Ariskant and Plathegel

Before Plato Oral tradition about matters of language is almost solely restricted to myths about the origins of language; for example, the mythical figure Cadmus, son of the king of Phoenicia, supposedly brought the alphabet to Greece (see Chapter 1). It is hard to imagine a body of linguistic work being produced before the creation of writing systems; and with the advent of writing, we are in historical times. Certainly, the Western Classical Tradition is a matter of history, not prehistory; so too are the Indian, Chinese and other traditions (see Lepschy (ed.) 1994, Vol. I; Koerner and Asher (eds) 1995; Auroux et al. (eds) 2000-6, Vol.1). In the ancient world all texts were handwritten in scripta continua (script without spacing between words, also called scriptio continua), sometimes without punctuation as in the Greek inscription on the British Museum’s Rosetta Stone (Figure 2.1).1 In early Egyptian papyri, and later in demotic and Greek texts, the first line of a stanza was sometimes indicated by the use of red instead of black ink; and line ends were marked by red dots. In Figure 2.22 the red parts of the original are in outline grey. 1. Based on a facsimile by Stephen Quirke, frontispiece to Parkinson 1999. 2. Based on Parkinson 1999, Plate 8. Figure 2.1. The Greek inscription in scripta continua on the Rosetta Stone, with a superimposed enlarged fragment from the top left corner. Plato on language 23 Here are examples of scripta continua in English; in the first, line ends are marked by grey dots. ALLTHEWORLDSASTAGE?ANDALLTHEMENANDWOMENMERELYPLAYERS?TH EYHAVETHEIREXITSANDTHEIRENTRANCES?ANDONEMANINHISTIMEPLAYSMA NYPARTS?HISACTSBEINGSEVENAGESATFIRSTTHEINFANT?MEWLINGANDPUKI NGINTHENURSESARMS?THENTHEWHININGSCHOOLBOYWITHHISSATCHEL?AN DSHININGMORNINGFACECREEPINGLIKEASNAIL?UNWILLINGLYTOSCHOOL eventhoughthechoicebetweenalternativeexpressionswillalwaysdependuponcontextitwouldbetr uetosaythatordinarypeopledoperceiveexpressionstobesomehowintrinsicallyeitherorthophemisti ceuphemisticordysphemisticforexampletermsfordiesuchaspassawayandsleepareeuphemisticwh ereascroaksnuffitandpegoutarenot There were also boustrophedonic (from bou-strophos “ox-turning”) texts, in which text went left-to-right then right-to-left, e.g. (using an example in English3 ) IHAVEADREAMTHATONEDAYTHISNATIONWILLRISEUPANDLIVE OTSHTURTESEHTDLOHEWDEERCSTIFOGNINAEMEURTEHTTUO BESELFEVIDENTTHATALLMENARECREATEDEQUAL In very ancient Greek new paragraphs were sometimes marked by a stroke, |, speeches by different personae by a dash, –. By the third century BCE three punctuation marks were in use: the high point ? marking a full pause (teleia), counterpart to our full stop (period); the mid point · (mes?) marked a long pause roughly counterpart to our semicolon; and the low point . (hupostigm?) was roughly counterpart to our comma. It was common for majuscules (upper-case letters) only to be used – as on the Rosetta Stone (Figure 2.1). On papyri cursive script was sometimes used and it was this that gave rise to minuscules. Initial capital letters were used for proper names and often to mark the beginning of a paragraph or quotation, but 3. A Greek example would more likely begin right-to-left. Figure 2.2. Egyptian demotic papyrus c. 1210 BCE (it is read right to left). 24 The Western Classical Tradition in Linguistics not otherwise sentence initially. Words were not regularly spaced apart until about the fifth century CE. 

Modern punctuation arose with the development of printing in the fifteenth century. Because they were handwritten, no two copies of an ancient text are identical; they needed to be cross-checked for differences, especially for scribal omissions, additions, errors, and emendations. The first step towards understanding is identifying the letters; the second is to identify the word and sentence boundaries. In poems especially, the identification of syllables is necessary for metrical purposes; although at a time when reading aloud was the norm, it was almost as necessary for reciting prose works. The alphabet was learned in sequence alpha, beta, … omega and backwards omega, psi, chi, … alpha and paired as in alpha-omega, beta-psi, gamma-chi, etc. Familiarity with the alphabet was important because the ancient Greeks used letter symbols for numbers (as did the Romans). The fact that there were seven vowels was correlated with the seven notes on their musical scale and the seven celestial bodies known as ‘planets’, namely the Sun, Moon, Mars, Mercury, Jupiter, Venus, and Saturn.4 The term letter and its translation equivalents were used from ancient times through to the early twentieth century to refer to one or all of the three properties in Table 2.1. Table 2.1. The three constituents of a letter. GRAMMA (Gk) LITTERA (Lat.) LETTER example onoma nomen name alpha charakt?r figura form ekph?n?sis potestas pronunciation /a/ Likewise, Ælfric, Abbot of Eynsham, c. 1000 CE wrote ‘æ? lc stæf hæf þr?o ðing: [...] nama and hiw and miht’ (Ælfric 1880: 5). Thus letter denoted the combined orthographic form, name, and pronunciation, all of which needed to be known. Letter was an abstraction; the orthographic form and pronunciation are manifestations of the abstraction in much the same way that a phone is the manifestation of the abstraction we call a phoneme. In ancient Greece, once the letters were mastered, the child learned to identify syllables, starting with two-letter syllables, next three-letter, and then practice with nonsense syllables. After this came the identification of one-syllable words before moving up to multi-syllable words. Included in the syllabus were exercises on proper names, archaic words in Homer’s Ionic dialect, tongue-twisters, and phrases made up from every letter of the alphabet (an Attic Greek counterpart to the quick brown fox jumps over the lazy dog). From words, students moved on to the recitation and interpretation of texts, often in sing-song chorus. 4. The sequence given here is roughly that which corresponds to days of the week in e.g. English and French (Sunday, Monday, mardi, mercredi, jeudi, vendredi, Saturday). Plato on language 25 Papyrus (made from a grass-like aquatic plant) was expensive and fragile; sheets about 150 mm wide were pasted together into rolls up to about three metres long.5 It was too expensive for use in school, so students copied text onto wooden or waxed boards which could be wiped at the end of the lesson. They practised writing out character-building aphorisms such as The learning of letters is the beginning of wisdom, Homer was not a man but a god; and others on topics such as wealth and possessions, family and friends, the gods, fate and fortune, women, education and mental development. Teresa Morgan 1995: 87f describes these moral exercises as follows: ‘Family and friends are always a boon, although in general women are to be avoided: they are compared variously with lions, wild beasts, the sea and fire; they will trap you, marry you and then drive you to your grave.’ One example of such misogyny (from Marrou 1956: 218) is Seeing one woman giving advice to another, He said ‘The asp is buying poison from the viper’. One hopes that girls were not also forced to write such demeaning things. The Western Classical Tradition in linguistics has two parents: one is ancient Greek philosophy; the other is language teaching. The language-teaching progenitor developed several centuries later than its philosophical partner, and will be examined in due course. For the ancient Greek philosophers, language analysis was a tool: it was a means of identifying relationships between things, how to recognize statements that are true from those which are false, and how to draw valid inferences from statements – singly or in combination. The study of language was a means to several ends, including the development and understanding of dialectic (sound reasoning), logic (defining truth and valid rules of inference), rhetoric (persuasive use of language), and poetry. The method was one still familiar today: analyse the subject matter in order to determine its primitive parts and then figure out rules for recombining the parts. The drive to find primitives led the ancients to question the sources of word meanings, the relationships between words, and the connection between meaning and form. Recognizing that language combines form with meaning, the ancients used both semantic and formal criteria for linguistic analysis. Grammatical analysis developed out of advances in dialectic, which required knowledge of propositional types, propositional structure, the forms of propositions, and valid inferences that may be drawn from propositions. For the ancient Greeks, a proposition (the content of a clause) was expressed as a declarative sentence (one with a potential truth value and therefore also called an assertion, judgment, or statement). In other words, the symbolic form of a declarative proposition is a natural language expression. Consequently, the analysis of the forms and structures of declarative propositions (single or combined) was concomitantly an analysis of language forms and structures. The first person known to have made any observations that can properly be said to be ancestral to the Western Classical Tradition is Protagoras (481–411 BCE). I am of the opinion, Socrates, [Protagoras] said, that command of poetry is the principal part of education; by which I mean the ability to understand the words of the poets, to know when 5. Books, codices, only appeared in the first century CE for compact editions of very long works such as the Bible. Paper is said to have been invented in China by Cài Lún (??)c. 50–121. It reached the Middle East in the tenth century and Europe in the eleventh or twelfth. 26 The Western Classical Tradition in Linguistics a poem is correctly composed and when not, and to know how to analyse a poem and respond to questions about it. (Plato, Protagoras 339a)6 The importance given to skill in composing, understanding, and reciting poetry in ancient Greece should not be underestimated (see Chapter 5 for discussion). Attention to such skills, and the teaching of them, had a large part to play in the development of linguistics, as we shall see. Aristotle wrote: A fourth rule is to observe Protagoras’ classification of nouns into male, female and inanimate [genders]. (Rhetoric 1407b 7, Aristotle 1984) In Sophistical Refutations 173b 19, Aristotle writes that Protagoras spoke against the mismatch between natural and grammatical gender, suggesting that the feminine nouns m?nis “wrath” and p?l?x “helmet” ought to be masculine, and that someone who calls m?nis a “destructress” (oulomen?) commits a conceptual solecism even though it is grammatically correct in gender agreement; conversely, someone who calls m?nis a “destructor” (oulomenos) commits a grammatical solecism but not a conceptual one. According to Diogenes Laertius, [Protagoras] was the first to distinguish the tenses of verbs [pr?tos mer? chronou di?rise]. He was the first to classify utterances into four [moods], namely: optative-subjunctive mood [euch?l?], interrogative [er?t?sis], declarative [apokrisis], imperative [entol?]. (Diogenes Laertius 1925, 9: 52, 53–54). We can now see these observations as precursors to the founding of the Western Classical Tradition by Plato, Aristotle, and the Stoics. Plato (429–348 BCE) was an aristocrat. His mentor Socrates, who participates in many of the dialogues as Plato’s mouthpiece, was condemned to death in 399. Plato immediately left Athens and travelled in Greece, Egypt and Italy, returning in 387 to found the Academy, a school dedicated to the Muses7 where students were prepared for public service by studying mathematics, dialectic, rhetoric, and natural science. After devoting his whole life to the Muses a man could confidently count on their patronage when he came to die; for they would summon him into their presence and lead him into the 6. The complete works of Plato are to be found in Plato 1997, but the translations used this book are not necessarily taken from it. Works by Plato and Aristotle use a universal referencing system based on published Greek texts. For Plato the system of pagination is based on page numbers in the various volumes of a 1578 edition of Plato by Henricus Stephanus (Henri Estienne) published in Paris. The numbers need to be used in conjunction with a title in order to make any sense. For Aristotle pagination is based on Immanuel Bekker’s 1831 edition of the Greek text. All editions and printings of Plato and Aristotle use the same numbers, though there is some slight variation. For instance, the reference below to Aristotle’s Rhetoric 1407b 7 refers to the second (b) column of the original page 1407. The final ‘7’ following the superscript b is the line number, and this does vary a bit from publication to publication. 7. There were at least nine Muses: Calliope was the Muse of epic poetry; Erato the Muse of lyric poetry; Polyhymnia the Muse of sacred poetry and mime; Euterpe the Muse of music; Terpsichore the Muse of dancing and choral song; Melpomene the Muse of tragedy; Thalia the Muse of comedy; Clio the Muse of history; and Urania the Muse of astronomy. Plato on language 27 astral spheres along with all those other souls who on earth had been similarly prepared for that great honour. (Marrou 1956: 145) Plato championed as a liberating force true knowledge based on rigorous demonstration; consequently he was unusual in condemning the myths propagated by poets (Republic X: 607–608a). Plato’s philosophy is largely presented in dialogues with at least two sides of an argument being offered; it is the engagement of (often conflicting) minds from which the truth emerges. A monologue is more likely to offer a biased or prejudiced view. Plato on meaning and grammar He first considered the meaning of grammar. (Diogenes Laertius 1925, 3: 25) The ancient Greek scientific method was to analyse the subject matter into its primitive parts in order to facilitate an understanding of it. We see this at work in Plato’s Sophist 218e–221c, where he seeks to define aspalieut?s “angler” by such questions as (1) and (2). (1) Is an angler X or not-X? (2) What is X(ing)? As usual in a Platonic dialogue, the wiser man (Plato’s mouthpiece) quizzes his interlocutor in such a way that enlightenment is reached through dialectic. In the following excerpt, the quizmaster primes the correct answer by establishing the pattern: Question 1 or 2? Answer 2. Is angling not a skill or a skill? It is a skill. Is the angler a creator or an acquirer? An acquirer. Is the acquisition by consent or by capture? By capture. Is it open-capture or stealthy-capture? Stealthy capture. Is it capture of nonliving or of living things (z?oth?rik?)? Of living things. Are the living things land animals or water animals? Water animals. Are the water animals caught waterfowl or fish? Fish. Are the fish caught by a net or by striking a blow? By striking a blow. Is this done by using fire at night or by using barbs in the day-time? Using barbs by day. Are the fish struck by a trident from above or struck from below with a hook? From below, with a hook. The procedure could be represented, if somewhat inadequately, by a series of choices between binary features such as ±skill, ±acquisition, ±capture. In fact the quizmaster uses 28 The Western Classical Tradition in Linguistics something of the sort when summing up the definition of angling: Of skill as a whole, half was by acquisition; and of the acquisition, half was by capture; and of capture, half was by stealthy capture; [... etc.] (221b) The definition of aspalieut?s is “a skilled person who, by stealthy capture acquires living water animals, namely fish, by striking them from below with a hook during the day-time”. There was no intention that this be a semantic description of the word aspalieut?s; Plato’s purpose was to fix what a contemporary angler did to merit the title aspalieut?s. The result has a lot in common with defining that part of the semantic content of a lexical item which picks out relevant features of a typical denotatum.8 However, a semantic description additionally identifies the semantic relationships of the term – which would include comparing its meaning with those of related words.9 Plato’s statements on grammar were very basic and perhaps not original (Householder 1995b: 92). In Sophist 26le–263 Plato identified two phonic signs for the essence of things: onoma [pl. onomata] “name”, which is often translated “noun” but is usually understood as the equivalent of “noun phrase”; and rh?ma [pl. rh?mata] “attribute, verb, predicate”. The rh?ma denotes an action; those who do the actions are signified by onomata (262a,c). Any combination of onoma and rh?ma produces a logos “sentence, proposition” also translated ‘speech’, ‘utterance’, ‘statement’ (262c). Every logos is ‘of something’ (tinos = ti+GENITIVE); i.e. a logos has an argument or topic named in the onoma. Moreover it is about something which is, or is becoming, or has become, or will be (262d). This implicitly recognizes tense without assigning it to any particular sentence constituent, as Aristotle later did (it is not known whether Aristotle built upon Plato’s ideas). A logos does not simply name something, it makes an assertion about what is named that is either true or false (263d). The reason that Plato divides up the logos is apparently to show that the name can be correct even though the statement is false (263c, and see Rijk 1986: 206f). That is the substance of what Plato says about grammar. Of greater interest to the modern linguist is Plato’s Cratylus and his treatment of the problem of universals and abstract entities. Plato on the relationship between meaning and form in language The power of words derives from the connection between the words and the things they denote. A word combines form with meaning: e.g. the sequence of phones [k], [æ], [t] 8. What expression e denotes is what e is normally used to refer to: e.g. the phrase a cat denotes a cat (a feline mammal). We say that cat (the animal) is the denotatum of cat. The plural of denotatum is denotata. 9. It is a matter of logic and not linguistics, but the procedure for definition that Plato uses here is not readily transferable to other nouns because the sequence of questions asked is task specific: one can only relate the first question to the final answer if one already knows all (or most) of the intervening stages. The procedure serves to make explicit what kind of skill hooking a fish is, but the steps in the reasoning are not spelled out through rules of procedure. Aristotle’s rules for the construction of syllogisms would do so; see Chapter 3. Plato on language 29

combines with the meaning “feline animal” into the word cat. People tend not to separate the form from the meanings of a word, and moreover they tend to associate the meaning of a word with its denotatum. Saussure proclaimed as the first principle of linguistics that the correlation between the form of a language expression and its meaning is arbitrary, and also conventional in the sense that everybody in a language community tacitly concurs in using a certain form with certain meanings, cf. Saussure 1931: 100. The notion is not a recent one; here is Aristotle: A word signifies this or that by convention. No sound is by its nature a word, but only by becoming a symbol. Inarticulate noises such as are made by brute beasts may mean something: but no sounds of that kind are words. [...] Every sentence has meaning, not as an instrument of nature, but – as we observed – by convention. (On Interpretation 16a 30; 17a 1) This seems obvious enough: how does one explain (3) if there is supposedly a connection between the meaning “canine animal” and the form of the word bearing that meaning? (3) A canine animal is called dog in English, chien in French, Hund in German, pies in Polish, ájá in Yoruba, kare in Hausa, mbwa in Swahili, and so on and so forth for all languages. Yet human beings seek to explain many of the things that confront them in terms of causal relations. If the form~meaning correlation in a word is arbitrary, any causal relationship between the form and meaning of a word is denied. It is understandable, therefore, that people have postulated a causal relationship between the two. Over the ages many welleducated scholars, as well as ordinary people, have believed that the original meaning of a word gave rise to its original form: hence the study of word-history was called etymology, whose ancient Greek morphological roots mean “the study of true original form and meaning”.10 The belief embodies a hypothesis for the natural connection between meaning and form in which the form somehow derives from the nature of the denotatum so as to communicate its essence: for this reason the hypothesis is known as the ‘naturalist hypothesis’.

The naturalist hypothesis is worth considering since it is only by refuting it that we can demonstrate that word forms are correlated with their meanings on a purely arbitrary basis. The most important discussion of the relative merits of naturalism (phusis) versus conventionalism (nomos or sunth?k?) 11 is Plato’s Cratylus.

 Dating from c. 385 BCE this is the oldest extant European work on a linguistic topic. 

Plato’s purpose was to question whether it is valid to study the natural world through discussion of the language denoting things in the world. Were the naturalist hypothesis correct, then a word would reveal the essence of its denotatum, so that the study of language would be as valid in the quest for 10. ?????????? is based on ?????? “correct (or true) [form and meaning]” and ????? “discourse (about), study of”. 11. Phusis shifted meanings through “genesis” to “nature” and later “Nature”; nomos from “allotment” to “proper customs, law” to “custom”; cf. Joseph 1990a. 

The usual word for “custom” is ethos and for “convention” sunth?k?. 30 The Western Classical Tradition in Linguistics knowledge about the natural world as a study of the world itself; on the other hand, if the conventionalist hypothesis is correct, then talking about the world cannot take the place of studying it directly through the physical sciences. So, Plato had a practical philosophical purpose in comparing the naturalist hypothesis with the conventionalist hypothesis for the correlation of the form and meaning of words; he was not indulging in idle speculation about language. Cratylus12 is a Socratic dialogue in which Socrates questions the proponent of a thesis about its meaning and consequences, seeking to define the meaning of key terms, reasoning from these and by continued questioning to uncover contradictions or absurdities. 

In the first part, Plato’s mouthpiece Socrates, who does most of the talking, argues the naturalist hypothesis against the conventionalist Hermogenes (385–427). In the second part (428–40), Socrates refutes naturalism in discussion with Cratylus. We shall not stick rigidly to the text of Cratylus, but begin with the question of the origin of words and the diversity of languages; then examine the procedures for naturalist etymologies and the naturalist account of the supposed basis for the natural connection between meaning and form; then, finally, consider the remnants of naturalism today, and its implications. A problem for conventionalism is to explain how the original correlation between meaning and form – if it is arbitrary – became conventionalized. The naturalist explanation for the spread of words through the community is that people straightforwardly perceive the natural connection between form and meaning. But who coined the original words? According to Socrates in Cratylus, it was a wordsmith (nomothet?s – etymologically “lawgiver”; in 389a Plato instead uses onomatourgos “name-maker”), a craftsman comparable with the blacksmith or the carpenter. It had to be a craftsman and not just anybody because a word is an instrument of teaching, and for separating the natures of things signified, like the shuttle is an instrument for separating the threads in weaving. (388c) The wordsmith fashioned words from sounds in accordance with the natures of the things signified, just like the blacksmith forges tools from iron, or the carpenter shapes wood into a shuttle. If you are wondering how far back one has to go to find the original words, Socrates favoured polygenesis, claiming that every language had its own wordsmith and each wordsmith used different syllables and sounds to form the words, just as different pieces of iron are used by different blacksmiths to make the same kind of tool. 

This neatly accounts for the fact that different languages have different forms with the same meaning, as in our “dog” example, (3) above. But the notion of each language being invented by a wordsmith needs considerable revision in the light of present knowledge of the tangled web of relationships among languages. Who were the wordsmiths? Under the naturalist hypothesis, in order to ply his craft, a wordsmith would have to know and understand the Platonic Idea (or Form), which in this context means the true essence of a thing. In Plato’s view ordinary people are not capable of 12. There is a useful summary of Cratylus in Matthews 1994; see also Joseph 2000b for more extensive analysis. Plato on language 31 this (Letter VII 340e–344b), so the wordsmith could be no ordinary man (Plato did not admit the possibility of it being a woman). Nor is he a god, because in Cratylus 439c Socrates shows that the wordsmiths are inconsistent and fallible: they cannot therefore be gods.

So Plato leaves us no choice but to conclude that the wordsmith was a purely hypothetical construct, a ‘straw man’ who could not have existed at all: therefore there has to be some other explanation for the origin of words. It is already implicit that the naturalist hypothesis cannot be sustained because the form of words cannot directly reveal the Idea of what they denote, and therefore the nature of a thing cannot be known from its name. We turn now to the etymological procedures used by naturalists. They sought a ‘natural’ or ‘true’ connection between meaning and form, but they did not go so far as pretending that a word duplicates its denotatum: obviously word and denotatum are completely distinct (cf. 432a–d). Nor was the word form thought to be a matter of sound mimicry: otherwise onomatopoeic words like moo and cock-a-doodle-doo would ‘name that which they imitate’ (423c). What a ‘true’ word form must capture is the essence (ousian) of the denotatum. For instance, according to biblical legend the first man was named Adam because he was created from the earth (Genesis 2:7), which is called in Hebrew adamah. Although this demonstrates a connection between Adam and adamah (not widely accepted today), it leaves an important question unanswered: can the Hebrew form adamah be connected with the meaning “earth” in any natural way? This sort of question can be asked for each of Socrates’ etymological analyses (at least some of which go back to Anaxagoras, c. 500–428 BCE). They are extremely fanciful, with a cheerful disregard for the transposition of letters, their omission, or insertion. Socrates’ excuse, reused by etymologists for the ensuing 2,100 or more years, was that the original names have been completely buried by those who wished to dress them up – for they have added and subtracted letters for the sake of euphony, and distorted the words in every way for ornamentation. The passing of time also plays a part in this process. (414c) 

Let’s take just a couple of examples from Cratylus. According to Socrates the goddess of wild nature, the hunt, chastity and childbirth, Artemis appears to get her name from her healthy [artemes] and well-ordered nature, and her love of virginity; or perhaps he who named her meant that she is learned in aret? [“virtue”], or possibly, too, that she aroton misei [“hates sexual intercourse”] between man and woman; or he who gave the goddess her name may have given it for any or all of these reasons. (406b) It is possible that Artemis and artemes are linked; but there is no plausible formal connection between Artemis and either aret? or aroton misei. The name of the god of Bacchanalia, Dionusos, derives, says Socrates, from didous ton oinon meaning “giving wine” (406c). This would be very appropriate but it is a mystery how the three-word phrase was pared down in the one-word name. Greater imagination still is required to explain in any systematic way the collapse of anathr?n ha op?pe “looks up to see” into anthr?pos “man” (399c). Such proposals were intended seriously; Socrates was not simply poking fun at naturalist etymologies. Consider the naturalist etymologies constructed for two Latin words by Marcus Terentius Varro around 45 BCE. 32 The Western Classical Tradition in Linguistics Cervi [“stags”] because they gerunt [“carry”] big horns derives from *gervi [“?carriers”]; the word has changed G to C as has happened in so many words. [...] Volpes [“fox” is so-called] because it volat [“flies”] with its pedes [“feet”]. (De Lingua Latina V: 101; Varro 1938) Although Varro was a very learned man (see Chapter 4), both these etymologies are inaccurate in fact. In the first it is claimed that the verb gerere “carry” gave rise to the plural noun gervi, although in fact no such noun existed; nor was there any systematic unvoicing of initial stops in Proto-Latin. The second example seductively claims that volpes is a blend of volare “fly” and pes “foot”; that is, the word for fox derives from the description “fleetfoot”. Etymologies of this sort flourished in ancient and medieval times. 

However, we should note that criticizing these ancient etymologies from a modern standpoint mistakes their intention: modern etymologists seek to map the diachronic development of the meanings and forms of the word whereas the ancients sought to explain the meaning of the word in terms of its perceived component forms (Robins 1997: 27). They assumed that knowledge is embodied in word meanings and can be elucidated by reference to the original meaning; hence the original forms and meanings of words in what would today be called the proto-language were, supposedly, finessed. Although their explanations are often faulty, as we have seen, it is arguable that they had some success in focusing attention on the meaning of the word under consideration.13 Where the etymologist could not come up with any kind of componential analysis of the kinds exemplified above, he would conclude that the word under analysis was of foreign origin, and its etymology could only be given in terms of its language of origin, cf. Cratylus 409d–410a. Thus, the naturalist gave his imagination free rein to etymologize without any kind of restraint. The method of analysis was: * describe the denotatum of the word under analysis; * cast around for an appropriate phrase that bears some resemblance to the description, and which has a form bearing some resemblance to the word under analysis; * if this fails, the word must be of foreign origin and is therefore unanalysable. Such fanciful procedures brought etymology into disrepute until more rigorous methods were adopted by philologists from the seventeenth century.14 Most of the etymological investigation undertaken by Socrates in Cratylus is analysing words (however fancifully) into a combination of semantic components that he called ‘primary words’ and which we might think of as the precursors of morphemes, cf. the derivation of Adam from adamah, or of volpes from vol[are] + pes. 

The problem is to show a natural connection between the forms of the primary words and their meanings. But all that naturalism can offer is that primary words are constructed on the basis of sound symbolism: sound symbolism is the foundation on which the whole edifice of the naturalist hypothesis rests. Plato’s degree of confidence in this foundation is indicated by Socrates’ well-justified comment: 13. There is further discussion of etymologia in Chapters 4, 7, and 13. 14. One of the earliest systematic etymologists in the modern sense was Ménage 1650. Plato on language 33 I think my notions about the primary words are quite outrageous and ridiculous, though I have no objection to imparting them to you if you like, and I hope that if you can think of anything better, you will tell me. (426b) Sound symbolism is language specific and conventional, not natural. For instance, onomatopoeic words, which supposedly mimic natural sounds, differ from language to language and obey the phonological conventions of the language in which they occur: cf. English cock-a-doodle-doo, French cocorico, German kikeriki, Japanese kokekokko; and compare English clang with Tzeltal15 c/an, English chip with Tzeltal c/ehp, screech with kic/, and so forth. Languages have phonaesthetic networks which differ from language to language: e.g. the English words flail, flame, flap, flare, flash, flay, flee, flick, fling, flit, flood, flop, flounce, flourish, flush, fly have the common consonantal onset ‘fl–’ and all suggest sudden or violent movement; bash, clash, crash, dash, flash, gash, lash, mash, slash, smash, thrash have the common rhyme ‘–ash’ and all involve violent impact (Allan 2001: 132–40). 

According to Socrates, naturalism is founded on something like phonaesthesia; consider such postulates as the following: r represents motion16 (e.g. rhein, rho? “flow [verb, noun]”, tromos “trembling”, tr?chus “rugged”); s and other fricatives are pronounced with ‘great expenditure of breath [... and] imitate what is windy’ (e.g. phus?des “windy”, seiesthai “to be shaken”, seismos “shock”, zeon “seething”, psuchron “shivering”); l has a liquid movement in which the tongue glides and expresses smoothness (e.g. leion “smooth”, olisthanein “glide, slip”, liparon, “oily, sleek” (426d–427b). But individual phonemes or letters are not consistently used in a particular sense. E.g. the word smooth has fricatives at either end, but it does not indicate something ‘windy’; lollop, laugh, hall, and toll contain ‘l’ but no sense of ‘smoothness’ or ‘slipperiness’; the ‘r’ in rust, rot, and round has nothing to do with motion; the ‘fl–’ in flint, flock and flower brings no sense of sudden or violent movement to these words: nor do ash, cash, sash involve violent impact. Indeed, when arguing against naturalism in 434e–435c of Cratylus, Socrates himself cites counterexamples like these to force the conclusion that the form~meaning correlation in a word must be conventional and arbitrary. For instance, he points out to Cratylus ‘that which is called by us skl?rot?s is by Eretrians [who lived on an island northeast of Athens] called skl?rot?r’ and these are mutually intelligible and both readily understood to mean “hard”, therefore neither can be judged ‘incorrect’. Also the word in both dialects contains l which seems at odds with the meaning “hard”. 

Cratylus is forced to agree. Socrates concludes that it does not matter what the form of a word is if its meaning is ‘sanctioned by custom and convention’.17 Today, controversies over the earlier forms of words are generally left to experts trained in historical and comparative linguistics; but even so, there is a strong body of public 15. Tzeltal is a Mayan language, spoken in Mexico; cf. Berlin 1968. On sound symbolism see Hinton, Nichols and Ohala (eds) 1994; Allan 2001: 132–40. 16. Plato’s word for motion is kin?sis which itself contains no r. 17. Joseph 2000b: 75 points out that Plato would prefer naturalism to be the guiding principle because it approaches the Idea, but he recognizes that convention prevails. Plato disapproves of this because he holds the populace in very low esteem (435c2–d1), see Republic and Laws. 34 The Western Classical Tradition in Linguistics opinion that the proper meaning of a word is the supposed original meaning. This is revealed in school textbooks and frequent letters to newspaper editors or calls to radio talkback shows asserting that the new meanings for words are ‘misuses’. This is the view expressed in the following remarks about the English word nice. nice This word is very much overworked and misused. Its real meaning is precise, exact, and delicately, fine; e.g. A nice difference in meaning. A nice ear for music. The word is now often used to mean agreeable, delightful, pleasant, etc. – because it is easier to say nice than to think of a more suitable word. (Wright 1978: 91) You should ask yourself what Wright means by saying ‘it is easier to say nice than to think of’ some other word. He possibly means “lazier”; but easier also implies greater communicative efficiency, which is laudable not reprehensible. 

Regrettably, Wright’s remarks about nice demonstrate a woeful ignorance. Today, the ‘real meaning’ of nice includes “agreeable”, “delightful”, “pleasant” just as much as (and arguably more than) it includes “precise”, “exact”, and “delicately fine”, which were its seventeenth and eighteenth century meanings and are a shade archaic today. English nice derives from Latin nescius “ignorant”, and we can trace a path through its earlier senses to its present meaning: nice from Latin nescius ? ne sc?re “not know”, “ignorant” 14th –16th century “ignorant”, “stupid”, “foolish”, “foppish” 16th –18th century “foppish”, “fastidious”, “precise” 18th –19th century “precise”, “balanced”, “agreeable” 20th –21st century “agreeable”, “pleasant”, “pleasing” To speak of one or more of the earlier meanings being the ‘real’ meaning of nice is absurd. It suggests that the current meaning is in some way degenerate. The degeneracy of contemporary language is, in fact, a recurrent theme in the naturalist tradition. The naturalist hypothesis leads to a belief that the original word bore the proper form and proper meaning; therefore any subsequent change is a degeneration from the perfectly natural original form, a decline that should be halted or preferably reversed. However, we should ask the naturalist for evidence that the original language (or the original form of his or her language) was a better instrument of communication than the language of today, or, indeed, of another time in recorded history. The answer is, of course, that there is none. Inspired by Plato’s Cratylus, we have seen that the naturalist hypothesis will not stand scrutiny: the forms of words do not capture the essence of their denotata; in fact there is no natural connection between the form of a word and its meaning. As Plato wrote in Letter VII: There is nothing to stop things which are at present called round being called straight, and vice versa; and their stability would be in no way impaired if everyone made this transposition. (343b) Plato on language 35 Particulars, universals, and abstract objects Plato’s interest in abstract objects arises from practical questions for his epoch such as How do we teach people courage? This question presupposes that we can recognize courage and bring others to know what courage is. How can anyone know what courage is when the best we can offer as evidence for our knowledge are instances of acts of courage from the past and, on rare occasions, the present? As we shall see, the difficulty extends to the language we use when talking about concrete entities. Consider Plato’s discussion in Letter VII 342a– 343b of a circle, which we can reasonably take to be an inquiry into the denotatum of the lexeme circle. It is important to be clear that what we are talking about here is the lexeme circle plucked from the dictionary (as it were) and not being used to refer to some particular object such as the noun phrase a circle does, which can be drawn ? and referred to a second time by the noun phrase the circle ten words back. The circle drawn here and, by now, three times referred to is a particular object of our experience. By contrast the denotatum of the lexeme circle is not; it cannot be drawn because it is an abstraction. In Plato’s terms it would be an Idea. Plato identifies three ways that we acquire knowledge of circles: through the name, through definition, and through sensory experience. The name circle is arbitrary and could have been different (it is kuklos in Greek, da’ira in Hausa, yuán in Mandarin) yet still denote the same thing. (4) A circle is a plane figure bounded by a line (the circumference) equidistant from a point (the centre) such that if the distance from the centre to the circumference is r, then the circumference has a length of 2?r and describes an area ?r 2 . But a lot of people know what a circle is without knowing the definition in (4). The Idea, which Plato refers to as ‘the circle itself’, is what is named in the lexeme and defined in (4). As we have seen, you can draw a circle, point to one, imagine one; but in each case it is just one instance of a circle. 

You can draw lots of (instances of) circles and then rub them out (or tear them up and burn them): such acts have no effect on the existence of the denotatum of the lexeme circle or the Idea of a circle. No instance of a circle matches the ideal: by definition a circle has no straight edge, yet any representation of a circle will be straight over some extent of its circumference, however small – it is, in other words, as impossible to draw the perfect circle as to draw a line that has length but no breadth. For practical purposes we overlook this fact; but it is interesting, because it does raise the question of what a circle really is. Plato lists ‘knowledge’ of what a circle is together with ‘understanding’ of and ‘true belief’ about what it is. According to Plato, knowledge of something involves understanding it, and cannot be false. True belief about something is based on assessment and interpretation, and may or may not involve an understanding of it; it does not have the permanent status of knowledge and can be an illusion – in which case it is presumably false belief. In Letter VII 342c Plato does not need to distinguish among knowledge, understanding, and true belief because he is presenting them as bound to a mortal person (however perspicacious) and necessarily therefore impermanent conceptions quite different in nature from the object of knowledge, understanding, and belief – the Idea, the ‘circle itself’. The Idea exists (or subsists) in another world, or on another plane, a 36 The Western Classical Tradition in Linguistics perfect and immortal world. Plato believed in the immortality of the soul, which in some pure state comes in contact with the Ideas so that when a soul is attached to someone s/he is able to use it to recollect the form. This is known as the doctrine of anamnesis. They say that the human soul is immortal; at times it comes to an end, which people call dying, at times it is reborn, but it never perishes. [...] So, since the soul is immortal and has been born again many times, and has seen both the things here on earth and those in the underworld and all things [including Ideas], there is nothing that it has not learned. So it is in no way surprising that it can recollect the things it knew before. (Meno 81b,c) As human beings we can only hope that our souls will recollect Ideas when stimulated by our knowledge and understanding of the world we experience. We readily interpret ? as a circle: our eyes tell us what it is and that it is the same kind of thing as ?, and we have learned that such things are called circles. 18 These are things we know from direct experience. It is much more difficult to figure out how we come to know the meanings of (5) to (7) because we cannot know any of these from direct experience. (5) the universal all circles (6) the description a perfect circle, and (7) the denotation of the lexeme circle To say that we extrapolate from our direct experience, or abstract from it, or generalize on the basis of it, at best describes what we do but nonetheless fails to account for our understanding of (5) to (7). Plato’s solution was that our knowledge of such things comes through recollecting the direct experience of our immortal soul. Thus Plato offers an explanation for how it is that everyone knows more than their direct experience in the world seems to warrant.19 It is a question raised by mother-tongue acquisition: somehow every child learns the grammar of their language(s) despite haphazard and impoverished linguistic input. To assume that the child’s immortal soul recollects language experienced in some former life would surely lead it to sometimes use archaic grammatical constructions; instead there is thought to be some biological cause for language acquisition. A better explanation for our understanding of universals, abstract entities, and partially perceived objects is that humans are biologically programmed to abstract and to generate gestalts. As to the understanding of universals, there has long been a great deal of controversy between two points of view. One is that things can be classed together because they all have some common characteristic(s); this is known as the ‘realist’ view because it presupposes that things are similar in the real world, and we therefore perceive them or conceive of them as similar. This was the view adopted with the two circles in the paragraph beginning above (5). The alternative is the ‘nominalist’ view that things are classed together only because they have the same name – an idea from Democritus (460–390 BCE), Gorgias (483–376 BCE), and the Epicureans (second century BCE) (Carré 1946: 41). Nominalism can smack of 18. The visually impaired need to replace visual perception with tactile perception; this complicates matters, but does not destroy the argument being made. 19. Chomsky 1988: 3f describes this as ‘Plato’s problem’; see Chapter 8. 

Plato on language 37 naturalism in that the name captures the Idea of the denotatum; but the usual version (shared with Peter Abelard, 1079–1142, and William of Ockham, 1300–49) has particulars as real world objects external to the human mind, and universals as either logical or conceptual entities. Thus abstract entities can exist (or subsist) in some world other than the real world, but one linked with it through the human mind. Plato’s view, recall, is that abstract entities exist as Ideas in a different world, accessible not to the mortal conscious mind, but through the immortal unconscious soul. The view that abstract entities exist in an immortal world has been advanced by Platonist linguists in the twentieth century (Carré 1946; Katz 1981a; (ed.) 1985; Katz and Postal 1991). Plato on meaning, form, and understanding Plato has little to say about grammar apart from dividing the sentence into a noun phrase naming the actor and a verb or predicate naming the action. The idea was probably not original with Plato, but we will pretend it was. 

The argument in Cratylus centres on the extent to which we can know the essence of things through knowing the words that name them. The argument is between the competing phusis or naturalist hypothesis and the thesis or conventionalist20 hypothesis. Naturalism claims that the form of a language expression was originally determined by its meaning and therefore change of form is degenerate; conventionalism claims that meaning is arbitrarily related to a form, and assigned to that form by tacit convention within a language community. When this Socratic dialogue is read carefully, there is plenty of evidence that the weight of argument is on the side of conventionalism even though Plato was not decisive or explicit about which side of the fence he was on. He presents the dialogue in a typically equivocal way, giving arguments for both sides, because his purpose was in part to teach through discussion. He firmly believed that the truth is reached through dialectic, discussion, and the exploration of ideas. It is truer of Plato than of many scholars that he argues a point rather than laying down the law. We saw that the naturalist hypothesis is founded upon sound symbolism; and although some vocabulary in every language reflects sound symbolism, this is determined by the phonological conventions of the language and the meaning of sound-symbolic vocabulary items is not regularly recognizable by someone who does not speak the language. 

The discussion of circles (roundness) in Letter VII shows Plato to be a conventionalist; but a few scholars who believe that Cratylus demonstrates Plato was a naturalist dispute the authenticity of this part of Letter VII. 

At the same time, however, Plato discusses the problem of abstract meanings in the context of asking how we know what a circle is. 

Discussed by H. P. Grice.

This discussion is relevant to the problem of how children learn the meaning of lexemes like circle and universals like all circles when they only experience imperfect instances of circles. 

In fact, the Greek term allows for the system to be created by one person and imposed on others, so it might be more correctly translated INventionalist; cf. Allen 1948: 36. But conventionalist is the standard translation. 

38 The Western Classical Tradition in Linguistics 

More generally, it raises the question of how children (and therefore adults) come to 

know so much more than seems warranted by the haphazard and impoverished sensory data that they encounter by direct experience.