that i t was a query to him, whereas in fact, i t was just an idle remark of mine addressed to no one. If I then accept his response to my utterance as relevant and continue the (unintended) conversation, is one entitled to characterize my i n i t i a l remark one way or the other? Was i t felicitous as one kind of illocu-tionary act and infelicitous as another? It did not secure uptake in any intended sense. It somehow elicited a response from the person who  heard the utterance and the response formed a part of the following con-versation although that particular response was not intended to be achieved by the speaker in the first place. In this case the speaker's intention which is supposed to be tied up with 'securing uptake', and the notion of hearer's understanding which also forms a part of 'securing uptake' are two forces working against each other and the retroactive analysis of the illocutionary act seems to break down under the stress. The notion of 'securing uptake' is essentially associated with the speech situation in which the speaker is involved in getting across some message to the hearer and is exercising his intentions accordingly. When we remind ourselves that sometimes utterances involving expressions 79 of propositional attitudes that are not necessarily directed to a hearer may function almost as pieces of intended communication, and be responded to, we realize that the notion of 'securing uptake' is not very helpful. This Ms so because the notion of 'securing uptake' is supposed to be decisive with regard to the questions of the felicity of a speech act, but the individuation of a speech act is what needs to be decided upon before we can go on to deciding its felicity. Next, we go to 'intentions'. The notion of a speech act is ex-plained in terms of al l sorts of intentions. There are intentions of the speaker to communicate with the hearer, his intentions to produce in the hearer an understanding of his utterance u relying on the con-ventional meaning associated with u, and there are also his intentions tied up with what Searle..calls1'the sincerity condition1, e.g., the intention to keep the promise he has given. We find that the intentions are tied up on the one hand x-rith the speaker's beliefs, and on the other hand xtfith the notion of securing illocutionary uptake. In the following pages, I shall try to sort out and analyze the intentions and see how essential they are to the notion of a speech act. That the speaker's intentions in general are essential is to be seen from the following argument by Searle.1"' He says that one of the things that is involved in our taking a noise or mark on paper as an instance of linguistic communication as opposed to just a natural 15 John R. Searle "What is a Speech Act" in Max Black (Ed.) Philosophy in America (Ithaca: 1965) 221-239. 80 phenomenon is that we take the noise or the mark as having been produced by a being with certain intentions; i.e., we take it as a speech act. Thus intentions in general on the part of the speaker are said to be essential for a speech act. In a later and more detailed analysis of speech acts Searle modifies the crude notion by saying that various kinds of intentions are individually necessary for the production of a speech act. His analysis also gives an impression that together a l l the intentions are sufficient for the production of a successful speech act. These conditions of necessity and sufficiency need discussion. We shall leave the intentions to communicate in general and the intentions tied up with 'the sincerity condition' for later discussion and first focus on the intentions which are brought in simply to explain the notion of meaning in a Gricean manner. Let us try to analyze a case of a purported promise with the help of Searle's account of what i t is to make a successful and non-defective promise and also with due regard to Austin's notion of infelicities of a speech act. Suppose that a well-intending father promises his son to take him to the circus in town. However, suppose that the circus is no longer in town, and that the father does not and cannot know this at the time of promising. The question i s : is this or is this not a successful or felicitous act of promising? Searle's (1970;57) account consists of nine conditions such that each condition will be a necessary condition for the successful and non-defective performance of the act of promising and taken collectively, the set of conditions will be a sufficient condition for such a performance. 81 Thus he says: Given that a speaker S_ utters a sentence T^  in the presence of a hearer E, then, in the literal utterance of T_, S_ sincerely and non-defectively promises that £_ to E i f f the following conditions obtain. 1. Normal input output conditions obtain, i.e., both S_ and H are speakers of a common language, are attentive, not asleep, or deaf, or dead, etc., etc. 2. _S expresses the proposition, that JJ in the utterance of T_. 3. In expressing that S_ predicates a future act A of S_. 4. II would prefer S_'s doing A to his not doing A. and S_ believes H would prefer his doing A to his not doing A. 5. It is not obvious to both _S_ and R that S_ will do A in the normal course of events. 6. intends to do A. 7. _S intends that the utterance of T_ will place him under an obligation to do A. 8. _S intends (i-1) to produce in H the knowledge (K) that the utterance of T^  is to count as placing S_ under an obligation to do A. _S intends to produce K by means of the recognition of i-1, and he intends i-1 to be recognized in virtue of (by means of) H's knowledge of the meaning of T. 9. The semantic rules of the dialect spoken by S_ and H are such, that T^  is correctly and sincerely uttered i f f conditions 1-8 obtain. Now, in our case, a l l the above conditions obtain. The speaker has the proper intentions to communicate, to be sincere, and he also 82 uses the conventional formula "I promise." in his utterance. According to Searle's biconditional i t ought to be regarded as a sincere and non-defective promise. But is i t really a non-defective promise? We find varied answers to this question from Austin, Searle and Alston. Austin seems to be in two minds on this point. He first classifies mistakes and misunderstandings as "a type of infelicity to which a l l utterances are probably liable" (1967., 41) and says that other inf e l i -citous cases may overlap with cases of this type. But then he goes on to say that "...mistake will not in general make an act void, though it may make i t excusable." (Ibid.; 42) He distinguishes the type of 'bad' advices, 'incorrect' or 'unjustified' verdictives from either 'misapplied' or 'insincere' ones and says about the former type that " s t i l l i t is not infelicitous in any of our senses; i t is not void (if the umpire says 'out', the batsman is out; the umpire's decision is final.) and not insincere." (Ibid.; 43) He concludes that to consi-der cases of misunderstandings is to introduce "an entirely new dimen-sion of criticism." What is the force of Austin's claim that a mistaken verdict is not infelicitous or void? Perhaps it is reflected in there being no breach of the practical consequences of?the verdict. He says that the umpire's decision is final. Maybe that is so. For this may be a case where a mistaken utterance is irrevocable due to some regard for a formal authority which is conventionally vested in the post of an umpire. But can this consideration be sufficiently generalized so as to be ap-plicable to other varieties of illocutionary acts as well? Can we say 83 that the ill-informed father's promise to his son in our story was not void or infelicitous? In Austin's case, the batsman was out, and the mistaken verdict was nevertheless effective. In our case, the promise cannot be ful-filled and this might seem to indicate that there is a difference in the two cases. But this is not really a significant difference, because we can imagine circumstances in which even i f the circus was in town, and the father knew about i t , the promise could not have been fulfilled.^, for before fulfilling i t the father may have died. So the absence of the practical consequences cannot be attributed solely to the father's ignorance of the circumstances. However, the difference goes deeper than that. It is reflected in the distance between different types of illocutionary acts: some of these can be brought out by uttering a more or less fixed formula in certain contexts on the mere strength of relatively formalized conventions about certain 'performances'; while some others do not have fixed formulas, nor formalized circumstances, nor a certain authority vested in the post of the speaker. These latter are farther and farther removed from the explicit performatives so far as the condition of their 'effectiveness' is in question. It is perhaps because a l l the speech-act analysts generalize from the relati-vely formalized forms of communication to a l l communication that these difficulties lie hidden in their analyses. Maybe one can go along with Austin in saying that the mistaken verdict was not void, and the batsman was out when the umpire said 'out'. But can we say the same thing about the father's promise? 84 First of a l l , unlike the unpire's verdict, a promise is not immediately glued to its effectiveness on the audience. The umpire's verdict is carried out at the instant by the batsman being out; the priest's wedding-pronouncement takes effect in that very instant in binding the bride and the groom in wedlock; but the father's words "I promise." put only him under an obligation, so there is a difference in the cash value of the 'effectiveness' which is also connected with the concept of being void. One might want to say that in the case of the promise in question, that it was not void, although mistaken, i f the father considers himself under obligation. But this is already granted. Does that solve our difficulty? The father considers himself under obliga-tion also because he is under the mistaken belief that the circus is in town, and that he can take his son to the circus. So i t would appear that i t is his belief rather than his intentions that plays a vital role in deciding the question of the felicity of his: illocutionary act. Perhaps i t is because of such a consideration that Searle (1970, 60) remarks, in sincere promises, the speaker believes i t is possible for him to do the act (or refrain from doing i t ) , but I think that the proposition that he intends to do i t entails that he thinks i t is possible to do (or refrain from doing i t ) , so I am not stating this as an extra condition. One may grant this and agree that i t was a sincere promise, but the question s t i l l remains: was i t a completely nondefective promise? It seems odd to call i t a nondefective promise, and yet i t f u l f i l l s a l l the given conditions of a full-fledged promise! As another try, one may argue that a l l i t takes to count an 85 utterance as a successful promise is not that certain circumstantial conditions obtain, but that the speaker believe that they obtain, and that his belief should be reflected in his intention. It is in this vein that Alston says in the context of 'making a request' that even i f the door is already open, i t is not logically impossible for me to ask you to open i t . For I might have been under a mistaken impression that i t was closed. In that case, you would not deny that I had made the request in question. On this analysis then proper beliefs entailed by the speaker's intentions appear to be sufficient for counting an utterance as a suc-cessful, not void illocutionary act of a certain type. But on closer look, i t turns out to be not so simple after a l l . It seems that for Austin and for Searle, at least in some cases, intentions peppered with beliefs are not sufficient for the performance of a successful speech act unless the corresponding circumstances actually obtain. Thus i t is not sufficient for a promiser just to believe that the hearer would prefer his doing A to his not doing A, but the hearer must actually prefer £'s doing A to his not doing A. (See Searle's condition 4 in the above list.) Or to take Austin's example, you may pick George as your partner in a game because you are under a mistaken impression that he is playing. But i f he says, "I am not playing.", then you have not picked George, in spite of your well-intended and sincere illocutionary act. So i t appears that over and above intentions and beliefs of the speaker, at least some circumstantial conditions must be fulfilled. Which ones deserve this honor remains unsettled or where i t is settled, i t appears arbitrary. 86 If mistaken verdicts or unjustified pronouncements are to be regarded as nevertheless not void or infelicitous, one wonders whether the same applies to other illocutionary acts as well. Take for ex-ample, 'side with'. This is Austin's example of the illocutionary act-verb of the 'commissive' type. Suppose that in a debate on' abortion I say "I side with George." believing as I say this that George is the one who is for the legalization of abortion. As i t turns out, I was mistaken in my belief, i t was not George but John who was for the issue. Do I s t i l l on the force of my utterance, side with George? In other words, is my utterance, like the umpire's decision, final? Maybe Austin would call myiutterance 'excusable' rather than 'void'. But what is the difference? I would rather claim that my utterance in question be null and void, so that I can side with John a l l afresh. Correcting your mistaken utterance is not so bad as simply changing sides with the whim of your fancy. Consider another case: I send Indira next door to see i f the neighbour is at home. Indira comes and informs me that the neighbour is not at home. Suppose that she tells me that because she sees the 'out' sign on the neighbour's door which was there, say, because of a mischief of the neighbour's son, or her oversight, or the wind or what have you. Can one not say that Indira misinformed me? If one grants that not a l l acts of misinforming someone are done maliciously, and that there can be a sincere act of misinforming, i t is not clear whether to say that in the above case Indira performed a full-fledged sincere illocutionary act of misinforming or an infelicitous, 87 unsuccessful illocutionary act of informing. The point is that the extent to which the speaker's knowledge or beliefs about the relevant circumstances determine the felicity (or even individuation) of an illocutionary act cannot be ascertained. In the absence of a nonarbitrary criterion for realizing the above notions, the talk about the speech act as the basic meaningful unit of communication is not illuminating. Let us now consider the intentions associated with the sincerity condition in the illocutionary act of promising. This should t e l l us more about the possibility of generalizing from this one case to other illocutionary acts. Condition 6 of Searle's nine conditions for making a sincere and non-defective promise is that the speaker intends to do A. In extending this analysis to other illocutionary acts he remarks that the appropriate sincerity condition for the illocutionary act of giving an order is that the speaker wants the ordered act done; for assertions, it is that he believes i t to be true. Searle also says that for 'greetings' there is no sincerity condition, and that there is no pro-positional content. This last remark, I think, needs modification. It may be that "Hello" (Searle?s example) has no propositional content and no sincerity condition. But it does not seem correct to say that a l l other greetings have no propositional content or the sincerity condition. Greetings such as "Good morning", "Merry X'mas", or "Happy Birthday" do seem to have different propositional contents, and i t seems that one could say any one of these without meaning i t , i.e., insincerely. One could say 88 "Happy Birthday" to someone while hoping that he have a rotten day instead. If in the above case, one argues that these expressions have the sincerity condition because they have some sort of propositional content, then i t would appear that having a propositional content is invariably linked with having a sincerity condition. But even this contention needs to be modified, for Searle claims that only where the act counts as the expression of a psychological state is insincerity possible. Thus he says, that one cannot christen insincerely, but one can promise or state insincerely. However, i t is not clear what i t is to be insincere in some illocutionary acts even where the act counts as the expression of a psychological state. Take for example, the i l l o -cutionary act of apologizing in saying "I am sorry.". One can say that to apologize in saying "I am sorry." counts as an expression of one's regret for one's own act. (Austin, 1970, 247) But what is i t to apologize without sincerity? It seems that whether or not we mean i t , to say "I am sorry." in appropriate circumstances, i.e., when one has done something wrong to another person, is to apologize. When a child, who has disobeyed and thereupon asked to apologize parrots reluctantly, "I am sorry.", i t is counted as a valid apology, even though he may mutter silently, "but I ain't.". It begins to look.as i f there is no criterion to count or even detect sincerity except for its hypothetical link with the proper in-tentions. Intentions tied up with sincerity are not necessarily re-flected in the associated behavior of the speaker nor can they be 89 unmistakably tested by looking at the relevant circumstances. For when one says, "I think i t is going to rain soon.", how do we t e l l whether or not he says i t sincerely? His answer is the last word. There need not be clouds in the sky at the time of his utterance. For even i f i t is a perfectly sunny sky, without a sign of a cloud, s t i l l his utterance could conceivably be a sincere one; because in the past, he may have experienced a correlation between a bright sunny sky and a sudden change in the atmosphere, bringing clouds and rain. So i t ap-pears that the only thing that is necessary for the sincerity of the 16 speaker in an utterance is that he have proper intentions. As to the question 'intentions of what sort?' there seems to be no answer that may suit a l l illocutionary acts, other than the vacuous one, 'to mean what one says'. This answer would be a circular one for someone like L.J. Cohen who maintains that the illocutionary force is just another aspect of the whole meaning of an utterance, and i t would be a vacuous one for Austin or Searle who try to separate meaning of an utterance from its illocutionary force, because the sincerity condition for them is linked up with the attitude reflected by the illocutionary force rather than the propositional content, and hence, 'to mean what one says* does not really touch the illocutionary force at a l l . 16 It is in keeping with the above line of analysis that the speechcact of lying can be said to require as a necessary condition the intention to deceive (mislead, misinform) the hearer with the utterance of a pro-position £ as true when the speaker believes i t to be false. Therefore in a case where the speaker's intention to deceive and his mistaken belief about the facts in question cancel each other out, producing a true assertion, we would like to maintain that the speaker li e d , al-though what he said happened to be true. 90 Therefore i t is not clear how one can cash the intentions associated with the sincerity condition so as to account for the proper i l l o -cutionary force. Let us dwell a bit more on the intentions 'to mean what one says'. Let us suppose that an analysis of 'to mean what one says' in Gricean terms avoids circularity, and see whether these intentions are necessary for the successful performance of a speech act. The analysis of speech acts is supposedly a retroactive analysis. That i s , i t takes an actual utterance by the speaker as its starting point and goes back to the con-ditions that must have helped it become a successful speech act. In such an analysis circumstantial evidence is a l l we have to rely on, and that can at times be dubious. The notion of success or felicity of a speech act is not clear., It seems to oscillate between moral and pragmatic (rather, amoral) considerations, as we shall see in the following discussion. Let us see how a single utterance of a sentence can be taken as a performance of different illocutionary acts. It is granted that an utterance token (including its phonetic aspects such as stress, intona-tion) may on one occasion have many different force potentials asso-ciated with i t . To take Searle's (1970, 70) example, i f , at a party, a wife says, "It's really quite late.", that utterance may be a state-ment of fact, an objection to her interlocutor's remark on how early i t was, or a request to her husband to go home. Searle's reason for this kind of one-many relationship between the utterance and the i l -locutionary acts is that the same utterance act may - be performed with 91 a variety of intentions. Thus i t seems that a l l the above illocutionary acts must first be intended by the speaker, or otherwise they would not be the illocutionary acts of_ the speaker in that utterance. However, there are times when we witness a gap between what the words mean and what the speaker may mean by the utterance. ('Mean' here is used in the wider sense so as to include even the 'force'.) One does not find the established conventions of meaning violated in this kind of event. Since, ex hypothesi, different illocutionary force potentials are already associated with a sentence of a language, on any one occasion of the utterance of the sentence, the speaker may intend to achieve only one of the several associated force potentials. But what i f the other associated force potentials somehow get activated and other illocutionary effects are achieved on the audience? Can we regard these others as successful illocutionary acts of the speaker? That i s , suppose in the above case, the wife intended only to decline an offer to stay and dance, but suppose her husband takes her to be reminding him of an early appointment the next morning, and requesting to go home^  He acts accordingly, and later when he thanks her for reminding him by that utterance of his appointment, she accepts the thanks without any clarifications of the lack of the specific intention on her part. Do we or do we not regard the illocutionary acts of requesting and re-minding as having been performed in the utterance of that sentence on that occasion? In practice, we care l i t t l e about individuating or pinpointing the speech acts. In theory, i t is not so clear that we have any hard arid fast rules to rely upon. In the above case, one feels 92 that since there was no specific intention on the part of the wife to remind the husband with the utterance, she did not perform the i l l o -cutionary act of reminding him although he took her to be reminding him. There seems to be a mismatch here between the speech act of the speaker and (if there is such a thing,) the hearing act of the hearer. What is not clear i s : in cases like the above where there is no breach of communication but there is a mismatch between the speaker's intention and the outcome on the hearer, which is the decisive factor in characte-rizing 'what happened'? Maybe we can try to patch up the theory of speech acts by distin-guishing between different ways of reporting an event. Maybe in the above case, one would want" to say, that the husband was reminded by her rather than saying that she reminded the husband. Maybe we can find some more cases of the sort where one way of reporting appears to be more accurate than the other. Maybe we can also adduce a tentative con-clusion that i f intention of the speaker is an essential aspect of an event, then we would be willing to infer the active voice report from the passive voice report, whereas in other cases, we will not be so willing. [l am not sure this is going to work: some wrinkles are al-ready showing. Consider the cases where although we do not attribute any intentionality to the grammatical subject, we are more willing to use the active rather than the passive report. For example, (1) The letter promised him a good salary, seems better than either (la) He was promised a good salary by the letter. 93 or (lb) A good salary was promised to him by the letter. (la) is simply awkward. (lb) almost forces us to read 'by the letter' as an instrumental phrase rather than an agentive phrase, and suggests an agentive insertion such as 'by someone' to complete the sense. So also (2) The sign said, "Smile. I like you.". Certainly no hint of intentionality on the part of 'the sign', but the corresponding passive (2a) "Smile. I like you."was said by the sign, sounds rather odd. Maybe these wrinkles can be smoothed out with effort, maybe not.] However, even i f we succeed in finding some cases and somehow link the feeling of propriety felt about one form of report rather than the other with the presence or absence of intentionality, where do xre go from there? Perhaps we will s t i l l be faced with a multitude of speech acts; some with the "proper" intentions on the part of the speaker, some without, some with the "proper" outcome on the hearer, some without, some with both on the part of both the speaker and the hearer. Which one of these gets called "the basic unit of linguistic communication" and why? The question is s t i l l open. To come back to our point, i t is not clear what intentions one should take into account when analyzing the notion of a speech act retro-actively. Take the case of the father who with the intention of intro-ducing his son to the stranger utters the words, "Meet my daughter.". The stranger understands the meaning of his words, and acts accordingly, i.e., says "Hello" to the daughter who is also standing by. The father swallows his words, accepts the hearer's response as appropriate, since 94 he was going to introduce his daughter to the stranger anyway, and then proceeds with the conversation. Was the father's first utterance of "Meet my daughter." a successful illocutionary act of 'introducing his daughter' or not? On the one hand i t seems to have a l l the elements of a genuine speech act of introducing one's daughter, including the usual outcome on the hearer. On the other hand, although the father has the intention to introduce his daughter to the stranger, he does not have the particular intention (so the story goes,) at the time of the utterance. It is only after the speech performance, even after the hearer's response that the speaker's intention to introduce his daughter i s , as i t were, charged and brought into contact with the by-gone utterance. And for a l l practical purposes, that would serve to make that speech act of his a successful speech act of introducing his daughter. The question i s : which intentions do we take into account when individuating a speech act? "Speaker's intentions" is not a sufficient answer. Even "speaker's intentions at the time of the utterance" is not a satisfactory answer. Because, as in the above case, his intentions in general may dominate over his Intentions at the time of one particular utterance, may influence our characterization of the resulting speech act. The point about fixing an optimum modulus for the utterance is relevant. On the one hand, there are moral issues connected with lying, being responsible, keeping one's word, etc., that are tied up with the intentions at the time of an utterance, and on the other hand-i there are numerous pragmatic (or, a-moral) considerations that are more prominent 9 5 in certain conversations where those moral questions are subdued to a great extent or are totally out of place. These two opposite forces can be working on an utterance, driving i t in two different directions: of the speaker's intentions tied up with his beliefs, and of the hearer's understanding. It is generally believed that both these are relevant for determining a speech act in an utterance, but what is not clear i s , how, i f at a l l , either one or both of them can help us to decide the questions regarding the individuation (let alone felicity) of a speech act. In the above case, we saw that there was a noticeable gap between what the utterance meant and what the speaker intended to convey on the particular occasion. The hearer's understanding the illocutionary force of the utterance also was a function of the hearer's knowledge of what the words uttered by the speaker mean (in general), and not a function of knowing what the speaker's intention in uttering these words was. Cases like the above suggest that the attempt by some philosophers to explain the meaning of a sentence in terms of its use in performing a certain illocutionary act will have some awkward consequences. Such an analysis generally demands that the speaker's intention to perform an illocutionary act x be a necessary condition of the performance of an illocutionary act x, which illocutionary act is then used to decide questions of sentence meaning, sentence synonymy, and the like. How-ever, i f the hearer's understanding of the propositional content of the utterance (or, the meaning of the sentence,) is sometimes sufficient for him or anyone to take an utterance as a performance of a certain 96 illocutionary act, then this makes unnecessary the condition that the speaker must have the intention to perform the particular illocutionary act by uttering the sentence. If i t is possible that sometimes the speaker's intentions and the hearer's understanding may collide in such a way as to make the hearer's understanding dominant without any noti-ceable breach of communication, then-this proves to be an objection to the role of intentions in defining a speech act. Evidently, the attempt to characterize the concept of an illocutionary act in terms of the speaker's intentions (alone or coupled with the hearer's understanding of that intention,) fails to supply us with reasonable mark for indi-viduating an illocutionary act; and i t also suggests that an analysis  of what a speech act is will depend on an analysis of what the meaning  of a sentence is rather than supply foundations for i t . A similar point can be made about the intention to communicate. Austin's account of the speech acts originates with the notion of a performative utterance. Searle and Alston analyze one illocutionary act such as promising or ordering and they believe that their analysis of one illocutionary act-can be generalized so as to cover a l l other illocutionary acts. Their choice of such examples is crucial for their analyses, which yields the intention to communicate as one of the necessary conditions for the performance of a speech act. But upon close examination, i t is not so obvious that the intention to let a particular hearer or even an audience recognize something by uttering a sentence is a necessary con-dition for every act of speaking. Austin's categories of the illocutio-nary act verbs called verdictives, expositives, and behabitives supply us with a good number of examples of the illocutionary verbs such that 97 the intention to communicate does not seem to be a necessary condition of their proper use. The l i s t contains verbs such as the following: assume, take i t , understand, reckon, bless, curse, regard as, inter-pret, postulate, deduce, etc. It looks as i f one could not only do the activities denoted by the above verbs without the necessity of there being a hearer, one can even use these in the first person sin-gular present indicative active formula supposedly so characteristic of the illocutionary act verbs and yet not have any intention to com-municate with anyone. While watching the bombing in Viet Nam on the T.V. someone says, "Curse that President!". The speaker may not have any particular hearer to whom the remark is addressed. He may simply express himself. Or again, while solving a mathematical problem in solitary confinement a prisoner says, "I understand.". Again, no in-tention to let any hearer recognize anything by the utterance. Or suppose, while reading a detective novel I come upon a page where I see through the story and say, "It must be the doctor.". One may characterize this speech act of mine as assuming, concluding, infer-ring or stating as one may like. But what is clear is the fact that I have absolutely no need for a hearer in this speech situation or no necessity to communicate. Austin is doubtful about the inclusion of verbs as 'know,' 'wonder,' doubt' etc., in the group of illocutionary ' act verbs and one reason for his doubt may be thatcthese verbs do not seem to require a hearer. In other words, these are the verbs of what is called the propositional attitudes. Therefore they do not necessarily have the speaker's intention to communicate as a precondition for their use. 98 One reply to the above criticism would be to say that the counter-examples are a l l of the form in which the speaker talks to himself, and therefore do not exemplify the normal use of language. One may choose to regard a l l such cases as deviant in some sense or as not standard use of language. But this reply is not satisfactory for two reasons. First, i t is not obvious that the purpose of language is exclusively to communicate. Language was, is and will be used also to express oneself, and expressing oneself (which consists of several other activities than merely evincing emotions,) does not seem to be odd or deviant in any sense i f there is no intention on the part of the speaker to communicate. Secondly, one cannot rule out a l l talking to oneself, or writing a poem or writing in a private diary as non-standard use of language. Although one may argue that language has to be learnt first and used in the communica-tional situation, and that historically communication must have pre-ceded private expression, that does not make the use of language for expressing oneself without communicative intentions deviant. One reason for this is the fact that soliloquies can be overheard and private diaries and personal poems can be read and understood by others. That is to say, a piece of every such non-communicative use of language can also play a double role, as i t were, and can be understood as one would an instance of intended communication. If this is granted,then one can see that except for the speaker's intention to communicate, there need be no difference in a sentence as i t appears in my 'thinking aloud', and in my conversation with you. 99 This suggests that we cannot really restrict the class of the so-called illocutionary acts to instances of intended communication, because at every turn, cases from the 'non-standard' field of language may step into the restricted territory of intended communication, and for a l l practical purposes they are taken as instances of intended com-munication. The fact that some utterances carry dual citizenship in-dicates that there must be something common to both the 'standard' as well as Vion-standard' uses of language which makes these utterances pass for both. The common factor here is clearly the meaning of the  sentence which could be understood by the hearer, without knowledge of the speaker's intentions. This fact creates difficulties regarding the question of deciding what speech act was performed in an utterance of a sentence. It also creates difficulties for anyone such as Alston who believes that he can account for what meaning a sentence has in terms of what illocutionary act i t is used to perform on any one occasion. Since the meaning of a sentence turns out to be knowable quite indepen-dently of the speaker's intention, or of the knowledge of what speech act was intended by the speaker in the utterance of the sentence, as least from this point of view, a sentence appears to be a more basic unit of meaning than a speech act. The Speech Act-primacy Thesis; Alston's Version There are at least two different versions of the claim that the speech act is the most fundamental unit of language. Searle, (as we shall see later) holds that speech acts are the basic units of language while at the same time claiming that the speech act or acts performed in the utterance of a sentence are in general a function of the 100 meaning of a sentence. (Searle, 1970, 18) Alston, while he agrees with Searle in holding the speech act primacy thesis, claims exactly the opposite about the relationship of the sentence to a speech act. In short, Alston believes that the meaning of a sentence is dependent upon or a function of the illocutionary force or the illocutionary-act potential attached to the sentence. Alston's two claims, namely, that (1) speech acts are the basic units of language and that (2) the meaning of a sentence is dependent upon or a function of the illocutionary-act potential attached to the sentence are at least theoretically consistent. That i s , the foamier can be said to follow from the latter, and the latter can serve as a sufficient reason for the former. Searle's claim about the primacy of the speech act as the meaningful unit of language on the one hand and the claim about the speech act itself being a function of the meaning of a sentence on the other hand are apparently inconsistent. How can the significance of the so-called basic meaningful unit of lan- guage be dependent on another meaningful unit? Upon examination, Searle is right about the speech act being a function of the sentence-meaning, and wrong about the primacy of the speech act as the meaning-ful unit of language. Alston's claims, interlocked as they are, f a l l hand in hand. But before we see the end of the story, let me summarize Alston's position f i r s t , and then offer an argument against i t . Alston's theory about the speech act is developed in his book Philosophy of Language (1964) and in an essay "Meaning and Use"17. The 17William P. Alston, "Meaning and Use", Philosophical Quarterly, Vol. 13 (1963), 107-124, rept. with revisions arid additions in G.H.R. Parkinson (Ed.) The Theory of Meaning (London: Oxford University Press, 1968). 101 statements of his theory in both these works are almost identical. His claim of the primacy of the speech act, or more precisely, of the illocutionary act, is based on his interpretation of the Wittgens-teinian doctrine: the meaning of a linguistic expression is a function of its use. On Alston's interpretation, 'use' in the above sentence becomes equivalent to 'the illocutionary force' or 'the illocutionary-act potential'. Thus he gets the result that the meaning of a sentence is a function of its illocutionary-act potential. Alston elucidates the above thesis also by way of considering what i t is for two sentences to have the same meaning. ...the fact that two sentences are commonly used to perform the same illocutionary act (have the same illocutionary-act potential) is sufficient to give them the same meaning. A wider survey will re-inforce the impression that sameness of illocutionary act potential is what constitutes sameness of meaning for sentences. (Alston, 1964, 36) Elsewhere (Alston, 1968, 147) he starts from the simple formula, 'x' means y_ (the meaning of 'x' is yj) = df. 'x' and 'y' have the same use. He then takes 'have the same use' to be equivalent to 'are used to perform the same linguistic act of a certain type'. I shall try later to show that the notion of two sentences being used to perform the same linguistic act of a certain type is a very confused notion and that i f we lean hard against i t , i t breaks down. At this point, let me only mention that Alston is aware of the fact that we describe on different levels of generality what illocutionary act is performed in the utte-rance of any given sentence. For example, the illocutionary act per-formed in the utterance of the sentence 'What time is i t , John?' can 102 he described in various ways; i.e., as an illocutionary act of asking, of asking a question, of asking John a question, of asking John a question about time, and so on. However, Alston does express confi-dence in his way of analyzing by saying, ...given a level of generality, we can handle the concepts fairly well. (Alston, 1964, 40) He also feels that To the extent that this analysis i s , or can be made to be adequate, i t has the great merit of showing just how the fact that a linguistic expression has the meaning i t has is a function of what users of the lan-guage do_ with that expression. This result has been achieved by concentrating on the appropriate unit of linguistic behavior, the illocutionary act. If this is the line along which meaning should be analyzed, then the concept of an illocutionary act is the most fundamental concept in semantics and, hence, in the philosophy of language. (Alston, 1964, 39) Let us now see what Alston's claim about the dependence of the sentence-meaning on the illocutionary-act potential amounts to. Alston notices a fact about language, the fact that i f two different sentences (different in their words or maybe in syntax) are used by the speakers of the language to perform the same illocutionary act, say, of asking John what time it i s , then almost always, the two sentences turn out to be what people usually call synonymous, i.e., having the same meaning. The statement of this fact as i t appears here, involves a tautology, which will be explained below. Alston seeks to explain the above correlation by trying to determine which of the two correlates is dependent upon the other. He confesses his inability to find any cases in which sameness of meaning of two sentences clearly does not hang on the sameness of illocutionary-act potential. (Alston, 1968, 15) 103 This introspective result is expressed by him in the following formula: two sentences are synonymous i f one of them can be substituted in a context where the other is used without thereby changing the specific illocutionary act performed by the speaker in the use of the other. In conclusion, he submits that sameness of illocutionary-act potential is what constitutes sameness of meaning for sentences; or what is the other side of the same coin, that the meaning of a sentence is a function of its illocutionary-act potential. In fact, however, Alston's endeavor amounts to putting the cart before the horse. This is so because he has somehow confused two notions: one, a way of coming to know that one sentence means the same as the other, and two, the criterion of sentence-synonymy. Investigating what the speakers of a language cto in uttering a sentence is one way of coming to know what a sentence means; but that does not permit one to conclude that being used by the speakers to perform the same illocutiona-ry act is what constitutes having the same meaning for two sentences. An analogy will perhaps help us see what is involved. Suppose, my two brothers are exactly of the same age. One way I can come to know this is to find out the fact that they both are exactly three years older than me. But this does not allow me to conclude that their both being exactly three years older than me is what constitutes their both being of the same age. Saying this involves giving a very peculiar sense to 'constitutes*. By the same token then, I would be required to say that since both of them are exactly twenty-five years younger than my aunt, their both being exactly twenty-five years younger than her is what constitutes their both being of the same age. But this is wrong. 104 The fact that really constitutes their both being of the same age is their both being born on the same day of the same year. If they were both born on the same day, then and only then, would they have been exactly of the same age, even i f they did not have an aunt older than or a sister younger than them to compare their age with. In other words, the results of comparing their age with other people's age are only ancillary to the fact of their both being of the same age, and, therefore, cannot be legitimately regarded as what constitutes their being of the same age. Although knowing their comparative age-dif-ference from the others may help one determine their age, and/or the sameness of their age, their comparative age-difference from the others can only be regarded as a function of their age, and not the other way around. Let us now see why one cannot regard sentence-meaning a function of what the speakers of language do_ with the sentence in their utterance of the sentence. I shall produce an argument to prove what seems intui-tively clear. The main reason why one cannot say that the illocutionary-act potential constitutes the sameness of meaning for sentences is that one cannot even specify what the relevant illocutionary-act potential is unless and until one includes in i t a specification of what the senten-ces mean. In other words, the so-called analysans in this case cannot even stand on its own without taking a large measure of support from the analysandum and therefore i t cannot serve to explain the notion contained in the analysandum. If one undertakes to explain sentence-synonymy with the help of the sameness of the illocutionary-act potential, 105 one enters a vicious circle. Consider, for example, the following sentences: (a) "Have you had your lunch, John?" and (b) "Did you eat your lunch, John?" Now let us try to see how we can determine whether or not they have the same meaning by leaning on the sameness of their illocutionary-act potential. In order to do this, we first have to specify what i l l o -cutionary act is performed in their utterance. We can start from the cruder descriptions of the illocutionary act involved and then go on refining. Thus we can say that the act performed was of asking, of asking a question, of asking John a question, of asking John a question about his lunch, etc. It will be clear that many sentence—candidates can suit the description at the cruder levels, but that the candidacy becomes restricted as we refine the notion of the illocutionary act. However, as we go on, we come to an embarrassing stop, just one step before the fullest description of the illocutionary act involved in the utterance of the two sentences can be given. This happens because the fullest description of the illocutionary act involved must, of necessity, include a reference to the meaning of the two sentences, i f it is to help us in weeding out any other unsuitable candidates. In other words, every full-fledged illocutionary act must carry on its identity card its propositional content and the illocutionary force. The propositional content of an utterance plays an indispensable role in determining the identity of the illocutionary act in question. Until we specify the propositional content of the illocutionary act in question, we leave the door open for other linguistic expressions to come and claim 106 a seat in the house. In this case, for example, (c) "How was your lunch, John?" would suit the description of the illocutionary act one level before the full description. Once we say that the illocu-tionary act in question is the act of asking John whether he has eaten his lunch, we can be sure that the only sentences that can now claim a seat would be doing so by virtue of having the same meaning as specified by the propositional content of the act in question. But we never really can climb this step because therein we would be using the very same concepts of sentence-meaning or sentence-synonymy which we are seeking to explain. Thus we see that Alston's claims that (1) sameness of illocu-tionary-act potential is what constitutes the sameness of meaning for sentences and that (2) since sentence-meaning is a function of its i l -locutionary-act potential, the most fundamental unit of language is the speech act, both prove to be untenable. On the other hand, from what we have seen in the above case, we can safely support Searle's claim that the speech act or acts performed in the utterance of a sentence are in general a function of the meaning of a sentence. This in turn does cast a doubt on Searle's other claim, namely,that the speech act is the basic meaningful unit of language. More about that, below. Speech and Language Before proceeding to discuss Searle's version of the speech act primacy thesis, a brief note on one point is in order. The point is — the decision about the scope of the words 'speech' and 'language'. Normally, we distinguish between these two, although the exact nature 107 of this distinction may not be clear. Roughly, speech is the product of the activity of speaking, i.e., people making meaningful noises, whereas language is wider than that, as i t encompasses thought as well as speech, or at any rate, i t is not exhausted by the actual speech of its speakers. Searle attempts to obliterate the supposed distinction between 'speech' and 'language', and the consequences of this move are important for this theory, as we shall see below. Interestingly, we find a precedent for this kind of treatment of 'speech' and 'language' in Austin's work. Let us look into the matter more carefully. The general impression that one gets from reading Austin's How  to Do Things with Words is that Austin here is concerned with analyzing speech, that i s , the activity of people when they speak, or issue ut-terances. Thus he starts with instances of 'utterances' in a speech situation, examines the intentions of the speakers, the effects of the utterances on the hearers, and at the end, comes up with a classifica-tion of various kinds of 'speech acts'. All this is fairly clear. The question I wish now to raise i s , does he distinguish between 'speech' and 'language', and i f so, how? The answer is relevant to the topic with which I am concerned, namely, when people talk about the units of meaning, do they have in mind units of speech or units of language? Lecture VIII is one place to look, as here Austin talks about several units such as locutions, phemes, rhemes, illocutions, and per-locutions. I find that his views as expressed in this lecture are not clear, and at places I suspect there is a shift back and forth between 'speech' and 'language'. Austin starts with locutions and calls them 108 the full units of speech. I quote at length from his account of them: And we began by distinguishing a whole group of senses 'of doing something' which are a l l included together when we say, what is obvious, that to say something is in the full normal sense to do some-thing - which includes the utterance of certain noises, the utterance of certain words in a certain construction, and the utterance of them with a certain 'meaning' in the favourite philosophical sense of the word, i.e., with a certain sense and with a certain reference. The act of 'saying something' in this full normal sense I c a l l , i.e., dub, the performance of a locu-tionary act, and the study of utterances thus far and in these respects the study of locutions, or of the full units of speech. (Austin, 1967, 94) It is worth noting here that Austin describes locutions — 'the full units of speech' — as inclusive of the other abstractions such as the phone, the pheme, and the rheme. This is apparent from his elaborate description of 'the f u l l normal sense' of 'saying something' which incorporates the definitions of a phone, a pheme, and a rheme. However, later on he distinguishes between a pheme and a rheme by claiming that the pheme is a unit of language and that a rheme is a unit of speech. This distinction seems to contradict his early in-clusion of both phemes and rhemes into locutions which are according to him the full units ofcspeech. Let us look closely at the distinction between phemes and rhemes. This distinction is summed up In the following short passage The pheme is a unit of language: its typical fault is to be nonsense^meaningless. But the rheme is a unit of speech: Its typical fault is to be vague or void or obscure, etc. (Ibid., 98) The pheme is defined earlier as the utterance of certain vocables or words, i.e., noises of certain 109 types belonging to and as belonging to a certain vocabulary in a certain construction, i.e., confor-ming to and as conforming to a grammer, with a certain intonation, etc. (Ibid., 95) and the rheme is the utterance of a pheme using these vocables with a certain more or less definitive sense and reference, (which together are equivalent to 'meaning'.) This amounts to saying that a pheme which is not a rheme is typically meaningless. It may have words of say, English in i t , and English word-order too, but no sense or reference. For example, "quadruplicity drinks procrastination" is a pheme which is not a rheme, for i t lacks sense. Another example would be the following: while testing a micro-phone I say, "That boy is chasing that g i r l " without having in mind any boy or any girl at a l l . My utterance is a pheme which is not a rheme, for the words therein lack references. Thus a pheme can f a i l to become a rheme by either failing to have sense or by failing to have references for its words. This amounts to being meaningless in one way or the other. A rheme is a group of words not only with a word-order but also with a certain sense and reference, therefore, always meaning-ful. But it may be vague or obscure so far as precision of the reference is concerned. For example, "That is beautiful" said while watching a swimming competition. So far this helps us to understand the general distinction between a pheme and a rheme, but i t does not give us a clue as to why the former is regarded as a unit of language and the latter as a unit of speech. Perhaps by 'language' he means the raw material, i.e., the lexicon and the grammatical rules, and by 'speech' he refers to the 110 actual activity of the speakers of this language, i.e., their actual operations with the raw material. This is a guess that seems to ex-plain things in this case but not elsewhere. Before leaving this puzzle, one note of caution. Austin's remark on p. 95, i f interpreted carelessly, may lead us to believe that phemes and rhemes both belong to the realm of speech: phemes to the direct speech and rhemes to the indirect speech. Austin's remark is as follows': 'He said "The cat is on the mat" ' reports a phatic act whereas 'He said that the cat was on the mat' reports a rhetic act. From this and other such contrasting pairs one may think (wrongly) that reporting a pheme is reporting an utterance retaining its original words and word order, etc., i.e., in direct speech, whereas reporting a rheme is reporting an utterance in the so-called indirect speech. On this interpretation the difference between a reported pheme and a re-ported rheme seems to boil down to the difference between the two modes of speech: direct and indirect. There does not then seem to be any conflict between this characterization of phemes and rhemes and their inclusion as locutions which are also called the units of speech. But this interpretation misses the point and thus confuses the issues. The use of either mode of speech is merely a device or a framework which serves to bring out the phatic or the rhetic aspect of any utterance. The use of either mode does not 'change' the original utterance into a pheme or a rheme. It merely 'presents' the original utterance as a pheme or a rheme. In other words, when a person chooses to report an utterance in direct speech, he reports the utterance in its phatic aspect, i.e., apparently without I l l concerning himself with its sense and reference. In this way one may report certain words of Latin exactly as he heard them, without knowing anything about their meaning; then a la Austin, he reported the utterance as a pheme. In order to report an utterance as a rheme, you have to take into consideration its meaning. That you do in fact pay attention to its rhetic aspect in the indirect speech report is shown by the changes in the original words, word order, etc., which sometimes occur though preserving the meaning of the original. Thus the two modes of speech are regarded by Austin as modes which serve to bring out a particular aspect of an utterance and therefore i t will not be proper to take the phemes or rhemes as themselves being the units of either mode of speech. This takes us back where we started. We tried to see in what way the distinction between a pheme and a rheme corresponds to a distinction between language and speech, and then in what sense one can justify Austin's inclusion of the pheme (a unit of language) and a rheme (a unit of speech) into locutions or the fu l l units of speech. The move which I suggested earlier that Austin might mean by •language' the vocabulary and the rules of syntax while by 'speech' the actual performance of the speakers is not very promising. In the first place, i t runs counter to our intuitive grasp of 'language' which we regard as having wider scope than 'speech', because we believe language to be inclusive of speech, writing, as well as thought. But apart from that, what is more important, Austin's own use of the word 'language' a few pages later, does not seem to be in any way similar to the use of 112 'language! we are attributing to him in this interpretation. Austin discusses the notion of an illocutionary act on p. 99 and in the course of examining i t as a way or a 'sense' in which we use words, he shifts from 'function of speech* to 'functions of lan-guage' and from 'use of speech' to 'use of language'. Again at the cost of being tedious, I quote Austin in f u l l , because this shift is important and seems to have influenced other philosophers. When we perform a locutionary act, we use speech: but in what way precisely are we using i t on this occasion? For there are very numerous functions of or ways in which we use speech, and i t makes a great difference to our act in some sense- sense (B) - in which way and which sense we were on this occasion using i t . It makes a great difference whether we were advising, or merely suggesting, or actually ordering, whether we were strictly promising or only announcing a vague intention, and so forth. These issues penetrate a l i t t l e but not without confusion into grammar (see above), but we constantly do debate them, in such terms as whether certain words (a certain locution) had the force of a question, or ought to have  been taken as an estimate and so on. I explained the performance of an act in this new and second sense as the performance of an 'illocutionary* act, i.e., performance of an act in saying something as opposed to performance of an act of_ saying something and I shall refer to the doctrine of the different types of function of language here in question as the doctrine of illocutionary forces. It looks as i f at the beginning of this discussion he treats even illocutionary force as a function of or a way in which we use speech. But at the end of the passage that gets transformed into a function of language, and this shift remains unexplained. Later on he frequently uses the phrase 'uses of language' to refer to the illocutionary acts or illocutionary force.(p. 100, 104) Even i f we give Austin a benefit of doubt and grant for a moment 113 that he did somehow intend to distinguish between locutions as functions of speech and illocutions as functions of language there is l i t t l e hope of getting clear on the nature of this distinction with the help of a similar unclear distinction between rhemes as units of speech and phemes as units of language. There he tried to tie up being the unit of language with the fault of being meaningless, but in the case of the illocutionary force, al-though he calls i t a function of language, he wants to separate i t strictly from the questions of meaning. Locutions have the qualities of being meaningful or meaningless, and illocutionary force is a notion carefully kept apart from considerations of meaning. Austin is clear on this point. Admittedly we can use 'meaning' also with reference to illocutionary force - 'He meant i t as an order' etc. But I want to distinguish force and meaning in the sense in which meaning is equivalent to sense and reference, just as i t has become essential to dis-tinguish sense and reference within meaning, (p. 100) Besides, earlier he characterizes the unit of language (the pheme) as mere words — vocables with a word order — but bereft of sense and reference which suggests a stage before the emergence of a meaningful utterance, but illocutionary force is something which must have a meaningful utterance at its basis. In conclusion, I have to say that at both these places (p. 98, 99) where Austin uses 'speech' and 'language' in close proximity, there is only a semblance of a distinction, but on a l l interpretations that I have come up with, i t seems to melt away into nothing but a curious slip. 114 The Speech Act-primacy Thesis; Searle's Version The edifice of Searle's detailed analysis of speech acts stands on the foundations of Austin's insights about performative utterances and illocutionary force. On the whole Searle follows the master in his blueprint of the structure of speech acts; introducing changes sometimes in the terminology, sometimes a bit deeper than that. Austin initiated the study of speech acts, or of "the total speech situation" in order to understand some problems of logic and ethics in their f u l l perspective. However, he did no more than to direct philosophical investigation towards certain utterances. He did not, to my knowledge, claim anywhere that speech acts are the basic units of speech, language or linguistic communication as such. Most of the conclusions in his lectures at Harvard were put forward as tentative and they served to point out that there are other things besides stating or describing that we do when we say something meaningful, and that besides the truth-falsehood dimension, utterances are also susceptible to the 'infelici-ties' that vitiate the performative aspect of utterances. In his last lecture of the series he suggests the following: "The total speech act in the total speech situation is the Only actual phenomenon which, in the last resort, we are engaged in elucidating." Taking his cue from Austin, Searle develops a detailed theory of illocutionary acts and in his writings we find specific claims about a certain primacy of speech acts. Two points deserve mention at this stage, before we get into deeper waters. The first is a rather general point about 'speech' and 'language*. We saw before how in Austin's lectures one finds what seems to be a shift between 'speech' and 'language'. In Searle's work, we 115 find a deliberate attempt to obliterate the supposed distinction between the two and claim that the study of the one is the study of the other. In this connection, consider the following remarks: (1970; p. 17) It might be objected to this approach that such a study deals only with the point of intersection of a theory of language and a theory of action. But my reply to that would be that i f myeconception of language is correct, a theory of language is part of a theory of action, simply because speaking is a rule-governed form of behaviour. [emphasis mine.] It might seem that my approach is simply, in Saussurian terms a study of "parole" rather than "langue". I am arguing, however, that an adequate study of speech acts is a study of langue. [emphasis Searle's]. I will have occasion to say more about the above claims below. At this point i t is worth keeping in mind that this general claim about 'speech' and 'language' is related to his claim about the speech act being the basic unit of linguistic communication. The second point is rather more specific. Austin seems to be very careful in keeping apart the concept of the meaning of an utterance from the concept of the illocutionary force attached to i t . Roughly his distinction between a locutionary act and an illocutionary act corres-ponded to a distinction between saying something meaningfully and saying something meaningful with a certain force. Of course he regarded the locutionary act as much as the illocutionary act as an abstraction only: "every genuine speech act is both"(Austin, 1967, 146) or "To perform a locutionary act is in general, we may say, also and eo ipso to perform an illocutionary act, as I propose to call i t . " (Ibid., 98) But the purpose of his insistence on this abstraction was to avoid what he called 'the descriptive fallacy', and to bring to the forefront the 'performance' aspect of 'saying something meaningfully', so as to make 116 oneself sensitive to the ways in which an utterance can be unhappy or infelicitous rather than simply true or false. Searle follows Austin in keeping apart 'meaning' and 'force', or as he chooses to call them, the 'propositional content' and'the i l l o -cutionary force'. In my general account of speech acts I tried to see why Searle objects to Austin's terminology of 'locutionary' versus ' i l -locutionary' acts which corresponds to Austin's meaning—force distinction, In its place Searle substitutes (1968, 420; 1970, 29) utterance acts and  propositional acts. One can represent the difference between Searle's and Austin's classifications as follows: Austin Searle L 0 Q Phonetic Utterance J Phatic 0 N A Rhetic Propositional 0 R Y Illocutionary Illocutionary Perlocutionary — Perlocutionary Searle's category of utterance acts corresponds to the phonetic and phatic acts of the locutionary act category of Austin. In place of the rhetic act of the locutionary category Searle substitutes proposi-tional act. The other two categories mostly stand as they are. Searle claims often that his suggested difference amounts to more than just a taxonomical preference and that his classification is not subject to the 117 objections he raised against Austin's classification. I do not find his claims fully substantiated in his account of propositional acts. Let me discuss the matter in brief. His account of propositional acts is as follows: We need to distinguish in the total illocutionary act the type of act from the content of the act. ...We need to distinguish the illocutionary act from the propositional act- that i s , the act of ex- pressing the proposition (a phrase which is neutral as to the illocutionary force). And the point of distinction is that the identity conditions of the pro-positional act are not the same as the identity condi-tions of the total illocutionary act, since the same propositional act can occur in a l l sorts of different illocutionary acts. ...in this form the distinction is not subject to the objections we made to the original locutionary-illocutionary distinction. The propositional act is not presented, either in the symbolism or in natural languages, by the entire sentence, but only by those portions of the sentence which do not include the iri^-dicators of illocutionary force. Thus the proposi-sitional act is a genuine abstraction from the total illocutionary act, and so construed, no propositional act is by itself an illocutionary act. (Searle, 1963, 420) The two arguments in favor of his distinction are: (1) The identity conditions for the propositional act and the illocu-tionary act are different. (2) The characteristic form of a proposi-tional act is not a full sentence but only those portions of the sentence which do not include the indicators of illocutionary force. Let us consider (1). The reason for the claim in (1) is that the same propositional act can occur in al l sorts of different illocutionary acts. Thus e.g., the proposition expressed in "Today is Saturday." may be expressed to state, assert, inform, object, affirm, etc. But was not the same true of Austin's locutionary acts? It would be fair to say that as in the above case, where there is no explicit performative 118 serving as the illocutionary force indicator in the body of the ut-terance, Austin's locutionary act and Searle's propositional act both stand on the same ground. For in cases like the above, the locutionary act — the act of saying something meaningful has just the kind of one-many relationship to the illocutionary acts that the propositional act — the act of expressing a meaningful proposition has. Similarly for the utterances that contain self-sufficient explicit performatives such as the following)} "I thank you from the bottom of my heart." or "I protest the b i l l . " . Searle's objection to Austin's distinction was, that in cases like the above, the account of the locutionary act necessarily mentions the illocutionary act which is 'uniquely determined' by the meaning of the sentence uttered. As i t turns out, the same can be said about Searle's distinction. It appears that the specification of the proposition expressed in the above two utterances would necessarily bring in a mention of the illocutionary act, or in other words, in des-cribing the 'content' of the acts, one would be forced to describe also the 'type' of act, i f the illocutionary act is taken to be 'uniquely determined' by the meaning of the sentence uttered. One possible move at this point would be to argue that since in cases like the above, there is no escape from the illocutionary act, let there be only the illocutionary act credited to the utterance and let us allow that there is no propositional act or no propositional content to be expressed. Searle does elsewhere insist, for instance, that the i l -locutionary act of greeting has no propositional content. But this move is not satisfactory; for we do differentiate between greetings. We do 119 not greet by saying "Happy Birthday!" at a Xmas party, unless i t is also someone's birthday, and we do not greet by saying "Merry Xmas!" on a Hindu New Year Day. So even where the illocutionary force seems to overshadow the content or the meaning of the utterance, we have some ways of differentiating between acts as belonging to one or many types on the bases of their content. Now let us turn to (2). Here Searle talks of'the characteristic grammatical form of a propositional act'. Negatively speaking, the form is said to be not a full sentence, but only those portions of a sentence which do not include the indicators of illocutionary force. Positively speaking, the form is described as the parts of sentence such as the gram-matical predicates for the act of predication, and proper names, pronouns and certain other sorts of noun phrases for reference. Further, Searle holds that the illocutionary force indicating devices in English include at least word order, stress, intonation contour, punctuation, the mood of the verb, and the so-called performative verbs. However, some of these also serve to identify and distinguish one proposition from the other. Take for example word order. Given that the illocutionary force, say of 'stating' is common to the following two utterances, "Searle criticizes Austin." and "Austin criticizes Searle." do not express the same proposition, and their word order distinguishes the one from the other. The same holds for punctuation, or the mood of the verb, etc. Thus some of Searle's illocutionary force indicators turn out to be the same as the propositional content indicators. So i t is not entirely true that "the characteristic form of the propositional act is only 120 those portions of a sentence which do not include the illocutionary force indicators", for the identity conditions of the propositional and the illocutionary acts do not turn out to be necessarily different. We are told that we can represent the general form of very many kinds of illocutionary acts as F(p) where the variable "F" takes i l l o -cutionary force indicating devices as values and "p" takes expressions for propositions. We are also told that i f not on the surface structure level, surely on the deep structure level we will be able to maintain a sharp distinction between the two aspects of the total illocutionary act. I do not see very clearly how we can achieve this even on that level for acts involving simply some self-sufficient explicit performative such as "I protest." or "Thank you.". But perhaps that may not be so important. What is important right now is the claim that Searle's distinction is not subject to the charges similar to the ones made by him against Austin's distinction. Searle objected to Austin's distinction by arguing that the locutionary act is not a genuine abstraction, because (1) with regard to the explicit performatives where the illocutionary force is uniquely determined by the meaning of the sentence, one cannot describe the locutionary act without specifying the illocutionary act, and (2) since no meaningful sentence is completely force-neutral, every locu-tionary act is an illocutionary act of some kind or the other. The first argument against Austin's distinction, as we saw before, can just as well be applied to Searle's distinction. For i f i t is true that the force of the utterances involving explicit performatives is 121 uniquely determined by their content, then on his own argument, one cannot describe the proposition expressed in the utterance without alluding to the illocutionary act as well. If i t is not true, then Searle loses his case against Austin, and they both stand on the same ground. We saw earlier why the second argument against Austin doesn't • work, since i t overlooks the asymmetry between locutionary and illocu-tionary acts with regard to the felicity conditions. But apart from the act terminology, i f the second argument is supposed to show that no locution is a genuine abstraction from the illocution, since some kind of very general illocutionary force is always built into the mean-ing of every sentence, then I fa i l to see why one cannot say the same thing about propositions, substituting "propositions" for the'meaning of a sentence" In the above type of argument. In conclusion, I submit that Searle's distinction fares no better or no worse than Austin's distinction: ithat at worst, they both are shaky with regard to the act-aspect, and that at best, they remain simply as theoretical abstractions between 'force' and 'content', not necessarily abstracting two classes completely exclusive of each other. Now let us dig out the proper sense of 'primacy' in which speech acts, or to be more specific, illocutionary acts are said to be the basic meaningful units of language. At the outset we can rule out two fairly obvious (mis)interpretations of the claim. (1) A speech act cannot be 'basic' in the sense of the smallest, atomic, indivisible meaningful unit of language, since by hypothesis, a speech act could be 122 divided into smaller meaningful units such as words, phrases, sen-tences, etc. (2) A speech act cannot be 'basic' in the sense of supplying foundations for other meaningful units such as words or sentences. In general, we saw why this line of approach is wrong in our examination of Alston's version of the primacy thesis. Searle is not guilty of this kind of charge, since he holds that an illocu-tionary act is a function of the meaning of the sentence. Thus at least for him, a speechsact is not 'basic' in the sense of forming bases for the meaning of words or sentences. In what sense then is i t 'basic'? Here we must reconstruct an argument for Searle's position on the bases of his remarks in section 1.4 "Why Study Speech Acts?" of his book Speech Acts. There are at least two different lines of^approach that Searle uses with regard to the question of primacy of speech acts. The dominant argument i s : the speech act or acts performed in the utterance of a sentence are in general a function of the meaning of the sentence. Supporting this line of argument is what he calls 'the principle of expressibility', which states that whatever can be meant can be said or, in slightly better words, that in principle each speech act that one performs or could perform could be uniquely determined by a given sentence or set of sentences,given the assumptions that the speaker is speaking literally and that the context is appropriate. The conclusion from the above argument supported by the principle of expressibility i s , that since the principle enables us to equate rules for performing speech acts with rules for uttering certain linguistic elements, a 123 study of the meaning of sentences is not in principle distinct from a study of speech acts. They are not two independent studies but one study from two different points of view. This is a very general line of argument and I shall touch upon i t in the course of discussing the second line of argument, only in-sofar as the first argument suggests that performing speech acts ex-hausts the functions of the meaning of sentences. However the above line of argument allows us to infer that from a different point of view, sentences could be regarded as the basic meaningful units of language, and therefore, this line is not as strong asSSearle's other line of argument regarding speech-act primacy. This other line of argument uses the following steps: [wording is mostly Searle's] (a) Speaking a language is engaging in a rule-governed form of behavior. (b) Speaking a language is performing speech acts, acts such as making statements, giving commands, asking questions, etc. All linguis-tic communication involves linguistic acts. (c) The unit of linguistic communication is not, as has generally been supposed the symbol, word or sentence, or even the token of the symbol, word or sentence, but rather the production or issuance of a symbol or word or sentence in the performance of the speech act. To take the token as a message is to take i t as a produced or issued token. (d) More precisely, the production or issuance of a sentence token 124 under certain conditions is a speech act, and speech acts of certain kinds are the basic or minimal units of linguistic com-munication. (e) Lastly, a theory of language is part of a theory of action, simply because speaking is a rule-governed form of behavior, and therefore, an adequate study of speech acts is a study of langue. This line of argument, as I said, is stronger in that i t attri-butes primacy to the speech acts in exclusion of other units of lan-guage such as words, sentences; i t is also stronger in claiming that a study of speech acts is a study of langue. There is a trivial sense in which any empirical study of lan-guage would have to incorporate a study of speech, for a l l our data comes from observing people's behavior when they use words in utterances to communicate or express themselves. However, Searle seems to regard speech as not just a starting point of a study of language; he holds that i f his conception of language is correct, then "a theory of lan- guage is part of a theory of action, simply because speaking is a rule-governed form of behavior." Thus, he seems to blur the distinction between 'speech' and 'language' to the point of suggesting that the domain of speech exhausts the domain of language. This attitude is reflected in a curious use of 'speaking a lan-guage1 in Searle's argument. Notice that in (a), 'speaking a language' is ambiguous between 'speaking' as an activity and 'speaking' as a facility. This ambiguity between 'speaking' (i.e., issuing utterances at time t in a language), and 'speaking' (i.e., knowing a language, 125 having i t at one's disposal when needed, etc.) is covered neatly by the vague general term 'behavior' which encompasses both 'acts and dispositions'. However at stage (b), the facility aspect of 'speaking a language' slips away and 'speaking a language' is simply equated with 'performing speech acts'. In (e), even the ambiguous phrase 'speaking a language' gives way to a clearer, activity-denoting word 'speaking', on the strength of which i t is claimed that the theory of language is part of the theory of action. Thus the argument overlooks a general ambiguity about 'speaking a language'. 18 In this connection, some insights from Zeno Vendler's paper "Say What You Think" deserve mention. The main contribution of Vendler's paper is to remind us that language encompasses both speech and thought, and that propositions are the common objects of these two closely related yet distinct processes. "Thought in the strict sense of the word, is inconceivable without the idea of language, and the use of language, in the full sense of the word, is essentially the expression of thought." Vendler argues that the same things (i.e., propositions) that can be asserted, stated or denied inispeech can be realized, understood, believed, suspected or doubted in thought; that what one may decide to do, want, wish or intend to do in thought, one can also promise, vow or pledge to do in speech; what you wonder about in thought, you can ask about in speech. As Vendler puts i t , "you can say whatever you can think and you can think almost whatever Zeno Vendler, "Say What You Think",..^Studies, in Thought and Language (Ed.) J.L. Cowan. (Tucson, The University of Arizona Press, 1970) 126 you can say." [The first half of this formula roughly corresponds to Searle's 'principle of expressibility'. However, i t is the second half that is important in the sense that i t serves as a reminder of the fact that although very closely connected, speech and thought are not exactly identical domains. One indication of this as pointed out by Vendler i s , that the vocabulary of types and items of performatives (or that which you do_ in an illocutionary act) is richer than the vocabulary of thought, since speech is a primary phenomenon or to use Vendler's expression, thought is the suppression of speech.] It is with this insight into the close affinity between the domain of thought and the domain of speech that Vendler improves upon Austin's study of performatives. He carefully isolates the genuine performatives — a l l of which are 'achievement verbs', from Austin's original l i s t of illocutionary act - verbs. He contrasts the genuine performatives or 'the verbs of saying' with what he calls 'the verbs of thinking'. Vendler's verbs of thinking are verbs such as 'know', 'believe', 'remember', etc. — the verbs about whose inclusion into the category of the illocutionary acts Austin had some doubts. Since both types of verbs show preference for simple present tense, Vendler classifies them together as belonging to the wider class of al l 'propositional container verbs'. But the main difference between the verbs of saying and the verbs of thinking is their time schema. Whereas the simple present tense in the case of a performative (e.g., "I promise.") singles out the moment at which the illocutionary act occurs, in the case of a propositional attitude verb (e.g., "I believe.") 127 the same tense does not indicate a unique moment, but an indefinite time span which includes the moment of utterance. In other words, all performatives are achievement verbs, but certain verbs of thinking are state verbs or attitude verbs. The issuing of a performative utterance with a certain force — the illocutionary act-occurs at a given moment and does not last through the period of time. However, since verbs of thinking can be both state verbs and achievement verbs, a proposition can be realized at time _t, or believed In for over a period of time. We saw above how the performatives as achievement verbs differ in their time schema from the propositional state verbs, but i t is interesting to note that the performatives also differ in this respect from the other propositional achievement verbs which are verbs of thinking. The difference consists in their behavior with respect to the present tense. With the performative the first person present indicative active form is the most natural, most characteristic oc-currence. Members of the other group, however, do not accept this form except when scope-indicating grammatical adjuncts are present. Thus, for example, "I promise to pay you ten dollars." is perfectly normal; but not so "I decide to go home." unless i t occurs in a modified form such as "I always decide after consulting my wife." etc. Another indication of the difference between the two types of verbs (namely, the verbs of saying and the verbs of thinking) is found in the auxiliaries used for reverb alizing tha nomilalized verbs of each 128 type. The performatives when nominalized, (e.g., a 'promise', an 'order') take act-indicating auxiliaries such as 'give', 'make', 'issue'; whereas the propositional attitudes (e.g., a 'belief', a 'doubt') are tied with stater-indicating auxiliaries such as 'have', 'adopt', 'hold*, 'cherish', etc. Although the auxiliaries used in reverbalizing the nominalized verbs can be regarded as metaphorical, they are nevertheless useful in pointing out the different modes in which language is used in thinking and in saying. This analysis provides us with an important insight into the total domain of language which encompasses both speech and thought. The above evidence strongly suggests that although speech and thought are very closely related and sometimes overlapping domains, they are nevertheless distinct. The distinction between them is not just the trivial 'silent/spoken' distinction, but i t is a deeper distinction in the very nature of the two activities, reflected in the logical grammar of the verbs of saying as opposed to the verbs of thinking. What is most important from our point of view is that Vendler's analysis of language raises a big doubt about Searle's claim that the study of speech acts as the basic units of linguistic communication is a study of language. It can only claim to be a partial study of language, and furthermore, its implication that a theory of language is a part of the theory of action simply because speaking is a rule-governed behavior turns out to be oversimplified. An adequate theory of language, as i t appears from the above analysis, must take into account thinking along with saying, and must analyze mental states and 129 attitudes which also manifest a use of language just as much as the illocutionary acts do. Vendler holds that the common link between these two domains is the 'objects' of the propositional attitudes and of the illocu-tionary acts. These are, he believes, propositions. According to Searle, "when a proposition is expressed i t is always expressed in the performance of an illocutionary act," (1970; 29) As i t turns out, from Vendler's analysis, this need not be so. If we charac-terize an illocutionary act as performing a 'conventional' speech act at a given moment and carefully sort out the illocutionary act-indicative verbs which are all performatives from the other decepti-vely similar verbs, namely, the propositional state verbs and the pro-positional achievement verbs, i t becomes clear that expressing a pro-position (even in a speech act,) need not necessarily involve an i l -locutionary act along with either locutionary (Austin's) or proposi-tional (Searle's) acts. As an alternative, Vendler uses the term 'mental act' and suggests that a proposition (whether expressed in an utterance or not) may simply be entertained in a 'mental act', say of 'realizing' or 'doubting', etc. This suggests a simple solution to the problem of the basic units of language; namely, saving the speech act analysis in its present form and supplementing i t with a parallel account of 'mental acts'. Together these might be hoped to supply us with a theory of language, covering both speech and thought. However, I believe that the 'act' terminology is more misleading than helpful. [My use of 130 'act' here is in conformity with Vendler's use of the term elsewhere in his paper, e.g., when he sorts out the illocutionary act verbs from the propositional attitude verbs on the basis of their time schema , and notes that "the illocutionary act occurs at a given moment, and, ideally, does not last through a period of time."] The 'act' termino-logy is insufficient because in both speech and thought we use language in more ways than just 'performing' certain 'acts'. We have proposi-tional attitudes, we experience propositional states, and reach propo-sitional achievements. These differ from one another not only in their grammar but also in their speech situation, the communicative or simply expressive intention that accompanies them, etc. Since one can (but need not) use speech (the verbs of thinking) to express a propositional attitude, the line between 'saying' and 'thinking' is not precise. This is also a reason why speech acts cannot be regarded as the basic units of linguistic communication, in the hope that they would provide us with a study of langue; for both the 'speech' and 'act' aspects of speech acts give way to wider concepts of speech plus thought and acts  plus attitudes. We will have to revise Searle's claims about 'speaking a language' in the light of the above discussion. Once we see that speaking a language is not just 'performing' illocutionary acts, that i t is also holding beliefs, knowing a fact, having a doubt, dreaming a dream and so on, we can realize that 'speech act', which covers only a part of the communicative aspect of language, and is the pro-duction of issuance of a sentence at time _t, has l i t t l e hope of claiming the title of 'the basic unit of linguistic communication' and even less so of providing us with an adequate study of language. 131 CHAPTER 3  SENTENCES The Sentence-primacy Thesis: Quine's Version Philosophers of different times and opinions such as Bentham, Erege, Russell have treated a sentence as the primary vehicle of meaning. I choose Quine's version of the sentence-primacy thesis for discussion mainly because his treatment of the thesis is more extensive, more consistent, and with widespread implications in several related topics. As regards consistency: earlier we saw that Russell, in effect, treats sentences as epistemologically basic units of a person's lan-guage. But we also witnessed some inconsistencies in his overall theory regarding the basic units of language resulting from his failure to keep apart the first-learnt one-word sentences from the words of one's language having the same phonetic sequence as those sentences. In conclusion, we saw why his 'object-words' deserve to be treated both structurally and functionally rather as sentences and thus denied the supposed 'direct connection' with the 'objects' of experiences that they seemed to enjoy also as words. However, i f one decides to choose this horn of the dilemma, (as Quine does, in his Word and 19 Object,) one frees oneself of the inconsistencies resulting from the word/sentence confusion, only to confront once again the problem 19 W.V.O. Quine, Word and Object (Cambridge, the M.I.T. Press, 1960) Second paperback rept. 1965. 132 of explaining the relationship between the first-learnt units of language, and the words and sentences that are learnt later. Quine, following Russell, holds that the first-learnt units of language by a child are sentences as wholes and not single words. (1965, 13, 14) Like Russell, he also makes ample use of behavior-isti c notions such as 'stimulus-synonymy', 'stimulus-meaning', etc. We can hear distinct echoes of Russell's views in passages like the following: Words can be learned as parts of longer sentences and some words can be learned as one-word sentences  through direct ostention of their objects. In either event, words mean only as their use in sen-tences is conditioned to sensory stimuli, verbal or otherwise. (Quine, 1965, 17) However, Quine is much more consistent in his treatment of the basic units of language. For him, there is always an appreciable dif-ference between the phonemic sequences that figure as sentences in the first stages of our vocabulary and the same sequences as they appear later as words available for combination into new sentences. To sen-tences, he allows a monopoly of full meaning, in some sense, without denying that the meaning has to be worked out. To words, he allows meaning only in the sense in which i t can be abstracted from previously learned sentences. Thus he remarks, As the child progresses, he tends immediately to build his new sentences from parts; and thus i t is that one usually speaks of learning a new word rather than a new sentence. But even the sophisticated learning of a new word is com-monly a matter of learning i t in context — hence learning,~by example and analogy, the usage of sentences in which the word can occur. It therefore remained appropriate throughout § 3 and not just at the beginning of i t , to treat 133 sentences and not words as the wholes whose use is learned - though never denying that the learning of these wholes proceeds largely by abstracting and assembling of parts. (Quine, 1965, 13) Passages like this help to point out that Quine regards sen-tences as basic in more than one sense. In particular, he combines in his theory the insights and the approaches of Frege and Russell. Like Russell, he considers sentences as the first-learned (and, there-fore, basic) units of language of an individual. And like Frege, he considers the sentence as basic amongst the units of language (viewed synchronically) in that i t supplies foundations to the meanings of other smaller units such as words. Not that these two views are wholly unrelated either. For one can see how in expounding the Fregean out-look on what i t is for a word to have meaning only in the context of a sentence, Quine makes use of his sentence-oriented theory about the ways of learning words. Essentially, Quine's position is as follows: sentences are basic in that, (1) the first learned wholes of an individual's language are sentences (more specifically, one-word sentences,) directly conditioned to sensory stimuli, (2) and all words are learned strictly by depending on prior knowledge of sentences: either by segmenting previously learned sentences into conveniently small recurrent parts, or later, when the vocabulary is sufficiently enriched, by dictionary definitions, which are "mere clauses in a recursive definition of the meaning of sen-tences". In other words, knowing words is knowing how to work out the meanings of sentences containing them. 134 The merit of this position i s , that i t seems to avoid the ap-parent circularity which enters disputes over word-primacy versus sentence primacy. Roughly, the circularity threatens any uncritical account of what i t is to know the meaning of either a word or a sen-tence. Typical are remarks such as: Sentences mean what they mean because of what the words in them mean, and conversely, to know what a word means is to know what difference is made to the meaning of sentences in which i t occurs by the fact that that word is used and not others. (R.M. Hare, 1971, 95) Viewed uncritically, such remarks might give rise to the egg-or-hen type of primacy problem, since knowing word-meaning seems to be depen-dent on knowing sentence-meanings containing the word, and knowing sentence-meaning seems to result only after knowing the meanings of its constituent words. This seeming circularity or the flaw of mutual dependency is easily done away with in a theory that starts with re-garding sentences as the first-learned wholes of language for any indi-vidual. The network of language is then spread out smooth and wrinkle-free. We have the first layer consisting of sentences learned as wholes, the next layer of words gained by segmenting the first senten-ces, which layer almost merges into the layer where available words are combined to form an indefinitely large number of new sentences and new words are learned in the context of old sentences, Another merit of Quine's theory is that i t can answer satisfac-torily charges such as Michael Dummett's when he says, W.V. Quine says that Frege discovered that the unit of meaning is not the word but the sentence. Likewise grammarians debate whether the word 135 or the sentence is the primary element of meaning. This dispute seems to me empty and Frege's alleged discovery absurd. As Witt-genstein says in the Tractatus (4.032; cf. 4.026, 4.027, 4.03), the sentence is neces-sarily complex....Sometimes, too, i t is argued that the sentence is primary on the ground that we can learn the meaning of a word only by learning the meaning of the sentences in which i t occurs. But though i t is certainly true of some words that we can learn their sense only by learning the use of representative sentences containing them conversely there are some sen-tences e.g., "I expect Jones will resign within the next month." - whose sense we could not be taught directly, which we understand only by al-ready knowing the meaning of the constituent words. Any attempt to express the idea that the sentence is the unit of meaning, or even the idea that the meaning of sentences is primary, that of the words derivative, ends in implicitly denying the obvious fact - which is the essence of language - that we can understand new sentences which we have never heard before.20 As i t turns out, Quine's theory of the sentence-primacy does not deny the obvious fact that we can understand new sentences, nor does he have to deny that some (or even most,) sentences are learned by learning the meaning of their constituent words. He can accept both these views and yet maintain that a sentence is the primary vehicle of meaning, and this he does, as we have seen, by attributing epistemological primacy to sentences in his general account of an individual's language-learning. In other words, he has to contradict and correct only one statement from Dummett's charge, namely, that 'it is true of some words that we can learn their sense only by learning the use of representative sentences Michael Dummett, "Nominalism" Philosophical Review 1965, 491-505. 136 containing them'. Quine would want to say, I believe, that not just some, but all words are learned after learning sentences. At one place, Quine anticipates and answers yet another argument in favor of a certain obvious primacy of words as units of language. He says (1972, 293) that the recognition of the fact that the sentence and not the word is the unit of communication was long obscured by the undeniable primacy, in xme respect, of words. "Sentences being limit-less in number and words limited, we necessarily understand most sen-tences by constructing from antecedent familiar words." But he sees no conflict here. For he does not attempt to make al l the limitless sentences the first stepping stones of the structure of language. He only attributes sentence a full monopoly of meaning, and reserves the feature of sentence-dependent meaning for a l l words, without denying that the meaning of the sentences has to be worked out, largely by segmenting and combining the handful of first-learned wholes. Thus, sentence mainly has conceptual primacy over words, within the overall framework of language, and only a few first-learned sentences have functional primacy over words and other sentences in each individual's personal history of language learning. Indeterminacy and its Implications for Meaning At this point, we can consider a further detail in Quine's theory, namely, the issue of the indeterminacy of translation. The indeterminacy, though explained in terms of a program for a radical translation of a hitherto unknown language into English, is supposed to be also relevant to the question of the nature of language . 137 in general, especially to notions such as 'meanings', 'ideas' and 'propositions'. Let me sketch in brief some parallels between a child's task of language-learning and a linguist's program of a radical translation. Of course the linguist has advantages over the child in matters of age, experience and in already possessing at least one, i.e., his own native language. But in the program for a radical translation these advantages are virtually non-existent. The linguist, unaided by any interpreters, and relying only on the obser-vable stimuli and the data of the native utterances to form, test, and correct his hypotheses of translation from Jungle to English is almost in the same position as that of a child, unaided by any know-ledge of language, grasping his first utterances conditioned to some sensory stimuli, and learning his language with the help of his 'pre-linguistic quality space', i.e., his prior dispositions to treat one stimulation as more nearly similar to a second than to a third. Percei-ving the direct relationship of the utterances to the observable stimuli helps one —— the child or the linguist —— to collect a few 'occasion sen-tences' or whole utterances. And one may tentatively fix also some 'stimulus-meanings' for these first-learned sentence units. However, Quine claims that to go beyond this primitive stage and to get at the 'terms' of a language, these stimulus-meanings by themselves are not sufficient. Quine contrasts a 'term' with a 'sentence', (1965, 35) and in particular, with 'occasion sentences'. ["Gavagai!" and "Rabbit!"' are regarded as occasion sentences of the jungle language and of English; whereas 'gavagai' and 'rabbit' can be regarded as terms, although not 138 all xvords need be terms. Quine remarks (Ibid., 53) "Occasion sen-tences and stimulus meanings are general coin; terms and reference are local to our conceptual scheme." The observationality is what makes the former general coin, and the very lack of this quality makes the latter local to our conceptual scheme. It is thus that Quine maintains that the distinction between concrete and abstract objects, as well as between general and singular term, is independent of s t i -mulus-meaning. It is also at this stage that Quine scores a point over Russell, in that Quine observes what Russell failed to note: that the use of a word as an occasion sentence, however determinate, does not fix the extension of the word as a term. (1965, 53, fn. 2)] At this stage, 'analytical hypotheses' enter in a linguist's work, and with these does the germ of an indeterminacy. The thesis of indeterminacy is stated as follows: manuals for translating one language into another can be set up in divergent ways, al l compatible with one-another....The firmer the direct links of a sen-tence with non-verbal stimulation, of course, the less drastically its translations can diverge from one another from manual to manual. (Quine, 1965, 27) Quine's above thesis has been widely discussed and much attention has been given also to its translational aspect (see Words and Objections, 1969). However, the implications of this thesis for the general theory of language are of far more importance, and that is what I shall dis-cuss next. The role of analytical hypotheses looms large in Quine's indeter-minacy thesis. The analytical hypotheses about the approximate rende-rings of a foreign language to one's own are indispensable for a 139. translator. And the analytical hypotheses may differ from one linguist to another. "Rival systems of hypotheses can conform to all speech dis-positions within each of the languages concerned, and yet dictate, in countless cases, utterly disparate translations; not merely mutual paraphrases, but translations each of which would be excluded by the other system of translation." (1965, 73) The point of all this is that in lack of any correspondence of the total speech behavior with the rival theories, we cannot even talk about which one of the rival hypo-theses is the right one. There i s , Quine holds, simply no sense in trying to objectively compare the 'objective' references of terms in radically different languages. This is so because the terms and sen-tences of a theory or of a language are linked organically to each-other and there is no objective, interlinguistic, neutral 'meaning' which sen-tences of one theory bring out better than the sentences of the other. At this point Quine agrees with Wittgenstein in holding that with respect to the theoretical sentences, "Understanding a sentence means understanding a language." Here we see a parallel between a case of an individual's language-learning and the program of a radical translation. In language-learning, each individual's own forgotten past history may differ from another's; yet i f their net speech behavior is the same, then we do not see any sense in imagining any significant semantic dif-ferences between them. That they share the socially inherited system of'meanings' is borne out simply by their socially identical verbal behavior. Similarly, when two rival systems of analytical hypotheses offer apparently disparate results in some sentences of two radical 140 translations, there is no sense in weighing each against another in the hope that one rather than the other may have captured the 'meaning', 'the idea' or 'the proposition' which was there to be captured, free of any linguistic clothing. This denial of "some free-floating, l i n -guistically neutral meaning" is the main implication of Quine's thesis of indeterminacy. This position of his is also reflected elsewhere in his denials of any existence to 'propositions' in the sense of 'meanings of sentences'. This is a point where we can confront Quine's thesis of the primacy of the sentences as units of language against Vendler's suggestion that propositions might have a better chance of being regarded as the basic units of language, since they are the 'objects' of both speech and thought, i.e., of the totality of language. Sentences versus Propositions There is an old controversy over the question of the existence of 'propositions'. Propositions are posited by philosophers of quite varied opinions and for various purposes. Propositions and other sen-tence-meanings were wanted as transitional constants: as things shared by sentences and their translations. -Similarly they have seemed to be desirable as constants of philosophical analysis or paraphrases, as things shared by the analysandum and the analysans. Propositions are also posited as truth-vehicles and as objects of"the propositional at-titudes such as wishing, believing, etc. Quine argues that to posit propositions for either one of the above purposes is useless and exuberant. There are problems about the individuation of propositions, and propositions, when posited over and 141 above sentences, do not in any way significantly contribute to philosophical analysis. We have just seen above why Quine rejects the notion of propositions as translational constants. He also deems the notion of propositions useless to clarify the paraphrastic enter-prises of philosophical analysis. For he claims that in the para-phrastic enterprises of this nature, the line of analysis adopted will commonly depend on what is sought in the inquiry at hand; and since the purpose of inquiry will dictate divergent substitutions in one's paraphrases from time to time, there need be no question of the uniquely right analysis, nor of synonymy. As regards proposi-tions as truth vehicles, Quine proposes that we might stop one step earlier, i.e., with the eternal sentences of which propositions are said to be meanings. There seems to be no reason that compels one to posit propositions as truth vehicles in addition to the eternal sen-tences, because in order to individuate the *abstract' proposition as the meaning of this or that eternal sentence, we have to formulate the eternal sentence anyhow, so we might just use the sentences to measure the truth values rather than to burden our theory with 'abstract' propositions, themselves at the mercy of the sentences. As for the notion of propositions as the objects of the propositional attitudes, Quine suggests a pragmatic double standard. He elaborates on the problems of individuation of the so-called objects of proposi-tional attitudes, problems due to referential opacity and other such ailments, and then notes that so far as the theory in logic is con-cerned, we may simply dispense with the objects of propositional 142 attitudes, i.e., not view the 'that J J' of 'w believes that p_' as a singular term referring to objects. Of course, this does not suggest that the propositional attitudes are dispensable from our everyday discourse, but only that, in certain rigorous theories of a scentific nature, i t is better to avoid dealing with the obscure modes of ordinary language. Protests against even such a liberal and relativistic program are heard from the side of the propositionalists. Thus Vendler, for example, predicts that a l l attempts "to reduce the notion of a propo-sition, with respect to speech and thinking to the notion of an ut-terance, or an inscription, or a set of isomorphic utterances or inscriptions" are doomed to f a i l . He says, The object of speech and thought, the proposition, is not a thing, or a set of things, in the physical world; i t is an abstract entity. or elsewhere, What is said, the proposition, is not a sentence, or an utterance, nor any of these things as pro-duced in a given historical context; i t is the abstract unity of a paraphrastic set of imperfect nominals, which may be represented by any member of this set. Before we see what Vendler means by an 'abstract entity', let us see why he rejects the notion of a sentence or a set of sentences as the objects of both speech and thought. One reason is hinted at when he says that a proposition is not a thing or a set of things 'in the physical world'. It suggests a contrast between linguistic ele-ments as 'concrete things' in the physical world, and 'propositions' or sentence-meanings as 'abstract entities'. Now an utterance of a 14 3 sentence on a certain occasion by a human speaker may well be regarded as an event in the physical world, but it seems to me that sentences or sets of sentences are abstract enough to be regarded as the objects of speech and thought. Surely a set of sentences is not anything in 'the physical world', and yet i t raises fewer problems regarding 'identity' than the so-called 'abstract unity' of a paraphrastic set, for sets are identical i f and only i f their members are identical. If however, Vendler is contrasting simply a set of sentences against a set of imperfect nominals, and is calling the latter 'propositions', then .the quarrel seems to be less serious. For then one need not posit propositions in an obscure sense, for they are simply sets of linguis-tic expressions. Sentences versus imperfect nominals do not really pose a serious problem, for after a l l , the 'that'-clause of imperfect nominals need not be regarded as the mode of representing propositional attitudes. There are ways of analyzing the expressions of propositional attitudes in the ordinary language which result in an expression refer-ring to a speaker, the two place predicate such as 'said' or 'believes', and a demonstrative pronoun 'that' referring to an utterance. Davidson sketches one such way of analysis in Words and Objections. (1969, 170) Vendler argues further that as to the object of thought, there is not even a temptation to view it as a physical event or object simply because, unlike speech, thought need not be formulated at a l l . The ex-pression of thought, he admits, is usually done in words, but, he insists that words are mere carriers of what is said, of the thought now given voice. 144 Here again, I doubt whether anything useful is gained by positing propositions in place of sentences as objects of thought. For let us view critically the assumption: 'thoughts, unlike promises, need not be formulated at a l l . ' Does this suggest that we first have a thought, utterly unclothed in any linguistic element of however vague form, and then we express it_ at our will in words, of this or that lan-guage? I hope that Vendler is not counting on this sort of formulation, for i f he i s , then he seems to be wrong. Because on this formulation, thoughts- pure, naked, non-linguistic thoughts — would be forced back to the abysmal darkness of unidentifiable shadows, and we will have no hope of even catching a recognizable glimpse of these shadowy creatures, let alone catch them or choose this one as against the other as f i t for linguistic expression on a certain occasion. I think Vendler would not want to go to the extreme of stripping the thoughts altogether of any linguistic form, even i f they stay unuttered. I am supported in my belief here by his views in "Say What You Think." Consider, e.g., the following: Thought, in the strict sense of the word, is incon-ceivable without the idea of language, and the use of language, in the fu l l sense of the word, is essentially the expression of thought. This is so because the object of thought and the object of speech belong to the same category: what we say and what we think are things of the same kind. or again, ...speech is the primary phenomenon; thought is the derivative one. It is true that speech is the ex-pression of thought, but i t is also true, and I would like to say, more true, that thought is the suppres-sion of speech. It is like typing or just touching 145 the keys without pressing them. Nox«7, i f Vendler is right in the above remarks, and I believe that he i s , then he cannot count on the possibility of 'unformulated thoughts' as evidence for 'propositions'. If there are no thoughts without language, and i f thought is the suppression of speech, then we better not be misled by the worn out analogy of words as mere carriers of thoughts. Words and thoughts must be more intimately related than the relation that this or any such analogy allows for. Sentences thus are not merely necessary for the expression of thoughts, they are es-sential even for our having thoughts, i.e., for the very activity of thinking. We have no unformulated thoughts; we simply have more or less specific, more or less elaborate "expressions" — uttered, or unuttered. Viewing the matter thus would also resolve, I think, Vendler's search for 'the object' of speech and thought. The need for positing an object only arises from the impact of the uncritically applied ana-logiesiof the carrier and the carried, the clothes and the clothed. It falsely presents the activities of speech and thought as having an ob-ject apart from the activities themselves. Thus i t puts 'speaking' and 'thinking' on a par with 'throwing' or 'catching'. The latter pair of verbs denotes activities that have objects like a ball or a stone which have an existence quite apart from the throwing or catching operations that may or may not be performed on them. However, speaking or thinking are not activities like throwing or catching. They have no need for such external objects. We do not 'speak or think a proposition or a thought.' Saying, "I think a thought" is 146 redundant in a way in which saying, "I catch a stone" is not. It is enough to say that we speak or that we think. Period. In other words, 'propositions' have no objective existence apart from their linguistic 'expressions' or sentences. We can s t i l l talk of "JC'S belief" or "y_'s wish" and we can also make such claims in ordinary discourse as "John and Tom both believe the same thing." so long as we leave open the construal of the 'beliefs' and 'wishes' in terms of sentences, however general and unspecified they may be. Talk of beliefs and wishes need not demand that we attribute existence to propositions apart from the linguistic elements. We s t i l l talk of 'sunrises' and 'sunsets', even long after Copernicus, without feeling any need to attribute 'objective' reality to the sunrises and sunsets apart from the earth's revolutions around its e l f . 147 FINAL REMARKS In the preceding pages I have discussed the senses in which philosophers regard some linguistic units as primary or basic to language. It should be clear by now that there are some senses of primacy which are not relevant to the philosophical issues. For ex-ample, grammarians regard phonemes as the smallest significant units of language, but the size or shape of a linguistic unit is hardly an issue that matters in philosophical discussions. Nor is the criterion of primacy understood in terms of the non-analyzability inxother units, as it is for atoms in physics or chromosomes in biology. Words, sen-tences, speech acts — a l l these units are analyzable into further linguistic units for different purposes. These somewhat "structural" notions of primacy are not relevant in philosophical debates. The notion of primacy that seems to be relevant in such debates is a more or less "functional" (as opposed to "structural") notion. In other words, a linguistic unit that is claimed to be primary or basic is very often supposed to be essential in the analysis of the meaningful-ness of other linguistic units. Its function in explaining the notions of meaning, synonymy and the like is an important criterion of primacy so far as philosophical issues are in question. Thus Russell treats words as primary by stressing the fact that sentences are built up of words, and noticing that we can understand what a sentence expresses i f we know the meaning of its constituent words and the rules of syntax. Alston treats speech acts as primary to language because he believes that knowing what speech acts are performed in an utterance of a sentence will help one to decide what 148 meaning the sentence has, and also to decide questions of sentence synonymy and word synonymy. Similarly, Quine argues for the primacy of sentences because, (among other things,) sentences are supposed to be the starting points of an analysis of word-meanings and synonymy claims, i t is interesting that although these philosophers are con-cerned with the common task of intelligibly explaining notions such as meaning and synonymy, they come to regard different units as basic to language. This discrepancy is sometimes a result of overemphasizing only one aspect of the total issue, sometimes i t reflects more funda-mental differences in perspectives about the nature of language. For convenience of treatment I have divided philosophers' theories in word-primacy theories, sentence-primacy theories and speech-act primacy theories. However, there are noticeable differences even among the allies. Thus although Ziff and Russell both claim that words are primarily the linguistic units that have meaning, they do so on diffe-rent grounds. Ziff is mainly concerned with people's usage of the term 'meaningful' or 'having a meaning', and does not support his claim by any substantial reason other than dogmatically asserting 'that is how things are'. Russell is mainly concerned with epistemological priority of words, and tries to build a theory about an individual's language learning in which 'object words' occupy a preeminent position, in that they are the first learned words having direct connection with sensible objects and therefore are at the basis of meaning of other words and sentences. Similarly, the two speech act theorists — Alston and Searle — differ considerably, as we have seen, in their total accounts. 149 Although both claim that speech acts are basic to linguistic com-munication, Alston claims that the meaning of a sentence is a function of its illocutionary act potential which Searle claims that the illocutionary act is a function of the meaning of a sentence. The differences in Alston's and Searle's claims about the relationship between the sentence meaning and the illocutionary force reflect the basic differences in their perspectives about the nature of language itself. Alston's claims about the primacy of speech acts are founded on his belief that they explain the meaningfulness and synonymy of words and sentences, and not on the belief that language itself con-sists of speech acts. While regarding the speech act as the most ap-propriate unit of linguistic behavior, he tries also to maintain the distinction between speech and language, and abstains from claiming that the speech acts are basic to the language its e l f . Speech comprises the totality of verbal behavior that goes on in a community; whereas language is . the abstract system of identifiable elements and the rules of their combinations, which is exemplified in this behavior and which is discovered by an analy-sis of the behavior....It is s t i l l more impossible to identify a language with a series of events or aggre-gate of verbal behaviors. (Alston, 1964, 61) Thus Alston's claim is comparatively modest. It is about the primacy of speech acts as the units of verbal behavior insofar as they function in an analysis of word-meanings and sentence-meanings. However, Searle holds a contrary belief as regards the relationship of sentence-meanings to speech acts. According to him the latter is the function of the former, and not vice versa. His primacy claim for speech acts is based 150 on his conception of language. He blurs the distinction between speech arid language, and claims that the study of speech acts is (not just a study of parole or speech, but rather,) a study of langue or language simply because "speaking a language is performing speech acts", and "a theory of language is a part of the theory of action." There is yet another angle from which differences between the primacy claims can be brought out. Sometimes philosophers view lan-guage as i t is manifested in the communication patterns of adult native speakers. At other times, they view language as a facility developed by an individual from his childhood. Not that these two viewpoints necessarily conflict with each other, but they are not treated as essentially related to each other by many. Philosophers such as Austin and Searle do not find i t necessary to relate their observations about adult linguistic behavior to the process of a child's language learning. Russell and Quine, on the other hand, try to synthe-size their observations regarding these two aspects of language. By using a bit of jargon, one can say that language can be viewed learning-theoretically or meaning-theoretically; and accordingly the primacy of linguistic units may be epistemological or logical. I think that these two viewpoints are not unrelated but are rather complementary to each other, and that although one can study the phenomenon of language from either perspective, a synthesis of the two would give one a better understanding of language. In this dissertation I have tried to show negatively, why neither 151 words, nor speech acts nor propositions can be regarded as the basic units of language, and positively, why sentences can be so regarded. The units other than sentences are found to be dependent on sentences for their own individuation or meaningfulness. In connection with Russell's version of the word-primacy theory I tried to show that the so-called 'object-words' or the units of an individual's first language are actually sentences. They are sentences both on structural and functional counts. They are one-v/ord sentences used in an exclamatory manner, and have a 'propositional' meaning in the sense that they either express an affirmative, an imperative, or an interrogative proposition. The words that occur in these first learned one-word sentences are the words that have observable denotations and that is probably why they figure so early in an individual's language. The meaning of even these words, I submit, have to be grasped after coming across their different occurrences in sentential contexts. I conclude that, (1) the first learned units of an individual's language are one-word sentences; (2) all words have meaning only in sentential contexts; (3) the meaning of the 'object-words' insofar as they are sentences, is not just 'the ob-jects', but 'the objects' can be the denotations of the words that occur in the primary sentences. With respect to Ziff's version of the word-meaning primacy theory, I must confess that I did not find that he had given substantial ;arguments to support his claim. I point out a number of cases where people do not restrict the word 'meaning' or the phrases 'have a meaning' and the like to the talk about words, but apply them to sentences and other linguistic 152 units as well. So i f his claim is about people's usage, then i t meets with many counterexamples; and i f he decides to brush aside the counter-examples merely as reflecting philosophers' strained, non-standard usage, then he has yet to substantiate his claim regarding the common man's standard usage. The only argument for Ziff's choice of restricting 'have a meaning' to words as opposed to sentences seems to be based on his obser-vation regarding the limitation of human memory. This again is very puzzling. If he is arguing (as he seems to me to be) that because of our natural memory limitations we, as human beings, tend to focus only on ut-terances which are found to be members of the distributive sets of words, then the fact that other philosophers have actually focused on units other than utterances, (namely, speech acts,) and have come up with dif-ferent primarily meaningful units (namely, sentences,) ought to cast a doubt on Ziff's observation about our natural memory limitations. If however, Ziff does not put much weight on memory limitations, and is willing to accept that theoretically i t is possible to regard other l i n -guistic units as having a meaning, only that we do not call them meaning-ful, then he would be making a milder claim and a more vulnerable one too. Because then the question i s , i f a word is to be regarded as having a meaning simply on the grounds of its contribution to the sig-nificance of a sentence or an utterance, why cannot the same be said of a phoneme? And secondly, i f , as Ziff himself says, words, sentences, phonemes and the like are all meaningful units yielded on different levels of linguistic analysis, then why should the fact that one does not 153 explain sentence-synonymy in a manner in which one explains word-synonymy induce one to apply the phrase 'having a meaning' exclusively to>words rather than to sentences? My own impression is that Ziff's own procedure of explaining word-meaning by going through several sen-tential contexts suggests something important about the ultimate de-pendence of word-meaning on sentences which unfortunately he does not take into account. < The dependence of speech acts on sentences is of a different kind. Words and sentences both are, as i t were, on one level, since they can be regarded as the units of this or that language, as pieces of its vocabulary. But speech acts are the units of our verbal behavior. It should be noted that by definition speech acts stress the speech aspect and the act aspect of language. I find that the notion of speech acts as explained in the works of Austin, Searle and the like leaves much to be explained. The notions of intentions and securing 0 uptake pose extremely vexing questions regarding the identification, let alone the felicity of speech acts. The distinction between the locutionary and the illocutionary aspects of an utterance needs to be sorted out in terms of acts and in terms of the meaning/force distinc-tion. As regards the acts, there is an important asymmetry between the concepts of the locutionary and illocutionary acts, namely, the fulfillment of felicity conditions for the .latter, but not for the former. This requirement marks the difference between the two con-cepts as well as the two classes of linguistic acts, although there may be some intermediate cases. This applies both to Austin's original 154 distinction between locutionary and illocutionary acts and to Searle's improved version of the distinction in terms of propositional versus illocutionary acts. The distinction in terms of (locutionary) meaning and (illocutionary) force is at best shaky, for sometimes the illocu-tionary force indicators such as stress, intonation, word order, etc., mark the meaning or the propositional content as well. It turns out that for the utterances containing explicit performatives, their force is uniquely determined by the meaning of the sentence, whereas for the others, i t is a function of the sentence-meaning. This is an important consideration for evaluating Alston's and Searle's respective primacy claims. Among the two prominent speech act theorists, Alston's claims are, as I said before, based on his belief that speech acts function essen-tially in explaining the notions of sentence-meaning and sentence-syno-nymy. I have shown why this endeavor is inherently circular. I have shown that since in individuating a speech act to the ultimate level where i t is supposed to be helpful for deciding whether two sentences are synonymous, one has necessarily to specify the meaning of the sen-tence as an essential ingredient of the speech act i t s e l f , and therefore cannot expect the notion of a speech act to be useful in explaining the notions of sentence-meaning and sentence-synonymy. This leads me to conclude that speech acts are functions of the meaning of a sentence rather than the case being otherwise. Therefore speech acts are not basic to language on |he functional count, i.e., not essentially helpful in explaining the cluster of notions around the 155 nucleus of meaning, either for sentences or for words. There is yet another interpretation of the primacy of speech acts and that is represented by Searle's account. It is to regard speech acts as basic to language because speaking a language is supposed to be simply performing speech acts. I have tried to show that Searle's argument rests on an oversight. It neglects an inherent ambiguity in the phrase 'speaking a language'. Speaking a language has both an activity aspect and a facility aspect. The notion of a speech act necessarily involves the notion of a timeable act of 'uttering' a meaningful sequence of syllables, which is not what we do everytime we use language. We also use language when we think, realize, wonder, remember, entertain or examine beliefs, grudges or doubts, and so on. The difference between speaking or performing speech acts and engaging in these other activities or having these propositional attitudes is not just the silent/overt distinction, but rather the difference between their logical grammar, their time schemata, the communicative versus non-communicative situation and the like, as is suggested by Vendler's researches. Speech acts cannot be basic to language because neither speech nor acts is a l l there is to language. An adequate theory of language should take into account both the speech and the thought as-pects and also both the activity and the facility aspects of language. The fact that pieces of non-communicative utterances can be mistaken for communicative utterances, that soliloquies can be overheard, understood and answered back suggests that there is something common to both the speaking and the thinking aspects of language. The common element 156 employed in any linguistic act or attitude is the meaningful sentence, the knowledge of which helps a listener to understand what is being said, although not necessarily said to him. Coming to sentences, I conclude that sentences can be regarded basic both epistemologically and logically; or both learning-theoreti-cally and meaning-theoretically. Following Quine, I suggest that the meaning-theoretical primacy of sentences becomes clearer when we take into account their learning-theoretical primacy. Thus sentences are basic in that, (1) the first learned whole units of an individual's lan-guage are sentences; and (2) meaning of all words is grasped in senten-tial contexts: either by abstracting and associating severally re-curring parts of previously learned sentences, or later, when the vocabu-lary is sufficiently enriched, by dictionary definitions. This is not to deny the a l l too obvious fact that new sentences are built up of words and understood by projections from previously known words. In order to get out of the egg-or-hen type of problem about the mutual dependency of word-meanings and sentence-significance, we have only to remind our-selves of the (1) above, which claims epistemological primacy only for a few first learned sentences, and not to each and every sentence of any language. However, (2) claims the meaning-theoretical primacy for the concept of sentence over the concepts of other linguistic units. The study of sentence-meanings, as we have seen before, supplies foun-dations for the study of speech acts, as speech acts are the functions of sentence-meanings. I submit that the study of sentences involves something more than that. Looking into the other functions of sentence-157 meanings and investigating their logical grammar would also shed light on the uses of the verbs of propositional attitudes, i.e., especially on the thinking aspect of language. As to the question of whether propositions rather than sentences should be more reasonably regarded as basic to language, since they can be supposed to be the 'object's of both illocutionary acts and the propositional attitudes, I reply in negative. This is because just as the illocutionary acts that we perform involve utterances, the propo-sitional attitudes that we entertain are also formulated in sentences, rather than extra-linguistic propositions. Propositions themselves depend on sentences for their own individuation, and no talk of propo-sitions makes sense unless they are identified by means of this or that sentence. Therefore positing propositions over and above senten-ces is useless and unnecessary. I point out that the need for postula-ting the 'object' of speech and thought arises from the uncritically applied analogies of the cloth and the clothed, or the carried and the carried. The activities of speaking and thinking, unlike throwing or catching, do not depend on external objects which have existence quite apart from the activities themselves. One can catch a ball or a stone or a frisbeek0f what have you. But speaking or thinking cannot be car-ried on without meaningful sentences. In other-words, propositions have no objective existence apart from their linguistic 'expressions', or sentences. We have no unformulated thoughts; we simply have more or less well formulated linguistic expressions — uttered or unuttered. 15 8 BIBLIOGRAPHY The works cited in the dissertation are marked with an asterisk. Aldrich, Virgil. "Review: Ziff's Semantic Analysis." Philosophical  Review, 1962, pp. 243-248. . "Illocutionary Space." Philosophy and Phenomenological Research, Vol. XXXII,. No. 1. 1971, pp. 15-28. *Alston, William P. "Ziff's Semantic Analysis." The Journal of Philoso- phy, Vol. LIX No. 1. 1962, pp. 5-20. * . "Meaning and Use." Philosophical Quarterly, Vol. 13, 1963, pp. 107-124; reprinted with revisions and additions in G.H.R. Parkinson, Ed. The Theory of Meaning, London, Oxford University Press, 1968. . "Linguistic Acts." American Philosophical Quarterly. Vol. 1. No. 2. April 1964, pp. 138-146. * . Philosophy of Language. Foundations of Philosophy Series, Englewood C l i f f s , N.J.: Prentice-Hall, 1964. *Anscombe, G.E.M. "Critical notice: Ziff's Semantic Analysis." Mind, Vol. LXXII, 1963, pp. 288-293. . Intention. Oxford: Basil-Blackwell, 1963. Armstrong, D.M. "Meaning and Communication." Philosophical Review, Oct. 1971, pp. 427-447. *Austin, J.L. How to Do Things with Words. Cambridge, Mass.: Harvard University Press, 1962; reprinted as a Galaxy Book, New York: Oxford University Press, 1967. . "Performative-Constative." in Philosophy and Ordinary Lan- guage , Ed. Charles E. Caton, Urbana: University of Illinois Press, 1963. . Philosophical Papers. Ed. J.O. Urmson and G.J. Warnock, Oxford: Clarendon Press, 1961, Second edition, Oxford: Oxford University Press, 1970. *Black, Max. The Labyrinth of Language. New York: The New American Library, 1969. • "Some Troubles with Whorfianism." in Language and Philosophy: A Symposium,;Ed. Sidney Hook,New York: New York University Press, 1969. 159 . Margins of Precision; Essays in Logic and Language. Ithaca: Cornell University Press, 1970. *Bloomfield, Leonard. Language, New York: Holt, 1933.; rept. 1964. Brown, D.G. "Knowing How and Knowing That-What." in Ryle: A Col- lection of Critical Essays, Ed. Oscar P. Wood and George Pitcher, Garden City,. New York: Anchor, 1970. Brown, Roger. Words and Things. New York: The Free Press, 1958, First Free Press Paperback ed., 1968. Broyles, James E. "Language and Common Sense." American Philoso- phical Quarterly, Vol. 6. 1969, pp. 233-239. *Carnap, Rudolf. "Testability and Meaning." Philosophy of Science, Vol. 3, 1936, pp. 419-471; Vol. 4, 1937, pp. 1-40. Caroll, John B. Language and Thought. Foundations of Modern Psy-chology Series, Englewood Cl i f f s , N.J.: Prentice-Hall, 1964; second print, 1965. Caton, Charles E. Ed. Philosophy and Ordinary Language. Urbana: University of Illinois Press, 1963. "Review of Vendler's Linguistics in Philosophy." Philoso- phical Review, Vol. 78, 1969,:pp. 518-521. Cerf. Walter. "Critical notice: How to Do Things with Words by Austin." Mind, Vol. 75, 1966, pp. 262-285. Chomsky, Noam. Syntactic Structures^ The Hague: Mouton, 1957; Fifth printing, 1965. . "Degree of Grammaticalness,." in Fodor and Katz, 1964 . , . "On the Notion "Rule of Grammar;'.1 in Fodor and Katz, 1964, . Aspects of the Theory of Syntax. Cambridge, Mass.: M.I.T. Press, 1965. . Topics in the Theory of Generative Grammar. The Hague: Mouton, 1966. . "Recent Contributions to the Theory of Innate Ideas." Synthese, Vol. 17, .1967, pp. 2-11. . Language and Mind. New York: Harcourt, 1968. • "Linguistics and Philosophy" in Language and Philosophy (Ed.) Hook, 1969. 1.60 Cohen, L.J. "Are Philosophical Theses Relative to Language?" Analysis, March, 1949. * . "Do Illocutionary Forces Exist?" Philosophical Quarterly, 14, 1964, pp. 118-137. . "Searle's Theory of Speech Acts." Philosophical Review, Vol. 79, 1970, pp. 545-557. . "Review: Zeno Vendler's Linguistics in Philosophy." Foundations of Language, Vol. 7, No. 1. 1971, pp. 125-130. Copi, Irving M. and James A. Gould (Eds.) Contemporary Readings in Logical Theory. New York: Macmillan, 1967; second print, 1968. Cousin, D.R. "How Not to Talk: Pedantic Objections to Professor Austin." Analysis 15:4, March 1955, pp. 73-81. *Davidson, Donald, and Hintikka J. (Eds.) Words and Objections: Essays on the Work of W.V. Quine. Dordrecht, The Netherlands: D. Reidel Publishing Company, 1969-. "Dummett, Michael. "Nominalism." Philosophical Review, 1956, pp. 491-505. Fingarette, Herbert. "Performatives." American Philosophical  Quarterly, Vol. 4, 1967, pp. 39-Fodor, J.A. and J.J. Katz (Eds.) The Structure of Language: Readings  in the Philosophy of Language. Englewood Cl i f f s , N.J.: Pren-tice-Hall, 1964. •-'•Frege, Gottlob. Grundlagen der Arithmatik. Translated by J.L. Austin as The Foundations of Arithmatic, Oxford: Basil-Blackwell, 1950. . "On Sense and Reference." Translations from the Philosophical Writings of Gottlob Frege. (Eds.) Peter Geach and Max Black. Oxford: Oxford University Press, 1952. . "On Sense and Nomination." Readings in Philosophical Analysis by Herbert Feigel and Wilfrid Sellars. Appleton-Century-Crofts, 1949; reprinted in Copi and Gould. Fremlin, Celia. "Must We Always Think in Propositions?" Analysis, Vol. 5, No. 2, Jan. 1938, pp. 17-27. Gale, Richard M. "Propositions, Judgments, Sentences and Statements." The Encyclopedia of Philosophy. Ed. Paul Edwards, Vol. 6. New York: Macmillan. 16* . "The Fictive Use of Language." Philosophy: The Journal of the Royal Institute of Philosophy, Vol. XLVI, No. 178, Oct. 1971, pp. 324-339. Garner, Richard T. "Some Doubts About Illocutionary Negation." Analysis, Vol. 31, No. 3. Jan. 1971, pp. 106-112. Garvin, Lucius. "Do We Ever Think in Propositions?" Analysis, Nov. 1938, pp. 25-32. Geach, Peter. Mental Acts: Their Content and Their Objects. London: Routledge and Kegan-Paul, 1957; third print, 1964. Ginsberg, Michell. "How to Say It and Mean It." Philosophical Studies, Vol. 22, April 1971, pp. 43-48. Green, O.H. "Intention and Speech-acts." Analysis, Vol. 29, No. 3, Jan. 1969, pp. 109-112. *Grice, H.P. "Meaning." Philosophical Review, Vol. 66, 1957, pp. 377-388. . "Utterer's Meaning, Sentence Meaning, And Word: Meaning." Foundations of Language, Vol. 4, 1968. . "Utterer's Meaning and Intentions" The Philosophical Review, April 1969, pp. 147-177. %are, R.M. Practical Inferences. London: Macmillan, 1971. Henle, Paul. (Ed.) Language^ Thought and Culture. Ann Arbor Paper-backs: University of Michigan, 1965; Second edition, 1966. Henson, Richard G. "What We Say." American Philosophical Quarterly, Vol. 2, 1965, pp. 52-62. Holdcraft, David. "Meaning and Illocutionary Act." in The Theory of Meaning, Ed. G.H.R. Parkinson, Oxford: Oxford University Press, 1968. Honore", A.M. "Reference to the Non-existent." Philosophy, Vol. XLVI, No. 178, Oct. 1971. Hook, Sidney. Ed. Language and Philosophy: A Symposium. New York: New York University Press, 1969; Jacobson, Roman and Morris Halle. Fundamentals of Language. The Hague: Mouton, 1956. Katz, J.J. and Edwin Martin Jr. "The Synonymy of Actives and Passives." Philosophical Review. Vol. 76, 1967, pp. 476-491. 162 Katz, J.J. and P. Postal. An Integrated Theory of Linguistic Des- criptions. Cambridge, Mass.: M.I.T. Press, 1964. Khatchadourian, Haig. "Vagueness, Meaning and Absurdity." Amer- ican Philosophical Quarterly, Vol. 2, 1965, pp. 119-129. Kiefer, Howard, and Milton Munitz. Language, Belief and Metaphysics. In the four Vol. publication of Contemporary Philosophic  Thought at The International Philosophy Year Conferences (1967-68) at Brockport. Albany: State University of New York Press, 1970. Kitley, Murray. "Of What We Think." American Philosophical Quarterly, Vol. 5, No. 1, Jan. 1968, pp. 31-42. Lehrer, Adrienne and Lehrer, Keith. (Eds.) Theory of Meaning. En-glewood Cl i f f s , N.'J.: Prentice-Hall, 1970. Linsky, Leonard. Semantics and the Philosophy of Language. Urbana: University of Illinois, 1952; reprint, 1964. Locke, John. An Essay Concerning Human Understanding. Collated and annotated by Alexander C. Fraser, Vol. II, New York: Dover, 1959. Lyons, John. Ed. New Horizons in Linguistics. Harmondsworth, Middle-sex, England: Penguin, 1970. Nyiri, J.C. "No Place For Semantics." Foundations of Language, Vol. 7, No. 1. Feb. 1971; pp. 56-69. Ogden, CK. and I.A. Richards. The Meaning of Meaning: A Study of  the Influence of Language upon Thought and of the Science of Symbolism. New York:Harcourt, Brace and World, 1923; Eighth ed. 1946. *Parkinson, G.H.R. Ed. The Theory of Meaning. Oxford: Oxford Uni-versity Press, 1968. Patton, T. and Paul Ziff. "On Vendler's Grammar of "Good"." Philosophical Review, Vol. 73, 1964, pp. 528-537. Patton, Thomas E. "Syntactic Deviance." Foundations of Language, Vol. 4, 1968, pp. 138-153. Pears, David. "Wittgenstein and Austin." British Analytical Phi 1 osophy -.Internat.ional Library o£ Philosophy and Scientific Method. New York: Humanities Press, pp. 17-39. Peetz, Vera. "Propositional Content." Analysis, Vol. 32, No. 6, June 1972. 16 3 Quine, W.V.O. "The Problem of Interpreting Modal Logic." Journal of Symbolic Logic, Vol. 12, 1947, pp. 43-48. Reprinted in Copi and Gould. ' . From A Logical Point of View. Cambridge, Mass.: Harvard University Press, 1953; Second revised edition in 1961; Harper Torchbook edition; New York: Harper, 1963. ' "Meaning and Translation." in Fodor and Katz. * Word and Object. Cambridge, Mass.: M.I.T. Press, 1960; Second Paperback printing, 1965. . The Ways of Paradox. New York: Random House, 1966. * . "Russell's Ontological Development." The Journal of Philo- sophy, Vol. LXIII, No. 21, 1966; Reprinted in Bertrand Russell: A Collection of Critical Essays. Ed. D.E. Pears, 1972. . Philosophy of Logic. Foundations of Philosophy Series. Englewood C l i f f s , N.J.: Prentice-Hall, 1970. . and J.S. Ullian. The Web of Belief. New York: Random House, 1970. Rankin, K.W. "Wittgenstein on Meaning, Understanding and Intending." American Philosophical Quarterly, Vol. 3, 1966, pp. 1-13. Reach, K. "What is a Language" Analysis, June 1939, pp. 49-62. Rorty, Richard. Ed. The Linguistic Turn: Recent Essays in Philoso- phical Method. Chicago: The University of Chicago Press, 1967; First Phoenix edition, 1970. *Russell, Bertrand. An Inquiry into Meaning and Truth. New York: W.W. Norton, 1940; Reprinted Baltimore, Maryland: Penguin, 1969. * • Logic and Knowledge: Essays 1901-1950. Ed. Robert C. Marsh, 1956, Capricorn Books edition, 1971. . "Mr. Strawson on Referring." in My Philosophical Develop- ment. London: Allen and Unwin, 1959, pp. 238-245; Reprinted In* Copi -and" "Could. ': ' : Ryle, Gilbert. The Concept of Mind. London: Hutchinson's University Library, 1949. . "Knowing How and Knowing That." Proceedings of the Aristote-lian Society, Vol. XLVI, 1945-46, pp. I-I64, Reprinted in Ryle, Collected Papers, Vol. II, 1971. 16 4 . "Ludwig Wittgenstein." Analysis, Vol. 12, No. 1, Oct. 1951, pp. 1-9. . "Ordinary Language." Philosophical Review, April 1953, pp. 167-186; Reprinted in Caton, 1963. . "The Theory of Meaning." in Philosophy arid Ordinary Lan- guage. Ed. Caton, 1963. . "Use, Usage and Meaning." Aristotelian Society Supplemen- tary, Vol. XXXV, 1961; Reprinted, Oxford: Parkinson, 1968. . Collected Papers: Vol. I, Critical Essays, Vol. II, Col-lected essays 1929-1968. London: Hutchinson, 1971. Sapir, Edward. Language. New York: Harcourt, Brace and Court, 1921; Harvest edition, 1963. Saporta, Sol. (Ed.) Psycholinguistics: A Book of Readings. New York: Holt, Rinehart and Winston, 1961; Reprint 1966. Searle, John R. "Meaning and Speech Acts." Philosophical Review, Vol. 71, 1962, p. 423. * . "What Is a Speech Act." in Max Black (Ed.) Philosophy in America. Ithaca: N.Y.: 1965. * . "Austin on Locutionary and Illocutionary Acts." Philoso- phical Review, Vol. 77, 1968, pp. 405-424. * . Speech Acts: An Essay in the Philosophy of Language. Cam-bridge: Cambridge University Press, 1969; Reprint 1970. . "Theories of Meaning and Uses of Language." in Philosophical Issues (Eds.) James Rachels, Frank Tillman. New York: Harper & Row, 1972. Seasonske, A. "Performatives." The Journal of Philosophy, Vol. 62, 1965, pp. 459-468. *Siegler, Fredrick A. "Lying.5" American Philosophical Quarterly, Vol. 3, No. 2, 1966, pp. 128-136. Skinner, Quentin. "On Performing and Explaining Linguistic Actions." Philosophical Quarterly, Vol. 21, No. 82, Jan. 1971, pp. 1-21. Sloman, Aaron. "Transformations of Illocutionary Acts." Analysis, Vol. 30, No. 2, Dec. 1969, pp. 56-69. 165 Steinberg, David D. and Leon A. Jakobovitz. Eds. Semantics: An Interdisciplinary Reader in Philosophy, Linguistics, and Psy- chology. Cambridge: Cambridge University Press, 1971. Stevenson, Charles L. Ethics and Language. New Haven: Yale Univer-sity Press, 1944. . "Some Relations Between Philosophy and the Study of Language." Analysis, Oct. 1947. . Facts and Values: Studies in Ethical Analysis. New Haven: Yale University Press, 1963, Reprint 1967. *Strawson, P.F. "On Referring." Mind. Vol. 59, 1950, pp. 320-344. Reprinted in Copi and Gould. * . "Intention and Convention in Speech Acts." Philosophical Review, Vol. 73, 1967, pp. 439-460. Tarski, Alfred. "The Semantic Conception of Truth." Philosophy and Phenomenological Research, Vol. 4, 1944, pp. 341-375. Reprinted in Linsky. Thau, Stuart. "Illocutionary Breakdowns." Mind. April 1971, pp. 270-275. . "The Distinction Between Rhetic and Illocutionary Acts." Analysis, Vol. 32, No. 6, June 1972. pp. 177-186. *Ullmanri, Stephan. Semantics: An Introduction to the Science of Meaning. Blackwell, 1962; Reprint. Barnes and Noble, 1964. Vendler, Zeno. Linguistics in Philosophy. Ithaca, N.Y.: Cornell University Press, 1967. • "Review of J.L. Austin's Sense and Sensibilia and How to Do Things with Words." Foundations of Language, Vol. 3, 1967. . "Semantics and Aletheia: A Comment." The Philosophical Forum, Vol. 4, 1969. * • "Say What You Think." in Studies in Thought and Language., Ed. J.L. Cowan, Tucson, Arizona: University of Arizona Press, 1970. Walker, Jeremy D. B. "Statements and Performatives." American Philo- sophical Quarterly, Vol. 6, No. 3, 1969, pp. 217-225. Warnock, G.J. "Words and Sentences." in Ryle: A Collection of  Critical Essays, (Eds.) Wood and Pitcher, 1970. 166 Welker, David D. "Linguistic Nominalism." Mind, Vol. LXXIX, 1970, pp. 569-580. White, Alan R. "The Use of Sentences." Analysis, Vol. 17, No. 1, Oct. 1956, pp. 1-4. Wittgenstein, Ludwig. Notebooks 1914-16. Ed. G.H. Von Wright and G.E.M. Anscombe, with an English translation by G.E.M. Anscombe, Oxford: Blackwell, 1961; First Harper ed. New York: Harper, 1969. . Tractatus Logicd-Philosophicus. (finished in 1918) with an introduction by Bertrand Russell. The first English translation by C.K. Ogden, London: Routledge and Kegan Paul, 1922, A New English Translation by D.F. Pears and B.F. McGuinness. London: Ibid., 1961. . The Blue and Brown Books. (1933-35) Edited with a preface by R. Rhees. Oxford: Blackwell, 1958; First Harper Torchbook edition, New York: Harper & Row, 1965. Philosophical Investigations. Ed. by G.E.M. Anscombe and R. Rhees with an English translation by G.E.M. Anscombe. Oxford: Blackwell, 1953; Second ed. New York: Macmillan, 1958. Whorf, Benjamin Lee. Collected Papers on Metalinguistics. Washing-ton, D.C.: Department of State, Foreign Service Institute, 1952. . Language, Thought and Reality: Selected Writings of Benjamin Lee Whorf. Edited with an introduction by John B. Caroll, Cam-bridge, Mass.: M.I.T. Press, 1956; Second paperback printing, 1966. Ziff, Paul. "On H.P. Grice's Account of Meaning." Analysis, Vol. 28 No. 1, Oct. 1967, pp. 1-8. ( . Semantic Analysis. Ithaca, N.Y.: Cornell University Press, 1960; Reprinted 1967. . "A Response to "Stimulus Meaning." Philosophical Review, Vol. 79, pp. 63-74, Jan. 1970. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0100888/manifest

Comment

Related Items