More Roman Grice

Quintilian, Dionysius, and Donatus: the start of a pedagogic tradition 

From philosophy to pedagogy 

The pedagogical progenitor to the Western Classical Tradition in linguistics arose in Alexandria on the west bank of the Nile delta in Egypt. Because Alexandria was inhabited by Egyptians, Greeks, and Jews, it has been suggested that the growth of grammatical interest there was the result of the Greek community living among speakers of other languages. This naturally enkindled an interest in a Greek which had to be not only learned by foreigners, but also preserved as the language of culture and the traditions of the people who identified with Greek culture. Hellenism (Hell?nismos) was perhaps a more powerful force on foreign soil than in Greece itself and it was inclusive: Isocrates (c. 436–338 BCE) had written, 

‘The people we call Greeks are those who have the same culture as ours, not the same blood’ (Panathenaicus, quoted in Marrou 1956: 130). 

Cultural pursuits were funded by the government. Around 250 BCE the great Library of Alexandria held 120,000 books (Marrou 1956: 260); scholars of literature and science were encouraged to work and teach in the adjoining Museum “seat of the Muses”1 – which today would be called a university. Instead of being interested in grammar as an adjunct to philosophy (like Plato, Aristotle, and the Stoics), the Alexandrians were interested in grammar as a tool for literary study; so they focused on that part of the grammatical tradition that came from Aristotle’s expositions of rhetoric and poetry, along with those of the Stoics. ‘The Stoics teach that logic is divided into two parts, rhetoric and dialectic’, wrote Diogenes Laertius 1925, 7: 41. The importance of rhetoric or oratory2 was championed by Plato’s contemporary Isocrates. Although many people could read, all copies of every book were handwritten; consequently, reading material was expensive and rare. The public lecture was the prime source of information; silent reading was virtually unknown: poetry was normally recited, dramas acted, songs sung. When there were no print or electronic media, oratory was at least as important in politics and law as it is today. As we shall see, this oral culture explains much about the format of the Alexandrian grammars. Plato had favoured parallel education for boys and girls (especially during their midteens) in riding, archery, javelin throwing, and other militaristic sports; in music, dancing, and singing (Laws VII: 804c–805b, 813b). He believed that geometry in particular and mathematics in general was a good basis for developing rational thought (Republic VII: 526a–527c) and thought them best studied during the late teens. Education began at about the age of ten with the study of grammar, that is, with the exegesis of literature. Perhaps 1. See Chapter 2, footnote 7, p.26. 2. A distinction can be made between rhetoric and oratory, but for our purposes they present the same needs. Quintilian, Dionysius, and Donatus: the start of a pedagogic tradition 81 

because Homer (ninth or eighth century BCE) 3 wrote of gods and heroes and history (mythical or real), there was a view that the poet’s function is to educate. Much of Homer, especially the Iliad, was learned by heart (Marrou 1956: 70).4 Other favoured writers were the tragedian Euripides (c. 484–406 BCE), the comedian Menander (c. 342–292 BCE), and the speech-writing orator Demosthenes (c. 384–322 BCE). It is possible that Hellenist educators were, in part, concerned with preserving the Greek language from ‘decadence’ (at every period of history some people have insisted that their language – whichever one they spoke – was being degraded by contemporary speakers). It was the duty of the grammarians to find out how Greek should be written and spoken, so as to fix it in the ‘correct’ form. Thus the notion arose that the function of a grammarian is to prescribe correct usage, rather than to describe observed usage as the philosophers and Varro had done. The difference is mostly a consequence of the purpose for which linguistic study is undertaken: someone teaching a second language or an older variety of the language is necessarily prescriptive; but the forms of language that are being taught ought nonetheless to be based upon the analyses of data collected by a descriptive linguist. Quintilian Some two generations after Varro died, Marcus Fabius Quintilianus (35–95 CE) was born. Quintilian headed the foremost school of oratory in Rome and composed the Institutio Oratoria “Institutes of oratory” (Quintilian 1920-22) around 88 CE. A manual for the public speaker, it contains some very commonsensical views on language, at least some of them traceable to Varro – although Quintilian castigates Varro, among others, for proposing absurd etymologies (ibid. I.vi.32–38). Quintilian regarded the study of grammar as a vehicle for developing knowledge of correct speech (recte loquendi scientia) and nurturing the ability to read and recite with understanding. Despite claiming ‘I did not set out to write a treatise on grammar’ (I.v.54), he nonetheless wrote in Book I about letters, syllables, words,5 parts of speech, correct ways of speaking, avoidance of barbarisms (faults of pronunciation), solecisms (grammatical errors), use of onomatopoeia, orthography, regularity (analogia), etymology, authority, and usage. 

The parts of speech he identifies (I.iv.18–20) are verbs, nouns, conjunctions (he prefers convinctiones to coniunctiones as a translation of sundesmoi), articles, prepositions, common nouns (appellationes) – which he says are sometimes subsumed to nouns (nomines),6 pronouns, participles, and adverbs. ‘Our 3. 

Although Homer is credited with being the author of the Iliad and the Odyssey (Homer 2003a; b), these epics may well have been created by a number of different rhapsodists – the people who recited the poems in public performance. 4. There are nearly 15,700 lines in the full version of the Iliad but early papyri often have at least 2,000 fewer; the Odyssey has more than 12,100 lines, though early papyri have fewer than half this number (Marrou 1956: 525). 5. This includes attention to prosodic features, e.g. ‘palus means a “stake” if the first syllable is long, a “marsh” if it be short’ (I.vii.3). P?lus, -i (M), vs palus, -udis (F). 6. As we shall see in the Techn? Grammatik? and Ars Grammatica, discussed later in this chapter. 82 The Western Classical Tradition in Linguistics own language dispenses with the articles, which are therefore distributed among other parts of speech. But interjections must be added to those already mentioned.’ Roman grammarians used the demonstrative in place of the Greek article, e.g. haec Musa “this Muse.NOM”, huius Musae “this Muse.GEN”, huic Musae, hanc Musam, etc. (Donatus 1961b: 355f; Marrou 1956: 371f). Quintilian holds the Greek language and classical Greek authors in very high esteem: 

‘For Latin is largely derived from that language, and we use words which are admittedly Greek to express things for which we have no Latin equivalent’ (I.v.58). Horace (65–8 BCE) had written Captive Greece took captive her savage conqueror and brought civilization to rustic Latium. (Letters II.i.156, Horace 1929) Roman education was the transfer to Latin of Greek education, and although Latin was the national language from the first century BCE, educated Romans were fluent in Greek; and in most Greek areas of the Roman Empire, Greek remained the lingua franca, as it was in Byzantium in the days of Priscian (sixth century; see Marrou 1956: 297ff). No other language within the Roman Empire enjoyed the status of Greek with respect to Latin. 

However, Quintilian had too much common sense to despise Latin, and he even proposed that the instrumental function of the Latin ablative should count as a seventh case, an idea subsequently ignored. Regularity (analogia) ‘cannot be universally applied, as it is often inconsistent with itself’ (Quintilian 1920-22, I.vi.12). 

For analogy was not sent down from heaven at the creation of mankind to frame the rules of language, but was discovered after they began to speak and to note the terminations of words used in speech. It is therefore based not on reason but on example, nor is it a law of language, but rather a practice which is observed, being in fact the offspring of usage. (I.vi.16) 

The importance that Quintilian gives to consuetudo “usage, custom” (i.e. the conventions of language use) strikes the twenty-first century linguist as very modern. 

Language is based on reason, antiquity, authority, and usage. Reason finds its chief support in analogy and sometimes in etymology. […] Authority as a rule we derive from orators and historians [but not poets …] Usage, however, is the surest pilot in speaking, and we should treat language as currency minted with the public stamp. (I.vi.1–3) 

Usage remains to be discussed. For it would be almost ridiculous to prefer the language of the past to that of the present day, and what is old language but the old manner of speaking? But even here judgment is necessary; we must make up our minds what we mean by usage. If it be defined merely as the practice of the majority, we shall have a very dangerous rule affecting not merely style in language but life as well, a far more serious matter. For where is so much good to be found that what is right should please the majority? [...] I will therefore define usage in speech as the agreed practice of the educated. (I.vi.43–45) Quintilian’s views were generally respected during the ages that followed, but the notion of correctness in language usage (or, whose dialect should be chosen as standard), though it was echoed by Sextus Empiricus (AM I: 176–240), did not really become an issue for discussion again until the seventeenth and eighteenth centuries (see Chapter 7). Like other educators, Quintilian believed that children ‘should study what is morally excellent’ (I.viii.4). Quintilian, Dionysius, and Donatus: the start of a pedagogic tradition 83 It is therefore an admirable practice which now prevails, to begin by reading Homer and Virgil, although the intelligence needs to be further developed for the full appreciation of their merits: but there is plenty of time for that, for the boy would read them more than once. In the meantime let his mind be lifted by the sublimity of heroic verse, inspired by the greatness of its theme and imbued by the loftiest sentiments. The reading of tragedy also is useful, and lyric poets will provide nourishment for the mind. (I.viii.5–6) In lecturing, the teacher of literature must give attention to minor points as well: he will ask his class after analysing a verse to give him the parts of speech and the peculiar features of the feet which it contains:7 these latter should be so familiar in poetry as to make their presence desired even in the prose of oratory. He will point out what words are barbarous, what improperly used, and what are contrary to the laws of language. (I.viii.13) Teachers should discuss metaphors and figures of speech and the style of presentation. Students should write aphorisms, moral essays (chriae) and discuss the character of personae (ethologiae) (I.ix). All these grammatical-cum-literary studies (grammatice) should precede the study of rhetoric and oratory (I.x.1, II.i.3–4). 

The latter includes, among other things, knowing when it is appropriate to substitute one expression for another. Instead of “I know,” we say “I am not ignorant,” or “the fact does not escape me,” or “I have not forgotten,” or “who does not know?” or “it can be doubted by none.” But we may also borrow from a word of cognate meaning. For “I understand” or “I feel” or “I see” are often equivalent to “I know.” Reading will provide us with a rich store of expressions such as these, and will enable us not merely to use them when they occur to us, but also in the appropriate manner. For they are not always interchangeable: for example, though I may be perfectly correct in saying “I see” for “I understand,” it does not follow that I can say “I understand” for “my eyes have seen.” (X.i.12–14) Quintilian warns against imitation in favour of new discovery, because ‘whatever is like another object must necessarily be inferior to the object of its imitation, just as the shadow is inferior to the substance’ (X.ii.4–11). Quintilian’s grammatical sketch is probably indebted to Quintus Rhemnius Palaemon (first century CE, Palaemon 2001), reportedly his teacher. It is Palaemon who is credited with writing the first Latin ars grammatica, introducing the interjection in place of the article among the parts of speech, and describing Latin declensions and conjugations – although Quintilian has no term for conjugate. 

Although Quintilian lived at least four generations after Dionysius Thrax, he makes no mention of him. Yet many of Quintilian’s recommendations are consistent with what Dionysius describes as the function of grammar; and the sequence of instruction seems to match. One can only conclude that the brevity of the Techn? Grammatik? and of Donatus’ De Partibus Orationis (the Ars Grammatica Minor) is what led to their wider use in pedagogy than Quintilian’s much more thorough – though at the same time less detailed – discussion in his Institutionis Oratoriae. 7. A practice taken up by Priscian and many of his followers; see Chapters 6 and 7. 84 The Western Classical Tradition in Linguistics The Techn? Grammatik? of Dionysius Thrax The Alexandrian tradition was established by Dionysius Thrax (c. 160/170–85/90 BCE) and more particularly by Apollonius Dyscolus (c. 80–160 CE). It was taken over into Latin by the Roman Aelius Donatus (315–85 CE), who adapted Dionysius’ work, and by Priscianus Caesariensis (490–560 CE), who adapted and built upon the work of Apollonius (see Chapter 6). 

In Europe the hegemony of the Latin language lasted until modern times, and throughout the centuries what was taught was founded on the grammars of Donatus and Priscian. Aristarchus of Samothrace (216–144 BCE) was one of the earliest Homeric scholars and has been credited with a number of developments in grammar, but nothing of his work remains. Aristarchus lived and lectured in Alexandria, where he taught Dionysius Thrax (socalled because his father supposedly came from Thrace, Thrákis).8 Dionysius became a Homeric scholar like his teacher, wrote some literary commentaries, and taught rhetoric in Rhodes and Rome. The work for which he is most celebrated is Techn? Grammatik? “The art of grammar” (hereafter TG), written in Rhodes about 100 BCE (Pfeiffer 1968: 271). Technai are handbooks or manuals (Marrou 1956: 126). TG runs to 25 sections in Anecdota Graeca (Bekker 1965: 627–43) and less than 3,000 words of Greek text. In Dionysius 1987, which we shall use for reference, it is a mere 20 sections.9 Although traditionally attributed to Dionysius, doubts about his authorship of the extant text have circulated since ancient times (Pinborg 1975: 103ff; Dionysius 1987: 169f; Law and Sluiter (eds) 1995). There are a number of bases for doubt. Everyone accepts that Dionysius is responsible for the first five sections of TG. The sceptic Sextus Empiricus in Adversus Mathematicos writes (in the late second century): Now Dionysius the Thracian says in his Manual [tois parangelmasi] that grammar is mainly expertness regarding the language of poets and composers [sungrapheusi], meaning by ‘composers’ (as is plain from its contrast with ‘poets’) none other than the writers in prose. 

For the grammarian appears to interpret the writings of the poets such as Homer, Hesiod, Pindar, Euripides, Menander and the rest; and he also takes it as his proper task to investigate those of the composers such as Herodotus and Thucydides and Plato. Accordingly, some grammarians have dealt with many of the composers, whether historians or orators or even philosophers, seeking to discover which of their writings are correctly and idiomatically expressed and which are faulty, and what for example is the meaning in Thucydides of zankloa [“sickle”] and torneutontes [“rounding off”]10, and in Demosthenes of ‘he shouted as though from a wagon’ [he shouted like a fishwife]; and how we should pronounce the word ?dos in Plato – whether the first syllable should be aspirated, or the first syllable unaspirated and the second aspirated, or whether both should be aspirated. It is because of such investigations that grammar has been called expertness regarding the language of poets and 8. He is known in French as Denys le Thrace “Dennis the Thracian”. 9. The translator is Alan Kemp who, following Uhlig (ed.) 1883 Grammatici Graeci I.i, includes Bekker’s §5 with §4; Bekker’s §13 with §12 (which become Kemp’s §11); Bekker’s §17 and §18 with §16 (Kemp’s §15); Bekker’s §22 with §21 (Kemp’s §17). 10. Not found in any extant work of Thucydides. Quintilian, Dionysius, and Donatus: the start of a pedagogic tradition 85 composers [empeiria t?n para poi?tais te kai sungrapheusi legomen?n]. (AM I: 57–60, Sextus Empiricus 1955) Sextus says (AM I: 72f) that such expertness is appropriate to conjectural (stochastik?) skills like navigation and medicine but not to grammar, which is a skill more akin to music and philosophy. Grammar only exists when being practised by the grammarian, ‘as walking is nothing apart from the walker’ (AM I: 74). Sextus favours the views of Crates, Tauriscus, and Asclepiades that grammar should deal with (1) the exegesis and criticism of language and grammatical tropes (figures of speech) in poets and prose-writers, including distinguishing genuine from spurious works; (2) the technical part, dealing with dialects, Hellenism, and the parts of speech; and (3) historical information about the persons and places written of (cf. AM I: 92f). And Dionysius Thrax, in asserting that there are six parts of grammar […] includes amongst them the historical; for he says that ‘the parts of grammar as skilled reading according to the scansion, explanation concerning the tropes which the poems contain, exposition of the phrases and histories, the discovery of etymologies, the reckoning of analogy, the judging of compositions.’ (AM I: 250) 

This is consistent with Dionysius’ reputation as a Homeric scholar. Dionysius’ first and foremost concern is the exegesis of poetical texts, especially Homer’s epics. He is a grammatikos, a scholar, someone active in the field of letters in a very wide sense. Linguistic studies as such are not his concern. (Schenkeveld 1995: 43f) Given what we have seen of the place of literary studies in education, and in the light of Sextus’ references to Dionysius, Schenkeveld’s comment seems justified: Dionysius did not set out with the same intention as a modern linguist would. Nevertheless, TG has come to be viewed as a linguistic work. Here’s another caveat: If the primary use of grammar was to help children to read literature, then we should expect to find more grammars of Homeric Ionic than anything else; since we do not, it seems unlikely that this was how grammatical texts were normally used. (Morgan 1995: 81) Morgan makes an interesting but misconceived point. For whatever reason, the ancient Greeks championed the Attic dialect of Athens as they pursued Hell?nismos; Ionic is a closely related dialect, spoken within 100 kilometres from Athens on many Aegean islands east of Attica and coastal Asia Minor. Classical Attic was also preferred in Hellenistic education to the common spoken version of it, koin?. TG begins with a statement of the purpose and function of grammar. §1 Grammar is the practical knowledge of the general usages of poets and prose writers. It has six parts: first, accurate reading aloud [anagn?sis] with due regard to the prosodic features [viz. accent, rough/smooth breathing, vowel length, phonotaxis]; second, explanation of the literary expressions in the works [ex?g?sis]; third, the provision of notes on phraseology [gl?ssai] and subject matter [historiai; e.g. on historical figures and events]; fourth, the discovery of etymologies; fifth, the working out of regular paradigms [analogi?s eklogismos]; sixth, the appreciation of literary compositions [krisis poi?mat?n], which is the noblest part of grammar. 

Consistent with what is known about Dionysius, there is emphasis on literary appreciation and the importance of studying good style. This will remain characteristic of the 86 The Western Classical Tradition in Linguistics pedagogical branch of the Western Classical Tradition in linguistics; indeed, we have already seen it in Quintilian’s Institutes. It is exactly the kind of study that Diogenes Laertius reported as characteristic of the Stoics when they were concerned with rhetoric and consequently with Hell?nismos. It builds upon the work of the Stoics, and it is consistent with what Varro did. So far then, there is continuity in the tradition as we have seen it develop. The difference is the abandonment of philosophy and a shift of focus to the preservation and perhaps restoration of literary classics of the Greek tradition. Hence Dionysius talks of grammar in terms of ‘practical knowledge’. Material was drawn from the written texts of selected authors whose usage justifies his descriptive statements. The same method was used in grammars for the next twenty centuries – with the drawback that constant reference was made to the descriptions and conclusions of earlier grammarians who may have been writing about a different language. We often find the English grammarians of the seventeenth to nineteenth centuries looking to the usage of Greek or Roman writers and arguing that that was what should be copied in the English of the day (see Chapter 7). The introduction, §1 of TG, sets out a much wider programme than is covered in the existing text, and the inconsistency suggests multiple authors. Pfeiffer 1968: 268–72 disputes the doubts of authenticity, while admitting that the text is incomplete and has been recombined by some early anonymous editor. It has been suggested, e.g. by Robins 1995, that TG may have changed as it was constantly and anonymously updated; earlier versions were lost and inconsistencies not resolved. This hypothesis is feasible. 

Nothing like copyright existed until after the invention of printing; authors did not necessarily insist on being identified; and revision while copying a text was far from uncommon. Robins demonstrates that Gray’s Anatomy of 1954 is in places vastly different from the original text of 1858, and that the different sources for Saussure’s Cours de linguistique générale also diverge significantly (see Chapter 11). Thus, even in modern times, different editions may show significant textual divergences. We may conclude that the extant text of TG comprises the work of Dionysius together with that of anonymous scholars who lived several centuries later. Instead of Dionysius being the precursor to Apollonius that Donatus would later be to Priscian, it is possible that the progressive portions of the extant text of TG date from the two centuries between Apollonius and Donatus. §2 of TG is on reading aloud, the significance of which in antiquity has been mentioned more than once. TG exhorts correct pronunciation and a delivery appropriate to the subject matter; for example, an elegy should be spoken in plaintive tones, an epic declaimed vigorously. Unless these rules are carefully observed, the true value of the poetry is lost, and the reader’s whole approach becomes subject to ridicule. §3 instructs how to read the accent diacritics ´, `, ^ or ˜. 

Greek accents (earlier referred to by Aristotle in Poetics 1456b 31, see p. 46 above) were originally pitch accents (tones) in Attic, Ionic, and Aeolic dialects. The rules are complex, and the following is an oversimplified sketch. Quintilian, Dionysius, and Donatus: the start of a pedagogic tradition 87 * A grave accent falls only on a final syllable and only when another word follows with no intervening punctuation. * An acute accent may only occur on the penultimate syllable when the final syllable is long: ???????. If it occurs on the antepenult (i.e. a syllable before the penultimate syllable), the final syllable must be short: ????????. * A so-called circumflex accent (often symbolized by ˜ ) is in fact a length marker; it falls only on a long vowel or diphthong: ?? , ?, ??, ?? , ?, e?˜, etc. It was Alexandrian scholars who invented these diacritics for marking the proper pronunciation of classical Attic Greek words. In addition to marking prosody, they must have aided word division and understanding in scripta continua. The Alexandrians also introduced the two breathing diacritics. The early Greek alphabets write H as a letter for [h], but this pronunciation was lost in Ionia; when Attic adopted the Ionian alphabet H was used for the front vowel eta, but that created written ambiguity between e.g. ? “or” and h? “which”; so the Alexandrians contrasted the smooth breathing diacritic ? “or” with the rough breathing diacritic ? “which” (see Householder in Apollonius Dyscolus 1981: 236). §4 is on how long to pause at the three punctuation marks: the high point ? marked 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 . (hypostigm?) was roughly counterpart to our comma. §5 tells that a ‘rhapsody’ is part of a poem which contains a theme of its own. Traditionally, the people called rhapsodists recited and also offered commentary on Homer’s epics (Pfeiffer 1968: 11). The aforementioned sections §§2–5 offer some detail on the first part of the practical study for the student of grammar promised in §1. 

Perhaps §§6–10 on letters and syllables do so too; or perhaps these sections should be classed with the rest of TG as ‘working out [the] regular paradigms’ – which is the fifth part of grammatical study identified in §1. However, verifiable direct quotations of §§6–20 are not found before the fifth century CE and so they may have been authored many years later than §§1–5. What is certain is that, of the parts of grammatical study described in §1, the second (exegesis), third (critical and historical commentary), fourth (etymology), and sixth (critical appreciation) are not dealt with anywhere in TG; it is even questionable that we have an account of the regular paradigms (part five), rather than merely an introduction to setting them out. To further the teaching of reading aloud, in §6, TG gives an account of the phonetic values of the 24 letters of the Greek alphabet. The letter (gramma or stoicheion) was, of course, regarded as having three parts: name, form, and pronunciation. Dionysius (for convenience we will assume he is the author11) identifies seven vowels – ph?n?enta, named thus because each is a ph?n? “sound” in its own right. This is recognition of the sonorant characteristic of vowels. He writes of vowel length and the construction of diphthongs; although as Sextus points out (AM I: 115–18, 169), the digraphs oi, ei, ou, and ai had long been monophthongs (respectively [y], [i], [u] and [e]). There are 17 sumph?na “withvowels” 11. This is comparable with assigning authorship of the Iliad and Odyssey to Homer. 88 The Western Classical Tradition in Linguistics like the Latin loan translation consonant (borrowed into English). 

Consonants need an accompanying vowel in order to be pronounced – is pronounced /ti/ not /t/. Consonants divide into h?miph?na “semivowels”, in today’s terms sibilants and sonorants which can readily be pronounced in isolation, and aph?na “mutes”, which include stops and the dasea “rough” aspirated voiceless stops ? /th /, ? /ph / and ? /kh /, which by the second century CE (if not earlier) had become fricatives /?/ , /f/ , and /x/ or . They are phonotactic counterparts to the voiceless stops, called psila “smooth”, when these are immediately followed by an aspirate: e.g. *eip hop? becomes eiph hop?, *autik ho becomes autich ho, etc. What today we call voiced stops were called mesa “middle”. The letter ? is said to consist of ?+? – presumably a voiced dental affricate. Aristotle had claimed this to be a matter of dispute in Metaphysics 993a 5 and perhaps was suggesting that takes the sibilance of and the voicing of , which accords with Sextus giving it the phonetic value [z] (AM I: 169, 174). The letter ? is composed of ?+? , ? of ?+? . The liquids (hugra) l, m, n, r do not change in verb or noun inflections and are therefore called invariable. Five letters occur at the end of masculine nominative singulars (-n, -x, -r, -s, -ps), eight in feminine nouns (-a, -?, -?, -n, -ks, -r, -s, -ps), six or seven in neuter nouns (-a, -i, -n, -r, -s, -u and perhaps -o); -a, -e, -? occur in the dual; and -i, -s, -a, -? in the plural.12 So TG combines phonology with morphology here. §§7–10 are on syllables. A syllable (sullab?) is defined in §7 as being ‘a combination of consonants with one or more vowels’ or a vowel on its own. There are eight kinds of long syllable (§8) determined by vowel length, a diphthong, or more than one consonant. Short syllables (§9) are defined on short vowels. There is also a ‘common syllable’ (§10), which is a syllable joining with the following syllable to form a single syllable (a sort of sandhi13 rule). In §11 the word (lexis) is defined as ‘the smallest part of a properly constructed sentence’. 

This is similar to Aristotle and not so thorough as the definition left by the Stoics. 

The sentence (logos) is defined as a combination of words in prose [and poetry14] that makes complete sense in itself. There are eight parts to the sentence: noun, verb, participle, article, pronoun, preposition, adverb, conjunction. Stoic influences persisted in Greek speaking areas well into the middle ages. According to Apollonius and various scholiasts,15 Dionysius was much more of a Stoic than the present text suggests. He reputedly distinguished onoma from pros?goria as distinct parts of speech, whereas TG has the common noun (pros?goria) as a subcategory of the noun (onoma). Dionysius defined the verb as denoting a predicate (lexis kategorian s?mainousa), whereas 12. Ancient Greek had a three-term number system: singular “one”, dual “two”, and plural “more than two”. 13. Sandhi is the assimilation of word-final or stem-final phones to the initial phone of the following word or suffix. 14. 

This does not appear in the older manuscripts. 15. See Uhlig (ed.) 1883 I.iii:124 lines 8–14; 160 lines 27f; 161 lines 6–8; 356 lines 12f; II.i:5 lines 18f; Robins 1995; Schenkeveld 1995. Quintilian, Dionysius, and Donatus: the start of a pedagogic tradition 89 TG gives it a formal morphosyntactic description (see discussion of §13 below). And like the Stoics, Dionysius classified personal pronouns as arthra deiktika, 16 which, again, is not verified in the existing text of TG. Dionysius was probably influenced by both peripatetic (Aristotelian) and Stoic schools, and may have adopted opinions consistent with the different schools in different books written at different times. Certainly the Stoic distinction between mer? lexe?s “parts of expression” and mer? logou “parts of a proposition” is not maintained by the Alexandrians, who incorporated these logical or semantic categories into a more formally oriented grammar. The eight parts of speech (i.e. constituents of a phrase, clause, sentence or utterance) in TG are: noun, verb, participle, article, pronoun, preposition, adverb, conjunction. The sequence is reckoned to be ‘natural’ on the following basis: nouns are first because substance is most basic and necessarily prior to any action, undergoing, becoming, or attribute; the verb is second because it predicates the noun; the participle is a combination of verb and noun; the article and pronoun are both dependent on the noun (the article precedes the noun in a phrase, and that may be why it precedes the pronoun in this listing of parts of speech); the preposition (covering both what we call prepositions and also prefixes) precedes both nouns and verbs; the adverb is dependent on a verb; the conjunction conjoins any of the preceding parts of speech. These parts of speech were retained – normally in exactly this sequence – throughout the Western Classical Tradition; except that, because Latin has no article, this part of speech was replaced in Latin grammars by the interjection (as we saw when discussing Quintilian I.iv.19–20). 

Both article and interjection were recognized to be necessary for grammars of ‘modern’ languages such as English, French, and German. Each of the parts of speech is subcategorized according to various parepomena “consequential attributes, subcategories”. §12, the longest in TG, is on the noun and its subcategories. A noun [onoma] is the part of speech inflected for case, denoting a body [s?ma] or thing/event [pragma]. This combines a formal with a semantic description in the Stoic manner. There are five subcategories of the noun, all of which have further subclasses. To the traditional three genders are added ‘common’ (for nouns like horse which may be either masculine ho hippos “the.M stallion”, h? hippos “the.F mare”) and ‘epicene’ (for nouns like aetos “eagle” of which the grammatical masculine denotes a bird of either sex). Morphological analysis starts with complete word and isolates prefixes and suffixes; there is no notion of a morphological root; a proper treatment of morphology had to wait for Priscian (Law 1995a).17 There are two types (eidos) of nouns: primitive (pr?totupon) e.g. g? “earth” and derived (parag?gon) e.g. gai?ïos “sprung from the Earth”. There are seven kinds of derived nouns: patronymics, possessives, comparatives, superlatives, diminutives (hypocorisms), denominal (e.g. the name The?n from theos “god”), and deverbal (nouns derived from verbs). Some nouns are simple in form (sch?ma), others are compounded from either complete or incomplete words. Some are super-compounds, having a derivational affix on a 16. Arthra is the plural of arthron. 17 .

 On practical exercises in morphology for ancient Greek students see Marrou 1956: 238, 242. 90 The Western Classical Tradition in Linguistics compound: simple (haploun) Memn?n, compound (suntheton) Agamemn?n, supercompound (parasuntheton) Agamemnonid?s. There are three numbers: singular, dual, plural; some singulars are collectives, applied to plural denotata such as d?mos “people”; some plurals apply to singular denotata such as Ath?nai “Athens”, some to duals, e.g. amphoteroi “both”. Nouns have five cases: nominative (orth? “upright”, onomastik? “nominative”, eutheia “direct”), genitive (genik? “generic”, kt?tik? “possessive”), dative (epistaltik? “epistolary”), accusative (kat aitian “causal”), vocative or salutatory. Notice how the typical functions of the cases are named as an aid to identification and memory. Finally, 24 species of nouns are identified including proper, common, collective, interrogative, and onomatopoeic. One of these types, epitheton “attached noun” (the third mentioned), is what today we call the adjective ‘placed next to either proper or common nouns, and conveys praise or blame’. At the end of this section §12, Dionysius divides the noun into active (e.g. krit?s “judge, one who judges”) and passive (pathos), the latter being adjectival (e.g. kritos “judged, one who is judged”). §13 is on the verb, and here we see the influence of both Aristotle and the Stoic grammarians, though a verb’s predicative function is not mentioned. The verb [rh?ma] is a word without case inflection, indicating tense, person and number, and showing activity or passivity. 

The verb has eight subcategories [parepetai]: mood, voice, type [eidos], form [sch?ma], number, person, tense, and conjugation [suzugia]. The five moods identified are the ones retained in the Western Classical Tradition: indicative (which includes both declarative and interrogative clause types), imperative, optative, subjunctive, and infinitive.18 The three voices are those identified by the Stoics: active, passive, and middle (focusing on an action of benefit to the subject, e.g. middle elousam?n “I bathed [myself]” vs. active elousa “I bathed [someone else]”). The next three subcategories parallel similar ones for nouns: primitive and derived types of verbs; simple, compound, and super-compound forms of verbs; and singular, plural, and dual number. There are three persons: first, second and third. The first is the source of the utterance, the second is the one to whom it is addressed, and the third is the person spoken about. This is very similar to what Varro wrote in De Lingua Latina, though it supposedly predates it by about 55 years: The persons of the verb are also of three kinds: the one who is speaking, the one addressed, and who or what is spoken about. Every verb has three persons, which [to some extent] explains why there are so many verb forms. (Varro 1938, VIII: 20) There are said to be three tenses, but four kinds of past tense: imperfect, perfect, pluperfect, and aorist. 

There are three close relationships between these tenses, namely between present and imperfect, perfect and pluperfect, aorist and future. This is something of a misunderstanding of the Stoic account we saw earlier in Chapter 4 Table 4.1. But in TG, the perfect is aligned with the past tenses, creating a lack of system. 18. The optative typically expresses counterfactuality and remote possibility, while the subjunctive expresses nonfactive and less-remote possibilities. Allan 2006a argues against counting the infinitive a mood, but it is certainly a clause-type. Quintilian, Dionysius, and Donatus: the start of a pedagogic tradition 91 The loss of the completive aspect to the category of past tense persisted in the Western Classical Tradition through to the twentieth century. Aspect is not labelled as a separate category, though it is seemingly recognized in the words h?n sungeneiai eisi treis “the relationships of which are three”. 

However, the relationship envisaged is formal rather than functional: the pluperfect egegraphei is built on the reduplicated stem of the perfect gegraphe; the imperfect egraphe is built on the stem of the present graphei; the aorist egrapse and the future grapsei each have an -s- stem (which is probably not an etymologically validated connection). On this interpretation, aspect is ignored in TG. §14 is on the conjugations (suzugiai) of ancient Greek:19 ‘Conjugation is a type of inflection belonging to verbs.’ The different conjugations are broadly classified according to those with no stress on the final syllable; those with stress on the final syllable; those ending in –mi. There are many finer subclassifications. Whether the model for Varro, or inspired by him, TG identifies the participle (metoch?) as a distinct category, defined in §15 as a word partaking of the characteristics of both nouns and verbs. It has all the subcategories of nouns and those of verbs, except for mood and person. §16 is on the article (arthron) – no longer referring to Aristotle’s range of connectives. An article is a part of speech which takes case inflections and may precede or follow a noun. When preposed it takes the form ho, when postposed [it is a relative pronoun with the form] hos. Its subcategories comprise three genders (M, F, N), three numbers (SG, DU, PL), and five cases (NOM, GEN, DAT, ACC, VOC). Note the formal and distributional description. A pronoun (ant?numia), §17, ‘is a word which substitutes for a noun and indicates definite persons.’ It has the same five subcategories as a noun and, in addition, person. A preposition (prothesis), §18, ‘is a word preposed to other parts of a sentence in compound forms and in syntax.’ TG included with prepositions what we would today call prefixes, such as sun– and kata– as in kataphron?. Dionysius listed 18 prepositions. The adverb (epirrh?ma), §19, ‘is a part of the sentence without inflection [cf. Varro], in modification of or in addition to a verb.’ 

This combines a morphological with a syntactico-semantic description. The Latin adverbus (whence English adverb) is a loan translation of epirrh?ma. TG proceeds to identify 26 kinds of adverbs (Dionysius 1987: 183–85). They include interjections like babai “Good heavens!”, euhoi “Hallelujah”, exhortatives eia, age, phere “Come”, prohibitives m?, m?dam?s “Don’t!” and oaths ma “No, [by god]”, n? “Yes, [by god]”. Finally, in §20, the conjunction (sundesmos) ‘is a word that binds the discourse together, giving it order, and fills gaps in its interpretation.’ This functional description is supported by the exemplification of nine kinds of conjunctions, disjunctions, adversatives and discourse markers. Padley 1976: 256 seems wrong to have judged this definition inadequate compared with the others in TG. 19. Kemp’s decision to follow Uhlig (ed.) 1883 and include these sections as further specification of the verb seems appropriate. 92 The Western Classical Tradition in Linguistics How does TG compare with what we have seen of the Stoics, Varro, and Quintilian? Most significantly, there is no theoretical underpinning for the linguistic categories offered in TG. By contrast the Stoic theory of the lekton underlies at least some aspects of their linguistic analysis; and Varro’s theory of language permeates De Lingua Latina. Quintilian’s approach is closer to Dionysius in that he is primarily focused on the proper materials for teaching literary exegesis and literary appreciation as preparation for instruction in rhetoric and oratory. Quintilian presents reasoned arguments for the topics he proposes for student instruction but these are almost entirely lacking from TG. On the other hand, TG is more of a teaching manual than the Institutes of Oratory because TG presents a more detailed description of the language being studied. Although the TG definition for verbs and participles uses the defining parameters of case and tense in a manner similar to that of Varro, TG lacks the systematic and principled basis of Varro’s linguistic categorization in which these parameters are used across the board. 

There is less breadth in Quintilian than in Varro, but more than in TG. Whether it is because Dionysius expressed no theory of the lekton, or because there is no syntax or logic within TG, the Stoic notions of subject and predicate as functional sentence constituents are effectively lost in favour of lexical classes. TG improves upon the Stoic noun subcategories insofar as we know them; but the same cannot be said of the verb subcategories. Aspect disappears as an independent (sub)category; and there is nothing on illocutionary type, only mood. Both aspect and illocutionary type are ignored by Quintilian and, with the exception of the insight of Apollonius Dyscolus into illocutionary force (see Chapter 6), they remain out of the Western Classical Tradition until the twentieth century. Lastly, the lexical relations and notions of derivational and inflectional morphology seen in Varro, and to a lesser extent in Quintilian (who mixes diachronic with synchronic derivation), are not to be found in TG. Perhaps mercifully, there are no etymologies in TG; but etymologies were intended to teach semantic parsing, and one would expect a pedagogical manual to include such examples as luchnos “lamp” from luein to nuchos “to blot out the night”; proskephalaion “pillow” from pros and kephal? “that which is placed under the head”. Quintilian’s sceptical discussion of etymology would surely have been a useful pedagogical aid. The major omission from TG is any statement of Greek syntax, although the word class system and the morphological analysis did form the basis for later syntactic statements. Varro’s work on syntax is lost, and Quintilian had as little to say about syntax as Dionysius. Practical exercises in morphology are first found in the third century CE, e.g. writing out a verb in all its voices, moods, tenses, persons, and numbers or a sentence in all possible forms, cases, number, persons, etc. [TG] summed up in a concise form the results of the past and became a school book in the future, suffering the corruptions and alterations unavoidable in this sort of literature. The brief and abrupt sentences in a staccato style20 called forth copious explanatory notes through the centuries. (Pfeiffer 1968: 268) TG makes no contribution to the potential development of theoretical linguistics as the philosophers and Varro did. It was translated into Armenian and Syriac very early on and 20. 

A similar style was later adopted by Donatus and it had a similar effect, as we shall see. 

Quintilian, Dionysius, and Donatus: the start of a pedagogic tradition 93 much discussed and commented upon in the middle ages. TG is a source for about 70 metalinguistic terms such as hypotaxis, diphthong, prototype, patronymic, anaphoric, deictic, neuter, dual, participle, conjugation, among others that remain in the Western Classical Tradition. TG functions very well as a school Greek primer. Its provenance as a pedagogical tool is all the more obvious in Donatus’ Latin version of it, as we see in the next section. The Artes Grammaticae of Aelius Donatus Latin took over from Greek as the language of European scholarship. Vulgar Latin was the colloquial spoken lingua franca; but classical Latin was the ideal for scholarly writing. Because it was nobody’s first language, classical Latin needed to be taught, and the most influential teaching texts were the De Partibus Orationis (Ars Grammatica Minor) “On parts of speech (Short grammar21)” and the Ars Grammatica Maior “Major grammar” of Donatus, and the Institutiones Grammaticae of Priscian (see Chapter 6). Donatus’ De Partibus Orationis is a condensed, explicitly pedagogical version of the Ars Grammatica (Major22) that is restricted to defining the eight parts of speech. Aelius Donatus (c. 315–85 CE) was obviously very strongly influenced by the Techn? Grammatik?, because the Artes Grammaticae are very similar to it in construction and presentation. 

The parts of speech identified are exactly similar to those of TG, the only difference being that, following Palaemon and Quintilian, the article is omitted and the interjection introduced; thus the number of parts of speech remains eight. 

Like TG, Donatus’ Artes omit any significant discussion of syntax. Donatus wrote principally (Law 1997: 58, 130) for people who already spoke Vulgar Latin, contemporary common koin? Latin which shows many features that were to emerge in Romance languages, but which are not found in classical Latin. 

In the Ars Maior (Donatus 1961a) Donatus defines vox “speech” as aer ictus sensibilis auditu “an audible puff of air”. 

He says that all speech is either anomalous or meaningfully articulated (articulata) and can be written (est quae litteris conprehendi potest, Donatus 1961a: 367). 

This seems to recognize the difference between utterance as a physical event that is used not only to carry meaningful sentences in a language, but also babbling and nonsense noises (even a dog’s bark could be said to be uttered – though it is unlikely that Donatus had any such thing in mind). We recognize utterance through brute perception; we understand meaningful sentences (and other linguistic fragments) via cognitive processes. 

These facts seem to be what Donatus is indicating. He writes: ‘the letter is the smallest part of articulated speech’ (Donatus 1961a: 367). Perhaps this identifies the most significant part of today’s definition of the phoneme; nevertheless, it is inferior to Aristotle’s description of the letter in Poetics 1456b 22 and 21. Ars grammatica is more literally rendered “the art of grammar” or even “the art of letters”. 22. Latin major is also spelled maior and usually pronounced with a medial semivowel [j], i.e. not with the alveolar-palatal affricate [?] used in English major. 94 The Western Classical Tradition in Linguistics furthermore it is an inaccurate description of a letter such as , phonetically [ks]. Donatus also discusses syllables, feet, tones and junctures, punctuation, barbarisms, solecisms, and other faults such as tautologies, verbosity, inappropriacy, and abbreviation. 

Finally, he names and illustrates several kinds of tropes and figures of speech. The Ars Maior also contains a more extensive discussion of the parts of speech than is to be found in the Ars Minor: for instance, noun and verb are said to be the principal parts of speech; the Latins count the interjection in place of the Greek article; and three parts of speech – noun, pronoun and participle – take six case inflections (Donatus 1961a: 372). In De Partibus Orationis (the Ars Minor) Donatus starts right in with a robust and vivacious pedagogic style: Partes orationis quot sunt? Octo. Quae? Nomen pronomen verbum adverbium participium coniuncto praepositio interiectio. DE NOMINE Nomen quid est? Pars orationis cum casu corpus aut rem proprie communiterve significans. Nomini quot accidunt? Sex. Quae? ... “How many parts of speech are there? Eight. What are they? Noun, pronoun, verb, adverb, participle, conjunction, preposition, interjection. CONCERNING THE NOUN What is a noun? 

A part of speech inflected for case which denotes a person or thing either specifically or generally. How many subcategories does a noun have? Six. What? ...” (Donatus 1961b: 355; Salus 1969: 92) One can well imagine generations of pupils repeating the answers to these questions as they struggled to learn how to parse Latin. Each of the eight parts of speech is introduced in the same manner, and a surprising quantity of information is packed into the roughly 3,650 words of the Ars Minor. The limitation is that only one example is given for each point made. For instance: How many degrees of comparison are there? Three. What? Positive as learned; comparative as more learned; superlative as most learned. What nouns are compared? Only common nouns signifying quality or quantity. What case is the comparative degree used with? The ablative without a preposition, for we say doctior illo “more learned than he”. What case with the superlative? Only the genitive plural, for we say doctissimus poetarum “the most learned of the poets”. (ibid.) Donatus’ definitions of the parts of speech are as follows. 

A noun is a part of speech inflected for case denoting a proper name such as Rome, Tiber, or a common object such as town, river. (Donatus 1961a: 373) This is identical in style with the formal and semantic description of Techn? Grammatik?. Six subcategories of noun are identified. Qualities may be proper or common: the latter included adjectives on the basis of the formal characteristics (number, gender and case) shared with nouns. Comparison was applicable only to what today we call the adjective; but Donatus did not identify this restriction. 

The five genders of TG are also mentioned in Ars Grammatica: M, F, N, common and epicene. 

Only two numbers are listed, singular and plural, no dual. Nouns are typed simple or compound, with four subtypes of compounding identified. There are the same six cases as in Varro and Quintilian (nominative, genitive, Quintilian, Dionysius, and Donatus: the start of a pedagogic tradition 95 dative, accusative, vocative, ablative). Donatus gives examples of each subcategory as illustrated in the following quote. Magister [“master, teacher”] is a common noun of masculine gender, singular number, simple form, nominative and vocative case, which will be declined thus: in the nominative, hic magister [“this master”]; in the genitive huius magistri; ... (Donatus 1961b: 355f; Salus 1969: 93) He continues through the declensions in singular and plural for all genders, e.g. M.NOM.PL hi magistri, GEN horum magistorum, DAT his magistris, ACC hos magistros, VOC o magistri, ABL his magistris; F.NOM.SG haec Musa “this Muse”, huius Musae, huic Musae, hanc Musam, o Musa, his Musis; etc. The example given for common gender is M or F sacerdos “priest(ess)”; that for epicene gender is the adjective felix “happy”, which may be M, F, or N. Regularities of various kinds are demonstrated, e.g. Those nouns which have the ablative singular in a or o make the genitive plural end in what? In rum, the dative and ablative plural in is. (Donatus 1961b: 356; Salus 1969: 94) For example, puell?~puell?rum~puell?s; domin?~domin?rum~domin?s. A pronoun is a part of speech that is often used in place of the noun to convey the same meaning and may refer to a person previously mentioned. (Donatus 1961a: 379) Pronouns have six subcategories: quality, gender, number, form, person, case. The qualities are definite (personal pronouns) and indefinite (relative and interrogative pronouns). There are three persons and only four genders (M, F, N, common). In other respects the subcategories are as for nouns. 

As for verbs: A verb is a part of speech with time and person and without case; it denotes doing something [active transitive], or undergoing something [passive], or neither [i.e. is intransitive]. (p. 381) There are seven subcategories of the verb: quality, conjugation (coniugatio), voice (genus), number, form, tense, and person. Qualities include the moods (modi): indicative (which includes declarative and interrogative), imperative, optative, subjunctive, infinitive, and impersonal. There are the four forms (formae) among qualities: undefined (lego “I read”), desiderative (lecturio “want to read”), frequentative (lectito “read often”), and inchoative (calesco “grow warm”); these are aktionsarten or aspects conditioned by the interaction of the verb form with its basic meaning, like the English verbs want, keep [doing], begin [doing]). There are three conjugations, formally defined, e.g.: What is the third? That which in the indicative mood, present tense, singular number, second person, active and neuter (intransitive) verb, has a short i or a long i before the last letter; in the passive, common, and deponent, in place of i, short e or long i before the last syllable as with lego~leg?s, legor~leg?ris [“read”], audio~aud?s, audior~aud?ris [“hear”]; and the future tense of the same mood ends in am and in ar as with lego~legam, legor~legar, audio~audiam, audior~audiar. It can be seen immediately in the imperative and in the infinitive whether the letter i is short or long. For short i is turned into e; if it has been made long it is not changed. When does the third conjugation end the future tense not in am only but also in bo? Occasionally when it has had the letter i not shortened but lengthened as with eo~?s~?b? [“go”], queo~qu?s~qu?b? [“be able”]. (Donatus 1961b: 359; Salus 1969: 97) 96 The Western Classical Tradition in Linguistics A possible reason for the discussion of the future is that in Vulgar Latin it was often periphrastic, e.g. instead of legam “I.will.read” it was legere habeo “to.read I.have” (Matthews 1994: 71). The five kinds of verb seem to include aspects of voice and transitivity: active, passive, intransitive (neuter – e.g. for active sto “I stand” there is no passive *stor “I am stood” in Latin), deponent (passive form, active meaning), common (e.g. deponent osculor te “I kiss you” and osculor a te “I am kissed by you”). There are two numbers (SG, PL). There are simple or compound forms. And three tenses, of which the preterite divides into imperfect, perfect, and pluperfect (as in TG). 

Other than the aktionsarten identified in the formae, aspect is ignored – perhaps because it was judged semantically irrelevant. There are three persons. The given sequence of subcategories seems arbitrary today, but was probably thought natural by Donatus. ‘An adverb is a part of speech which being added to the verb explains or completes its meaning’ (Donatus 1961a: 385). Adverbs have three degrees of comparison: positive (docte “knowledgably”), comparative (doctius “more knowledgably”), superlative (doctissime “most knowledgably”). Again, there are simple and compound forms. Donatus identifies 24 types (significationes) of adverb which are similar to those in TG, excepting interjections. ‘A participle is a part of speech which is part noun and part verb’ (p. 387). With the noun it shares gender and case; and with the verb time and denotation. The six subcategories of participle are therefore: four genders, six cases, three tenses, five voices, two numbers, and two forms. ‘A conjunction is a part of speech joining and ordering [the constituents of] a sentence’ (p. 388). There are five functions (potestas) of a conjunction: copulative (e.g. et, que “and”), disjunctive (e.g. aut, vel “or”), completive (expletivae, e.g. quidem “truly”, videlicet “obviously”, autem “moreover”), reason (causales, e.g. si “if ”, quando “when”, quidem “for certain”, seu “whether”), and logical (rationales, e.g. itaque “thus”, enim “for”, ergo “therefore”). A conjunction has either a simple or a compound form. It may be classified according to its relative position: preposed, postposed, or common (i.e. placed between the conjoined constituents). ‘A preposition is a part of speech which is put before other parts of speech to change, augment, complete or diminish their meaning’ (p. 389). 

The only subcategory of a preposition is case. Some prepositions are bound forms; i.e. in today’s terms they are prefixes. Palaemon reportedly defined the interjection as having no referential but only emotive or expressive meaning. Donatus writes much the same: An interjection is a part of speech interjected between other parts of speech as an expression of an affected mind such as fear, hope, grief, or joy. (p. 391) Interiectioni quid accidit? Tantum significatio “What subcategories has an interjection? Only its meaning.” The similarity of Donatus’ Ars Grammatica to Techn? Grammatik? is obvious. The only difference would be that the Ars Minor (De Partibus Orationis) is more explicitly pedagogical. As with TG, Ars Grammatica presents no theory of language structure, and there are many similarities in the definitions for parts of speech as well as in the format in Quintilian, Dionysius, and Donatus: the start of a pedagogic tradition 97 which they are presented. 

Donatus was fortunate that the categories induced for the analysis of Greek language structure transfer comfortably for use with Latin. Where there are obvious differences between the languages as with the additional Latin case and the lack of an article in Latin, the difference is readily accommodated. Dubious is Donatus’ assumption of the distinction which existed in classical Greek between the optative mood expressing counterfactuality and remote possibility and the subjunctive expressing nonfactive and lessremote possibilities, a distinction which did not operate in Latin (Lyons 1977: 815ff). The transference of grammatical categories from the analysis of one language to that of another was a feature of linguistic analysis until the twentieth century. The parts of speech in the pedagogical tradition It may be that the extant text of Techn? Grammatik? is almost contemporary with the Ars Grammatica of Donatus rather than nearly 500 years older. We assume nonetheless that the Latin text is inspired by the Greek rather than vice versa. Both are principally pedagogical texts; both comprehensively describe the parts of speech in their respective languages; both omit an account of syntax. Techn? Grammatik? begins with a general statement of the purpose and function of grammar; Ars Grammatica has nothing comparable. Donatus’ Ars Maior distinguishes the physical event of utterance from the cognitive content of language used in making a meaningful utterance; Techn? Grammatik? has nothing comparable. Dionysius spends time discussing what is needed for reading aloud, expounding pronunciation, letters, and syllables; the Ars Maior has sections on speech (vox), letters, syllables, feet, the three accents, and junctures (positurae); but these are lacking from the Ars Minor. As in Aristotle (and the Western Classical Tradition through to the late nineteenth century) the translation equivalents of the term letter were used of the name, grapheme, and (typical) pronunciation, e.g. pee,

or

, and /p/. Whereas Dionysius discusses the pronunciation of letters, the Ars Maior has the general statement that the letter is the smallest part of articulated speech. Dionysius defines both word and sentence, but Donatus does not. The definitions in Techn? Grammatik? are: * A word is the smallest part of a properly constructed sentence. * A sentence is a combination of words in prose and poetry that makes complete sense in itself (expresses a complete idea). Both these definitions continued through the Western Classical Tradition. Both Dionysius and Donatus recognized that there are eight parts of speech, i.e. constituents of a word (in the case of ‘preposition’), phrase, clause, sentence or utterance: noun, verb, participle, article for Dionysius or interjection for Donatus, pronoun, preposition, adverb, and conjunction. The Western Classical Tradition adopted all nine, eventually distinguishing adjective from noun adjective in the twentieth century. The definitions were essentially similar and were retained in the Tradition. * 

A noun is inflected for case and denotes a proper name or a common name for person, thing or event. 98 The Western Classical Tradition in Linguistics Other subcategories of the noun were also identified and although specific to either Greek or Latin are similar because the two languages are related. The notable difference is that Latin has an additional case, the ablative, often referred to as the Latin case.23 When the definition of noun was later applied to languages which do not have explicit case morphology (such as Chinese or modern English), the formal part of the definition was simply omitted, leaving only the semantic description. * 

A verb is without case, but indicates time (or tense), person, and number; it denotes activity or passivity (is transitive or intransitive). Notice that the definition is primarily formal and syntactic. The verb subcategories of transitivity and voice were more precisely identified in Dionysius and Donatus than in the verb definition just given above; also mood and tense were identified and the optative mood inappropriately applied to Latin (where it is identical with the subjunctive). Aspect was ignored and included along with tense – where it remained in the Western Classical Tradition until the twentieth century. * A participle partakes of the characteristics of both nouns and verbs: from the noun it has gender and case, and from the verb time and denotation. Essentially the participle is given the same definition in both Dionysius and Donatus. Notice the balance of formal and semantic definition. * An article is a part of speech that takes case inflections and may precede or follow a noun. This definition is found only in Dionysius because classical Latin did not have an article, although during Donatus’ lifetime Vulgar Latin was adopting the articles that we see in Romance languages today. On formal and syntactic (distributional) grounds, the article includes the relative pronoun – an application that quickly dropped from the Tradition under influence from Latin. * An interjection is located between other parts of speech as an expression of the affected mind, such as fear, hope, grief, or joy. Although Ancient Greek did have interjections, they were not recognized as a distinct part of speech, but were included under adverbs. Perhaps they were distinguished in Latin grammar because of the Roman love of oratory; but such a hypothesis ignores the fact that the Greeks also studied and valued rhetoric. 

The inclusion of the interjection as a distinct part of speech does seem justified. It is possible that the sole motivation was to keep the number of parts of speech at eight. 

Whatever motivated Donatus, he listed the interjection last among the parts of speech – perhaps because they stand outside of sentence structure; Quintilian (and presumably Palaemon) had listed them last, too. 

The Latin case, the ablative, was also listed last among cases, suggesting a habit among scholars. 

The interjection has stayed within the Western Classical Tradition, usually listed last (likewise the ablative case). 23. 

The ablative is strangely omitted from Alice’s memory of ‘her brother’s Latin Grammar, 

“A mouse—of a mouse—to a mouse—a mouse—O mouse!”’ (Alice’s Adventures in Wonderland, Carroll 1965: 33f). 

These respectively list the 

nominative, 

genitive, 

dative, 

accusative and 

vocative – 

the normal Greek sequence. 

The ablative would be "from or by a mouse."

Quintilian, Dionysius, and Donatus: the start of a pedagogic tradition 99 * 

A pronoun is a word that substitutes for a noun to convey the same meaning and indicates definite persons (i.e. persons previously mentioned). Fittingly, Donatus discussed pronouns immediately after nouns. Dionysius seems have thought that verbs preceded them in importance. The beliefs about priority of substance (already discussed in Chapters 1, 2, and 3) explain why the noun is the first part of speech to be discussed. After that there are competing reasons for selecting a sequence. The notion that a pronoun stands in place of a noun is widely held by English speakers today, even though it is strictly false: most pronouns function within sentence structure as noun phrases and not as nouns; e.g. in Ed kissed the tall girl with green eyes who was wearing dark glasses, and was surprised when she slapped him, although ‘him’ is coreferential with the proper name ‘Ed’ (apparently a noun though arguably a noun phrase), there is no doubt that ‘she’ is not merely coreferential with ‘girl’ but with ‘the tall girl with green eyes who was wearing dark glasses’. Greek onoma and Latin n?men ambiguously mean “name, noun, noun phrase”. The English pronoun one(s) does stand in place of a noun, as in Harry ate all the black olives and Marge gobbled up the green ones. Although ‘ones’ appears within a definite noun phrase it is not itself definite; and although it refers to ‘olives’, these are not the same olives as Harry ate. * 

A preposition is a word preposed to other parts of a sentence in compound forms and in syntax to change, augment, complete or diminish their meaning. Dionysius and Donatus had essentially the same definition for preposition. Both of them applied the term to prefixes, a practice that continued until at least the seventeenth century. * An adverb has no case or tense inflection and when added to the verb modifies or adds to its meaning. Dionysius says simply that adverb ‘has no inflection’ whereas Donatus describes its three degrees of comparison. 

What Dionysius is referring to is the lack of case or tense inflection, which is written into the bulleted definition here. * A conjunction joins the constituents of a sentence, binding the discourse together, giving it order, and filling gaps in its interpretation. This definition combines the essentially similar definitions given by Dionysius and Donatus. Once again this has remained in the Western Classical Tradition. The Techn? Grammatik? and Donatus’ Artes laid the foundation for word and paradigm grammars that developed in the early medieval period; but, notably, they lack anything but the barest sketch of a paradigm (see Chapter 7). 

Because the Techn? Grammatik? was written for speakers of a variety of Greek and the "Artes Grammaticae" for speakers of a variety of Latin, it may be supposed that a student’s existing knowledge of syntax could be called upon to apply the facts identified within these accounts of the parts of speech to the interpretation of classical texts without it being essential to study syntax as well. Nevertheless, a study of syntax would certainly be desirable because of the dialect differences between the everyday language spoken by the students (koin? for the Greeks, Vulgar Latin for the Romans) and the language of the classics they were meant to understand and perhaps imitate. The lack of an account of syntax in Dionysius and Donatus 100 The Western Classical Tradition in Linguistics 

renders their work inferior to that of their respective successors, Apollonius Dyscolus and Priscianus Caesariensis.