Chapter 1 Introduction
The aim of this book is to explain how the grammarians of the Graeco-Roman world perceived the nature and structure of the languages they taught. It is addressed in particular to linguists of the present day, primarily in western countries, and I write as one such linguist, not as a specialist in Classics. I cannot among other things assume that every reader will know Latin, let alone Greek. I will assume, however, some basic understanding of linguistics, and will refer for comparison to ideas current in this century and the last without explaining them in detail. e task may seem at rst sight to be easy. Some ancient texts have been translated, o en in terms that to a modern reader are in their modern senses perfectly familiar. For Greek onoma or Latin nomen a translator into English will write modern ‘noun’; for Greek sundesmos or Latin coniunctio the modern ‘conjunction’; for Greek lexis or Latin dictio the modern ‘word’; for Greek gramma or Latin lit(t)era the modern ‘letter’; and so on. For none of these terms are the equivalences exact. Ancient ‘nouns’ included adjectives, and ‘conjunctions’ in Greek included words that a modern treatment will class separately as ‘particles’. An ancient ‘letter’ was a unit as much of speech as of writing. Even, however, when such di erences have been acknowledged, the
What Graeco-Roman Grammar Was About. First edition. P. H. Matthews. © P. H. Matthews 2019. First published 2019 by Oxford University Press.
Introduction
history of grammar can be seen as one of individual re nements and improvements, in which scholars of successive eras have drawn distinctions that their forebears missed, in which new ndings have been added and new ideas assimilated, in which individual errors have o en been corrected. It is the history of a continuous tradition, in which linguists of our day are labouring in a vineyard that was planted by linguists of the ancient Mediterranean world, and the problems they were addressing are at heart ours also.
In part that is, of course, true. Ancient physics, for comparison, was not modern physics. e ancient theory, for example, of four elements, of earth, air, re, and water, is now simply dead. Yet linguists still talk, if not of the eight parts of speech, of a system of categories that include in large part similar distinctions. eir number varies, as does the basis on which they are established. But where ancient accounts of Latin distinguished nomina and verba modern grammars of, for example, English distinguish syntactic categories called ‘noun’ and ‘verb’. e modern distinction between nouns and adjectives can be seen as one of the same order as those we have inherited from antiquity, such as that of prepositions and conjunctions, or one that the tradition has since demoted, between verbs and participles. Nor does anyone doubt, or seem to doubt, that categories like the parts of speech are fundamental to the study of grammar, in the twentyrst century as in the rst.
Yet why, we might ask, are they so called? e term ‘part of speech’ has as a whole become opaque, but it translates in origin the one in Latin for the ‘parts’ quite literally of what was called an oratio. is can o en be translated ‘speech’: the speeches of the Roman ‘orator’ Cicero were his orationes or ‘orations’. But for the Roman grammarians it referred most nearly to what we might now call an ‘utterance’. e formation of the term was similar: as
Introduction
‘utterance’ is from the verb ‘to utter’, so oratio was formed transparently from a verb whose stem was or(a)-, meaning ‘to speak’. is was in turn related to the word for ‘mouth’ (genitive oris). An oratio was therefore anything said and anything represented, as if said, in writing. Its ‘parts’ (partes) were categories of units into which an utterance was again quite literally divided. We will address this topic in more detail in a later chapter. It is already clear, however, that an apparent continuity in terminology may mislead us into thinking that ideas too are unchanged. e past as always is another country, and the greater the di erences between periods the more our outlook must re ect it. e Roman empire was a society not only unlike ours. It was di erent too from that of the European Middle Ages, and from the way it came itself to be perceived in the Renaissance. Part of the history of grammar, therefore, in the west is of its adaptation to new circumstances and new pressures: to an educational system restricted in the Dark Ages to the church; to the teaching of Latin to speakers of Old English and other Germanic languages; to a new emphasis, at the height of the Middle Ages, on its philosophical foundations; to the development in the early Modern period of standard forms of national languages; to the description of unfamiliar languages in other continents; to the university system as developed in Germany in the nineteenth century; to later preoccupations nearer our own time. Its external history, if we may so call it, is a eld in itself. But its internal history has its own momentum. Any grammar is a partial description of a language, which identi es certain kinds of unit and relations of certain kinds between them. ose established in antiquity in analyses of Greek and Latin were later taken as a model for the description of languages whose structure was in one way or another di erent. Other units, however, and other relations came in time to be identi ed, which have since been taken up by scholars
Introduction
generally. e concepts, for example, of a root and an a x, which were no part of the Graeco-Roman model, were adopted gradually by Europeans from the sixteenth century onwards, from accounts of Arabic or Hebrew. A later, independent model lay in the ancient analysis of Sanskrit, when it became known to western scholars in the early nineteenth century. In response to these and other in uences a modern account of Ancient Greek or Latin, leaving aside all other languages, is di erent in substance from those current een and more centuries ago.
A central aim then of this book will be to make clear what the ancient model was. e term ‘model’ is an anachronism: an ancient grammarian, if teaching Greek, was thinking of Greek alone or, if teaching Latin, was thinking of just it and Greek. He had no professional interest in any other language with which speakers of either might be in contact. He had no motive like that of linguists nowadays, to develop a ‘theory’ of the structure of language that will be compatible with what we know of forms of speech in all societies. e moment, however, one says that a language has ‘words’, that they consist, as ‘words’ in antiquity were seen to consist, of letters and syllables, and that such ‘words’ belong to di erent classes within utterances, a system of grammar is implied which can in principle be abstracted and applied more generally. e texts that survive, in which ideas can safely be identi ed, date at the earliest from the rst century bc; most, however, were written four centuries or more later. ose in Latin include in particular the ones that were to prove most useful in the early Middle Ages, when it was taught increasingly as a foreign language. It is possible, therefore, that they are more homogeneous than they would be if survival had been more random. ey formed part, however, of a continuous tradition, in which grammar as represented by Quintilian, who was born in the formative
Introduction
years of the Roman empire, was still the discipline de ned by Isidore of Seville, writing in the early seventh century ad, long a er the empire had politically fragmented. ese and other authors are identi ed brie y in an appendix. Earlier writers will at times be mentioned, as far back as the owering of philosophy in ancient Athens. But in the later period at least, the model or technique of grammar became in essence frozen.
Before then it did have a history, which belongs especially to the period called Hellenistic, in o cially Greek-speaking states across the eastern Mediterranean that succeeded the conquests of Alexander in the late fourth century bc. It is a history, however, that we do not know directly. Original texts have not survived, and we must therefore rely on subsequent accounts in what would now be textbooks, and on scattered references by various authors, o en second-hand and sometimes hostile, and inevitably in uenced by ideas of their own day. is is true especially of our sources for the Stoic philosophers of the third and second centuries bc, whose theory of language, as understood and ingeniously pieced together by modern scholars, underlay a great deal of what followed. Our earliest extended texts in Greek, by the great Alexandrian grammarian Apollonius Dyscolus, date from the second century ad, and seem to attest a stage when many important details were still being worked out. ey were details nevertheless, and the nature of the discipline had been established already by pioneers of whom we mostly know at best their names. It would be perverse, if it were possible, to ignore this history entirely. e very term grammatica or ‘grammar’ has its origin in Greek in a period of which we have at least a partial understanding. e focus of this book, however, will be on the consensus that was broadly achieved. By the time the Roman empire reached its zenith, under the rule of Trajan and Hadrian and the other ‘good’ emperors,
Introduction
a grammarian had not only a secure place in the ancient system of education, but could take for granted a technical apparatus that was already well developed. Our main task is to try and think about the nature and structure of language in the way that, from their o en voluminous writings, it appears that they thought.
We will o en be forced to that end to suspend preoccupations that belong to later eras. To most linguists nowadays it is obvious, for example, that writing is in principle not speech. We may give illustrations in writing, as grammarians have done from the beginning. But written English or written French has its own structure, which has evolved separately in many respects from spoken varieties. It is an error in this light to think of written sentences as ‘utterances’ or, as linguists did before the twentieth century, of letters as having ‘sounds’. But ancient attitudes had not developed that far. To read in particular was to read aloud, from manuscripts that in general did not divide words. A doctor could therefore prescribe the physical exercise of reading as a course of treatment for some diseases. Reading silently was odd and the practice that to us is normal, of scanning texts at speeds that are o en much more rapid than speech, was facilitated by changes in the way a manuscript was written half a millennium a er the period we are concerned with. To write was to represent, letter by letter, what could alternatively be uttered; to learn to read was to reconstitute a text, letter by letter and syllable by syllable, in its primary form. Compare in that light a modern recording of someone talking. It is strictly not, itself, an ‘utterance’. It records no more than the sound made in an act of utterance, in abstraction from facial expressions, gestures and body postures, and so on. Yet many linguists blithely talk of a transcription of such a recording as their ‘data’. To an ancient writer a con ation of speech and writing might have seemed as innocent.
Introduction
To explain what ancient grammarians were up to, and in that way justify the title I have chosen, we need not argue that they were right. Even, however, where we know or believe that they were wrong, what they wrote may still make sense in the context in which they were writing. We must avoid in particular the temptation to think of them as ancient ‘linguists’: as the equivalents in antiquity to modern specialists in morphology and syntax, with the aims and preoccupations common to linguists of our day. ‘Linguistics’ is a term that dates from the early nineteenth century, and the boundaries between the study of language and other disciplines, such as philosophy or the study of literature, have since been determined largely by the growth of faculties and departments, with their own curricula and their own examinations, in universities. In the period of the Roman empire the role of a grammarian, or in Latin a grammaticus, was self-contained in a quite di erent way. He was professionally a secondary teacher, who took pupils whose parents wanted and could pay for it beyond a stage of primary literacy. If members of the governing elite, they were destined ideally to play a part in public life, in a society that valued skill in oral presentation. From a grammarian’s care they might therefore pass to that of a teacher of rhetoric, who could take for granted that his pupils were literate; that they were able to understand and study literary texts; that they could assign the words of any text to successive ‘parts of speech’; that they could identify the cases of nouns or the tenses of verbs; that they understood in general what made utterances complete and, in a modern term, ‘grammatical’. Some of this belongs to what is now linguistics. To talk, however, of ancient grammar as part of the history of this subject is to project a modern concept onto an ancient discipline that only partly corresponded to it.
In looking beyond grammar we must be yet more cautious. We have already referred, however, to the philosophers of the Stoic school, whose interest in language was not the earliest. e history of linguistics, or a projection of what is now linguistics, has therefore been taken to begin, some centuries before the Romans conquered everything in their path, in a Greek world dominated intellectually by Athens.
A leading text is Plato’s Cratylus. It is a dialogue named like others a er one participant, who maintains that relations of forms to meanings are ‘by nature’ (phusei). In an opposite view, defended by another participant, they are valid merely ‘by custom’ (nomōi) or by convention. In modern eyes this second opinion is obviously right. e relation is not natural but ‘arbitrary’, in a sense that can be traced directly, through the Middle Ages, to a Greek word for ‘convention’ as it was subsequently used by Aristotle. What Plato himself concluded, in the mouth of Socrates as a third participant, is open to varying interpretations. Let us assume however, as was largely assumed by scholars throughout antiquity, that the view we now take to be obvious is wrong. Words are subject as we know to changes in, for example, sounds. It is not so long ago that these were called, quite neutrally, ‘corruptions’. ey can also be replaced by ‘borrowing’, as linguists have come to describe it, from other languages. ese represent disturbances, however, to what could be thought in principle to be an ideal system, in which the forms of words, as established before they were corrupted or replaced, directly re ect reality. An ideal system cannot, of course, be wholly recovered. But if this is right it is perfectly reasonable to ask, for example, why men should be referred to by a form man or, conversely, what sort of entity a form such as man can appropriately refer to. It is in this light that we must understand especially the ancient concept of ‘etymology’. In the period on which this book will
Introduction
focus, etymology was in practice separate from grammar and could be said to overlap it at the edges only. Both terms, however, have been used continuously for two millennia. In both elds, therefore, it is tempting to assume a continuity of ideas, in which the aims of ancient writers were basically those that we have also. For etymology in particular that is strikingly not so.
For a skeletal history and some ancient de nitions see Box 1.1. But the term in Greek, etumologia, was transparently a compound of -logia, as in modern ‘-(o)logy’, and an adjective, though not the one most usual, with the meaning ‘true’. What was ‘true’ then, and the subject of what was in antiquity an etumon, concerned the relation of an original form to an original sense, and the objective was to recover it as far as possible, by analysis and imaginative intuition, from the overlay of history. is involved in part establishing relations between forms; and in the simplest instances they were ones which we too, though with a di erent aim and by quite di erent criteria, will approve as valid. e name Cicero, for example, was and is from Latin cicer ‘chickpea’; the noun amor ‘love’ from the verb ‘to love’ (in nitive amare). e proper meaning of amor, and the right way for this word to be used, was in that way made clear. Less transparent relations called, however, for deeper insight, and those proposed in antiquity, if misinterpreted as ‘etymologies’ in the modern sense, will o en seem absurd.
Some characteristic illustrations are in Box 1.1. Let us imagine, however, that an ancient etymologist were to apply his insight to the study of English. To ‘cover’, for example, might be explained in his view as a reduction of to ‘conceal overall’; ‘grass’ could be so called because it ‘grows fast’; a ‘television’, if we may be just a little bolder, because it ‘tells things that are visible’. ese should not be seen as jokes. We are dealing with an earlier theory of the origin of words, rooted in earlier assumptions about the nature of language,
Box 1.1 Ancient etymology
e earliest serious treatment is in Plato’s Cratylus. e origin, for example, of the Greek word for a god (nominative singular theós) lay, as the character Socrates proposes, in the verb ‘to run’ (in nitive theîn). e reason, he explains, is that the earliest deities to be recognized were bodies like the sun and moon, which were constantly moving. In a bolder hypothesis, which is part of the same t of inspiration, the noun ánthrōpos (‘man’ in the general sense of ‘human being’) is explained as a contraction of anathrôn ha ópōpe ‘considering the things he has seen’, thus re ecting our ability to reason. How far Plato himself believed the ights of fancy that he put into the mouth of Socrates has been a central problem for the interpretation of the dialogue. e term etumologia dates, from fragmentary sources, to the centuries that followed, and for the Roman scholar Varro, in the rst century bc, it described a discipline then familiar. We do not have the chapters (traditionally the ‘books’), in which he discussed and defended it in theory. e topic as a whole, however, was ‘how words were applied to things in Latin’,1 and the parts of his work that survive are a classic illustration of the explanations that were thought to be illuminating. Many are again a product of imaginative speculation. Vīta ‘life’, for example, is from vīs ‘force, physical strength’: Varro cites in support a line from a Roman poet, which said that vita est (‘is’) vis. A blackbird is in Latin a merula because it does not form ocks and is thus, in a word used normally of wine, mera ‘unmixed’.2
A de nition in Greek, in a grammatical commentary of a much later period, describes the subject as ‘the unfolding of words, through which their true meanings (literally ‘that which is true’) is made plain’.3 For etumon ‘that which is true’ this
Introduction
de nition substitutes the usual word alēthes. It is as if, the commentator goes on to explain, one were to talk of alethologia. For Isidore of Seville, writing in Latin seven centuries a er Varro, a typical illustration, famous indeed in the history of the discipline, was the derivation of the word for ‘copse’ or ‘grove’ (nominative singular lūcus) from the verb for ‘to shine’ (in nitive lūcēre); a conventional explanation was that in such places, which were o en sacred, light shines minimally. e term etumologia is applied in general to ‘the origin of words, where the force of a verb or noun is brought together through interpretation’.4 ‘For when the etymology is known’, as Isidore explains a few sentences later, ‘every study of a thing is plainer’.5 ese words are cited from a section headed ‘etymology’, in a work which was in e ect an encyclopaedia of ancient knowledge. But the plural Etymologiae is also the title given to the whole. e meaning of a word was not a problem separate in principle from that of its origin. Both played a central role in any inquiry, and in any inquiry the proper use of words and the proper distinctions between them, was essential. We now excoriate what we call the ‘etymological fallacy’. But Varro or Isidore, for example, might have found it very hard to see it as such.
1. quemadmodum vocabula essent imposita rebus in lingua latina (LL 5.1, trans. Kent 1938: 2).
2. LL 2.63; 5.76.
3. hē anáptuxis tôn léxeōn, di’ hês tò alēthés saphēnízetai (GG 1.3: 14, ll. 23–4; trans. Robins 1990: 26).
4. origo vocabulorum, cum vis verbi vel nominis per interpretationem colligitur (Etym. 1.29).
5. omnis enim rei inspectio etymologia cognita planior est.
that linguists have been forced to reject. If a Greek or Roman scholar were to be resurrected, he would indeed have di culty in understanding our priorities. He might well conclude that ‘etymology’ has become a sadly jejune discipline. People practising it now are simply ignoring what should be the central issue, of how the forms that a society has created relate appropriately to the things to which they are assigned.
e example of etymology can serve as a warning for the study of ancient grammar. We may try to project modern distinctions, between what is now and what is now not part of ‘linguistics’, but we must not lose sight of the intellectual context in which ancient scholars were working or their own, o en tacit assumptions. For grammarians, moreover, part of the context was itself formed by the practice of etymology, and by ideas that lay behind it. It mattered in this light that things should be named appropriately. e term for a noun, for example, was in origin the ordinary word for someone’s name: the onoma of Plato, or in Latin his nomen, was the nominative singular Plátōn. A word, however, like ánthrōpos ‘human being’ was not a name and in the earliest accounts was not an onoma. It was instead distinguished as in Greek a prosēgoria (literally an ‘addressing’ or an ‘identi cation’). In their grammar, however, names and ‘identi cations’ were alike, and by the end of the Hellenistic period an onoma was in one analysis a part of an utterance, as in later grammars, that included both. But the name of an individual remained an onoma that was kurion, a principal or ‘leading’ name, and became in Latin a nomen proprium, or name in a strict sense distinguished from others. is is the source accordingly of ‘proper noun’ in English, in whatever way the term may now be understood. e terms we use now tend in contrast to be more opaque. ‘Noun’ in English has no origin other than as the equivalent in
Introduction
late Middle English of nomen as it was used in Latin grammar. ‘Tense’, for example, is a similar equivalent of Latin tempus, which corresponded to Greek khronos as the ordinary word for ‘time’. Any linguist writing nowadays will insist, of course, that tense and time are not the same thing. One is an in ectional category whose ‘basic role’, if I may cite a formulation of my own, ‘is to indicate the time of an event, etc. in relation to the moment of speaking’ (Matthews 2014: 403). e other is a dimension of reality as people perceive it. It does not follow from a de nition such as mine that every form described as in, for example, the past tense must always refer to events or situations that in time too will be past. e point can be made clear in English with such utterances as I was naturally seeing you tomorrow. But for a Roman grammarian tempus meant quite literally ‘time’, and ‘past time’ (tempus praeteritum) meant precisely that. A verb such as amāvī ‘I loved, have loved’ had as a word the property of being situated on a time scale.
Another important category is that of ‘person’. For us, this too is a grammatical category, ‘distinguishing’ if I may cite myself again, ‘speakers and addressees from each other and from other individuals etc. referred to’ (2014: 296). But the term in Greek, prosōpon, had the ordinary meaning of a ‘face’; also of a ‘character’, distinguished by a mask worn by a performer, in the theatre. It was therefore natural to extend it to the individuals engaged in or referred to in an act of speech. e person speaking, whose utterance is the centre of attention, was the ‘ rst’ prosōpon, or in Latin the ‘ rst’ persona; a person spoken to the ‘second’. Any other entity an utterance might refer to was a ‘third person’. e same terms then applied to forms by which participants, as we may now call them, were identi ed. In the sentence, for example, in English I told you the man was there the form I refers to the
Introduction
speaker and would itself as a form have had the property ‘ rst person’; you refers to someone addressed and would itself be ‘second person’; (the) man, which refers to someone else, would for an ancient grammarian have been as a form ‘third person’. e assumption behind this is one that also inspired ancient etymology: that language was a rational representation of reality, invented by human beings who are distinguished by their power of reason from all other living creatures. Categories of reality and of language were therefore in principle the same. is belief also informed ancient understanding of how words with properties like these combined meaningfully, as we will see, in utterances.
Transcriptions
is is perhaps as good a place as any to explain the conventions by which forms in Greek especially will be represented. Where they are cited as examples I will follow a conventional transcription of the Greek alphabet, accents included: thus ánthrōpos for ἄνθρωπος ‘human being’. e acute on the rst syllable of ánthrōpos represented what was historically a high pitch; the circum ex on words such as theîn ‘to run’ or tôn ‘of (plural) the’ represented a falling pitch over a syllable whose nucleus was a long vowel or diphthong; a third accent, the grave, was used in writing when a high pitch was lowered in context. Note that the length of vowels was distinguished in the Greek alphabet for mid vowels only. Note too that a γ will be transcribed by n before a velar consonant: thus ángelos for ἄγγελος ‘messenger’ (phonetically [aŋɡelos]).
I have been persuaded to use the same transcription where passages in Greek are cited in the notes to boxes. Where terms,
Transcriptions
however, are cited in the text I will follow a more usual convention and omit accents: thus etumologia for ἐτυμολογία.
In representing examples in Latin I will add a macron, following the usual tradition, where the length of a vowel determined that of a syllable: thus lūcēre ‘to shine’. In citing terms, and in the notes to boxes, macrons are not added.
