by Scott McGill
To understand the constitution of this theoretical knowledge, it would be an error to think, as Karl Barwick (1922) did, that, except for some secondary differences that would be imputable to pedagogical practices, late grammarians passively adopted a putative Stoic model (the tekhnê peri phônês, for which see, e.g., Diogenes Laertius 7,44 = FDS 474) that supposedly circulated in Rome already in the age of Panaetius, disciple of Diogenes of Babylon and Crates of Mallos. The same pattern was allegedly followed even after the diffusion of Remmius Palaemon’s Hellenizing model (first century CE), which is believed to have combined Stoic roots with the Rhodian‐Alexandrian branches developed under Dionysius Thrax’a teaching (second century BCE). As a matter of fact, late Latin grammar included systematic manuals (“Schulgrammatik‐type”), like Donatus’s Ars maior and the Artes grammaticae written by Saint Augustine (extracts from an abridged version, Ars breuiata, of the original, lost work, in GL 5,494–496,14; full, new edition by Bonnet 2013), Marius Victorinus, Dositheus, Pseudo‐Asper, and Pseudo‐Victorinus (GL 6,187–205 ≃ Audax’s Excerpta, GL 7,320–349). Nevertheless, there were also simple collections of basic rules and alphabetical lists for nominal and verbal analogical inflection (“regulae‐type:” Law 1987; De Nonno 1990a, 633 n. 119), like Phocas’s Ars de nomine et uerbo (GL 5,410–439,7 = Casaceli ed. 1974, fifth century CE), Priscian’s Institutio de nomine et pronomine et uerbo (GL 3,443–456 = Passalacqua ed. 1999, pp. 5–41, including a section on participles), Pseudo‐Augustine’s Regulae (GL 5,496,15–524 = Martorelli ed. 2011), the Frg. Bobiense de nomine et pronomine (GL 5,555–566 = Passalacqua ed. 1984, 3‐19) and Regulae Palaemonis (GL 5,533–547,2 = Rosellini ed. 2001). Pseudo‐Probus’s instituta artium gives not only exhaustive inflectional paradigms, but also lists of technical inquiries like quaeritur qua de causa/quare…(40 for the noun alone: GL 4,123,38–130,34). We do not know exactly if these works are the epigones of the third subunit of the technical ars grammatica, as Barwick strove to prove, because the De latinitate treatises by Pansa and Flavius Caper (second–third century CE) are lost, but, interestingly enough, this typology also contaminated grammars of the Schulgrammatik‐type. So, for example, Sacerdos’s Artes (third century CE) are formed of three books, which were independently redacted (De Nonno 1983): After the study of the parts of speech and the virtues and vices of the speech (book 1, the beginning is missing), come the nominal and verbal inflections according to the endings (litterae terminales) of each form (book 2: GL 6,471–495; a parallel redaction is known as Catholica Probi [GL 4,3–43]); finally, metrics occupies book 3. Similar cases are found in Consentius’s Ars de nomine et uerbo (GL 5,347,8–31; 353,17–365,27), part of a systematic ars independently transmitted, and in the second book of a De uerbo written by Priscian’s pupil Eutyches (GL 5,467,16–489,8).
Moreover, some Artes grammaticae elude the threefold structure. Charisius’s work (ca. 360 CE, from Constantinople?) is formed by the juxtaposition of different sources, which mostly are declared. For example, in his chapters on adverbs, conjunctions, prepositions, and interjections (GL 1,180,27–242,12 = Barwick ed. 19642, pp. 233–315), Charisius first takes from Cominianus (early fourth century CE) a short, scholastic explanation; then he quotes the Alexandrian‐connoisseur Palaemon; finally he adds long extracts from Julius Romanus (second half of the third century CE), probably more an amateur than a professional grammarian (Schenkeveld 2004), whose Aphormai (Foundations [of the study of the Latin language]) were a collection of rare and archaic words. In all likelihood, Charisius was an archivist at the imperial court (magister scrinii) who stated in his praefatio that he used his free time to write a work for his non‐Latin‐speaking son (Uría Varela 2006; 2009, pp. 7–11), in which he chose relevant topics that included those outside the traditional curriculum (see below, Vices and Virtues). Another complex Ars grammatica, written by Diomedes (370–380; Dammer 2001), starts with the eight parts of speech, like Donatus’s Ars minor, but, like Charisius, it inserts before each of them specific remarks on their accidentia (genre, number, figure, and case) and a classification of the nouns according to their ablative; books 2 and 3 also present many peculiarities on rhetorical and literary matters (see below, Metrics).
One last important sign of complexity is that Donatus’s acclaimed work gave birth to commentaries, in which his doctrines were explained and sometimes criticized, and on occasion his language was quoted as a linguistic model. Thus Priscian, concerning the use of modo with the function of nunc (GL 3,284,3–4), refers to Donatus (mai. GL 4,373,5–6 = Holtz ed. 1981, pp. 614,5; see also Holtz ed. 1981, pp. 240–241) in this way: Donatus in secunda arte de nomine “sed modo nomina generaliter dicimus” pro “nunc”. Unlike the anonymous and impersonal scholia to Dionysius Thrax, these works were written by historical individuals who could develop personal observations. Cledonius, a Roman senator and teacher at Byzantium, wrote a commentary on both Donatus’s Ars minor and Ars maior (GL 5,9–79). Servius did the same (GL 4,405–448) and added a De finalibus (GL 4,449–455), conceived as a prosodic‐metrical supplement. Likewise, Iulianus Toletanus’s Ars (GL 5,317–324 = Maestre Yenes ed. 1973, ca. 687–690) included commentaries to both Donatus’s Artes, supplemented by a De finalibus syllabis for the first part of the Ars maior (see the corresponding section in Pseudo‐Marius Victorinus’s Commentarium: GL 6,231–239 = Corazza ed. 2011, pp. 36–63, with two independent redactions), and by a Conlatio de generibus metrorum, derived from Mallius Theodorus’s De metris, for its third part (excerpts from the commentary to the second part of Donatus’s Ars maior are found in GL Suppl. ccxii–ccxviii, but see Munzi 1980–1981 for a new, complete edition). The African grammarian Pompeius rewrote many Servian materials on the Ars maior in a pleonastic and discursive form (GL 5,95–312): His “presentative” ecce, the use of the deixis, and the didactic repetitions give us the lively feeling of the work that was done in the classrooms (Kaster 1988, pp. 139–168). The first book of explanationes in Donatum (incompletely edited by Keil, GL 4,486–565; a new edition has been announced by Paolo De Paolis) testifies to the adaptation of the ancient, Peripatetic zêtêmata or problêmata kai lyseis to the scholastic practice: Some fictional, lifelike characters (et interrogauit Filocalus…, et interrogauit Rusticus…, interrogatus…respondit) give a lively style to the exegesis of the magister (De Nonno 2010, pp. 187–202).
28.2 Sublexical Level
The standard beginning of an Ars grammatica contains sections on graphophonemes (de litteris) and syllables (de syllabis), that is, on what we would consider to be phonological analysis of the articulated voice (Desbordes 1990; on the general de uoce chapter see Ax 1986). The linguistic unit called littera combines the written and the oral dimensions. Definitions like “littera is the littlest part of the articulated sound” (see TLL 7.2,1515,21–30) are based on articulatory parameters. Thus < X>, which corresponds to Greek < Ξ>, cannot be a littera, because it has a biphonematic character (velar plosive plus sibilant cluster). Other definitions distinguish the phonological and graphematic levels: “littera is the sign of the phonological unit and, so to speak, an image of the sound, which is susceptible to written representation” (see TLL 7.2,1515,38–49). So the parallel between the Latin initial aspiration and the Greek rough breathing led the grammarians to consider < H > as a graphic sign (nota) of a vowel’s specific articulation, not a graphophoneme on its own.
The section on syllables (de syllabis; see Scialuga 1993) marks a profound difference from the supposed Stoic model of the Latin grammar in Barwick’s reconstruction. If, pedagogically, the syllable is an essential step after having learned the alphabet and before passing to words and utterances, it has no place at all in the Stoic dialectical tripartition phonê/lexis/logos (cf. Diocles ap. D.L. 7,55–57 = FDS 476). Latin grammarians paid specific attention to the prosodic features of the syllables because of the importance of the metrical scansion (Leonhardt 1989, pp. 24–71): So, for instance, the syllabae communes, wh
ose quantity is determined by the presence of sibilant, mute, and liquid clusters in the immediate context, are a recurrent topic.
Chapters on accents (de accentibus) and punctuation signs (de distinctionibus) refer to the suprasegmental units (see Ebbesen 1981; Müller 1964; still useful is Schoell 1876). The nature and position of the accent are studied through Greek categories, like the presence (dasia ⊢) or the absence (psile ⊣) of aspiration, and the joining (hyphen ͜) or separating (diastole ₎) of the words of a sentence. This transfer of Greek theories causes many inconsistencies and obscurities for a modern reader (cf., e.g., Donatus mai. GL 4,371,31–372,13 = Holtz ed. 1981, 610,16–611,9), but, notwithstanding these difficulties, a general distinction is made between toni breuis (˘) and longus (¯), which indicate the quantity of the vowels, and accentus acutus (´) and circumflexus (^), which designate the places of a tonal heightening, respectively protracted or cursory. Almost always added by a “hand” that is different from that of the original scribe, those signs were indeed put on the Virgilian papyri and reveal the performative dimension of the literary texts, which were read aloud at school. For example, the bilingual papyrus Nessana II 1 (Scappaticcio 2012 no. 6: end of the fifth century–early the sixth century CE) gives Aen. 2,99 as follows: Īn uulgum ambíguas et quaerere: cónscius árma (l. 557–560), with accents that do not indicate the metrical ictus. Among the extant theoretical treatises, a late De accentibus, where one can find Priscianic materials (GL 3,519–528 = Giammona ed. 2012), contains interesting remarks on accents, aspiration, and instances of sandhi. A specific monograph on aspiration (De aspiratione) is jointly transmitted with Phoca’s authentic work, but, in fact, it is a work of a humanist (GL 5,439,10–441 = Jeudy ed. 1976).
Orthography (orthographia) has a twofold aspect: It covers not only the correct writing but also the theoretical study of it (De Paolis 2010). Marius Victorinus’s grammar (half of the fourth century CE), which is incompletely conserved, is the only comprehensive work with an extensive section on orthography (GL 6,7,35–26,13 = Mariotti ed. 1967, 70,18–90,11). This knowledge is justified by the scholastic emendatio of the books of literary authors, as shown, for example, by this remark: non est, ut emendastis, “porca praecidanea,” sed “praecidaria” (GL 6,25,16 = Mariotti ed. 1967, 4,13–14). Victorinus’s work also testifies to the late trend toward inclusiveness (see above, A Textual Typology), since traditionally autonomous fields (the second subunit of the technical ars grammatica: see above, A Textual Typology) are incorporated in that of the technical grammar. Most of the ancient monographs on orthography are lost, but two works of the age of Hadrian are preserved: Scaurus’s and Velius Longus’s De orthographia (respectively GL 7,11–33,13 = Biddau ed. 2008 and GL 7,46–81 = Di Napoli ed. 2011). Eventually, the new needs of the average readership produced a shift from the philological to a new, prescriptive approach: The written and oral competences of an increasing number of non‐Latin‐speaking people demanded to be emended and improved. So the well‐known Appendix Probi, of uncertain origin and period, contains, besides miscellaneous materials, an orthographical antibarbarus (GL 4,197,19–199,17 = Asperti and Passalacqua eds. 2014, 20–27), which is immediately followed – in the MS Napoli BN lat. 1 – by a list of differentiae (see below, Lexicon, lexica), with problems of pronunciation, morphology, and semantics. The same features appear in a De orthographia dedicated to Eucherius (ca. 434–450), bishop of Lyon, by Agroecius, bishop of Sens (GL 7,113–125 = Pugliarello ed. 1978), and intended to be an integration of a homonymous compilation attributed to Caper (GL 7,92–107,2). These rather glossematic works had a strong influence on the high‐medieval treatises by Bede and by Alcuin (redaction a: GL 7,295–312 = Bruni ed. 1997; redaction b: PL 101,902–920), henceforth alphabetically structured. Providing many extracts from ancient authors not otherwise preserved, Cassiodorus’s own De orthographia aims at giving the classical linguistic and philological tools to a monastic community that devoted itself to the study and transcription of the Bible, so that errors in copying manuscripts could be avoided. Papiri(an)us’s original work, besides a short, independent extract (GL 7,216,8–14), is partially preserved in Cassiodorus’s excerpts (GL 7,158,9–166,8 = Stoppacci ed. 2010 § 4,1–109). Specific norms on < b > and < u>, which tended to overlap in their fricative pronunciation, are given in the sixth century CE by Martyrius’s De B muta et V uocali (GL 7,165–199); Cassiodorus used this work too, distributing its four sections in chapters 5– 8 of his treatise (Keil’s edition puts Martyrius’s text under Cassiodorus’s, each one with its own app. crit. on the same page).
28.3 Morpholexical Level
Chapters on speech and its parts (de oratione eiusque partibus) start with the definition of oratio as “utterance by oral production, which is structured by words, as if it was an ‘oral‐ratio’.” Only Priscian, who adapts Apollonius Dyscolus’s model (see below, Syntax), takes into consideration both the grammatical consistency (congruitas) and the semantic autonomy (perfectio): “oratio is a coherent combination of words, which shows a complete meaning” (GL 2,53,28–29, my translation).
The taxonomies of the parts of speech are a recurrent topic (Garcea and Lomanto 2003). After having observed that a lexical element can replace an entire oratio “when it shows a full sentence (plenam…sententiam)” (GL 2,54,1–4), for example when one uses the imperative forms or answers a question by a single word, Priscian adds that there is another form of independence besides that of the monorhematic expressions. He thus refers to an Aristotelian pattern, opposing the semantically necessary and sufficient constituents of an utterance to the other parts of speech (Ars grammatica group 2013, pp. 11–16): “For the dialecticians, there are two parts of speech, the noun and the verb, because they are the only ones that, combined together without anything else, form a complete utterance; the other parts are called synkategoremata, i.e. co‐significant (consignificantia)” (GL 2,54,5–7, my translation).
Chapters on morpholexical units can follow different structural orders (still useful are Schoemann 1862; Jeep 1893; new, important perspectives in Matthaios 2002). Notwithstanding Barwick’s synoptic tables (1922, pp. 3–87), not only the sequence but also the definitions of the parts of speech vary significantly: Some of them are, traditionally, etymological and morphological, whereas semantic‐referential definitions progressively became widespread, thanks to the diffusion of Neoplatonic terminology brought to Rome by Porphyry and assimilated by authors like Marius Victorinus and Donatus. So, for example, the stoic definiens “common vs. peculiar quality” of the noun (cf. Diocles ap. Diogenes Laertius 7,58 = FDS 536) recedes into the background in favor of the distinction between corporeal and incorporeal things (corpus as res corporalis vs. res as res incorporalis), be they individuals or collections. Starting from Donatus (GL 4,373,2–3 = Holtz ed. 1981, 614,2–3), this is considered the semantic proper feature of the noun, whose defining genus is being a part of speech and its species belonging to the case‐inflection (see Luhtala 2002; 2010, pp. 224–237).
Bare inflectional paradigms are a common topic of the grammatical regulae‐type (see above, A Textual Typology) and are best paralleled in the Greek tradition by Theodosius’s Basic Rules [Kanones] for Noun and Verb Inflection (fourth‐fifth century CE), commented upon by Choeroboscus and Joannes Charax (GG 4/1–2).
28.4 Vices and Virtues of Expression: Between Rhetoric and Grammar
According to Barwick (1922, pp. 95–101), chapters on vices and virtues of expression (de uitiis et uirtutibus orationis) form the “third part” of the allegedly stoic model of the Latin grammars, after the sublexical and the morpholexical sections; in fact, they are the most explicit signs of a late development in Latin grammar (Baratin and Desbordes 1986; case studies in Vainio 1999). Unlike Greek sources (cf. Diocles ap. Diogenes Laertius 7,59 = FDS 594), late Latin grammars treated multiple instances of transgressing the linguistic norm, both in a negative (uitia: barbarism, solecism) and in a positiv
e (uirtutes: metaplasms and figures of speech) way. So, as forms of ornamentation, virtues lie beyond the grammatical correction and pertain specifically to the rhetorical field of textual analysis, already found in Cicero’s ideal of the elocutio ornata (cf. De orat. 3,52–53).
The polar system “virtues vs. vices” is arranged in four categories, constituting the quadripertita ratio (Quintilian, Inst. 1,5,38: Desbordes 1983; Ax 1987): two quantitative variations, that is, addition (additio: A B C D) and subtraction (detractio: A B); a qualitative variation or mutation (transmutatio: A B E); and a permutation of the items (immutatio: A C B). Following a synchronic and normative point of view, which has no parallel in former texts (Rhet. Her. 4,29 on adnominatio/paronomasia; Varro Ling. 5,6; 6,2; 7,1 on etymology), late grammarians created a symmetric system, where the same type of phenomena was considered virtue or vice, depending on its presence or absence in poetry. An interesting example is given by the Gallic grammarian Consentius (fifth century CE), whose two independent sections of an autonomous reworking of Donatus’s tradition are left. In his De barbarismis et metaplasmis (Table 28.1) he chose to take his examples of barbarism not from literary prose texts but from the lively language of his country (GL 5,387,29–393,3 = Niedermann ed. 1937, 3,19–13,8).
