Introductory Remarks

Many cultures have schooling and literacy, all have ideas about language, but not all recognize anything resembling a discipline of ‘linguistics.’ Schooling, religion, and official culture are part of institutions for “making up people” (Hacking Reference Hacking2006), and, in the case of language, making experts and teachers who are empowered to establish language norms and standard forms. These norms and forms channel people into certain linguistic attitudes and activities and prepare them to be social agents, transformers, or resisters. Although schooling and literacy have historically been closely related to linguistics (grammar), the discipline of linguistics has always had a complicated relationship with dominant culture and other fields of knowledge touching on language and meaning.

Before 1500 ce, linguistics or ‘grammar’ emerged within practices of writing, interpretation, and translation, as well as within larger cultural movements, including the expansion of Christianity and Islam throughout the Mediterranean and Western Europe, the growth of universities, and attitudes toward correctness and linguistic purity in increasingly multilingual contexts. In those premodern worlds, the cultural and intellectual power of linguistics centered in reading, writing, and praying, or teaching others to do so. The emergence of grammar as a discipline depended on the social creation of the ‘grammarian’ or ‘linguist,’ responsible for literacy education and for speaking about linguistic form, usage, language origins, and meaning, topics some considered to be the province of philosophers or theologians. Grammatical discourse separates language experts from language users and emerges as a new branch of cultural authority. Language study, like language itself, is a form of symbolic power.

Ancient Writing and the Cultural Production of Linguistics

Thinking about language throws a pebble onto the watery surface of language. When we talk about language with language, how do we distinguish what we are referring to from the act of linguistically referring to it? When we teach about language in school, we objectify language as an abstraction or theoretical topic and also locate linguistic forms within fields of cultural difference. Sometimes written language is posited as the ideal object of perfect utterance. Other times, school language is presumed to clean up usage according to a literary or scriptural standard. Historically, linguistic thinking has presumed not only that language can be known but that it is capable of being regulated or purified. Linguistic thinking divides the water of active language into two parts, code and gloss. Language becomes the systemic code of knowledge, while gloss produces a disciplinary metalanguage authorized to speak about that language system.

A ‘linguistic turn’ occurs when language, instead of labeling experience and phenomena, is reimagined as mediating or constituting experience and phenomena. Although the concept of linguistic turn is associated with Wittgenstein’s (1953) critique of traditional philosophy, ancient and medieval philosophy, as well as some medieval approaches to grammar and theology, constitute important antecedents for the theory of language as mediating reality and how we know it. Moreover, we can expand the concept of a linguistic turn by focusing on how the premodern quest to ‘know’ language meant that speech and writing became schooled practices and topics of discourse. Eventually, premodern grammar became a master discourse, sometimes a rival of theology or philosophy but always related to literacy.

Writing began about 5,500 years ago, and almost immediately literacy prompted deliberate linguistic thinking and standardization. Writing systems tacitly categorize linguistic and sometimes mental ‘elements’ by identifying words, syllables, or discrete sounds as primary units of utterance. Adding or deleting graphic characters testifies to thinking about the existing inventory of sounds in a language and the capacity for representing sounds visually. Western writing systems evolved from pictograms to ideograms to arbitrary or symbolic characters in a process of ‘deiconization.’ During the first Egyptian dynasty (2000–1000 bce), scribes used stylized pictorial signs (hieroglyphs) to designate individual words, discrete sounds, or non-pronounced elements which differentiated meanings. Later, individual hieroglyphs acquired phonetic values derived from the initial consonant of the word or concept denoted by the sign. In Mesopotamia (c. 3500 bce), the Sumerians and Akkadians used groups of wedge marks in clay (cuneiform) to indicate entities, concepts, syllables, or individual sounds. Most early cuneiform texts were financial and commercial records, and Akkadian scribes’ bilingual word lists reflected this pragmatic literacy. Around 1200 bce, Phoenician scribes developed a consonantal alphabet, later disseminated among surrounding peoples (Hebrews, Samaritans) and then remarkably improved on by the Greeks (fifth c. bce) (Reiner Reference Reiner1973; Baron Reference Baron1981; Coulmas Reference Coulmas1996).

Writing systems make language visible, increasingly conventional, and retrievable over time and space. Writing, like linguistic description and analysis, objectifies language. Whereas Phoenician and later Hebrew and Arabic writing used largely consonant characters while relying on diacritical marks to indicate vowels and phonetic features (e.g., aspiration), Greek alphabetic writing created character–sound correlations for all available Greek sounds and clusters. Greek alphabeticization tacitly theorized linguistic sounds as consonants, vowels, and semi-vowels, thereby creating new linguistic categories below the word level. Pre-Socratic philosophers quickly recognized the usefulness of this atomic approach to language and used the Greek alphabet to model physical elements, ‘atoms.’ They regarded entities in matter and language as divisible or indivisible. Aristotle (1984; Metaphysics, 985b4–20) states that the atomists considered the elements (stoikheia) as “the causes of all other qualities.” They posited three different kinds of qualities among elements, based on alphabetic characters: shape, order, and position: “Thus A differs from N in shape, AN from NA in order, Z from H in position.” Some Pre-Socratics further claimed that letters or mental sounds are themselves material elements, although not iconic with what speech refers to. Stoics and later Christian grammarians explored the relations between material and immaterial aspects of language, focusing on incorporeal or mental reference, meaning, and understanding, for example, the theology of nihil (‘nothing’).

Ancient linguistic thinking was partly structured around the tension between whether language was naturally (really) or conventionally connected with reality. This debate was part of a wider argument as to whether immaterial concepts had real existence outside the mind. While Plato, Aristotle, and the Stoics all posited a realm of incorporeal meaning and reality beyond physical existence, they theorized language as a qualified conventional system for uttering and representing meaning or getting things said. Aristotle’s linguistic ‘realism’ emphasized the arbitrary correlations between written characters and spoken sounds and language’s conventional relation to realis. Speech sounds are “symbols (symbola) of affections/movements of the soul”; written characters are “symbols of spoken things.” Just as different peoples have different writing systems, so peoples have different phonetic inventories. Nonetheless, “the affections of the soul, of which these [spoken and written words] are in the first instance signs, are the same for everyone, and the objects of which these [affections] are images are also the same” (Aristotle Reference Barnes1984; De interpretatione, 16a3f.). For Aristotle, “no name (onoma) is a name naturally but only when it has become a symbol,” that is, when used in a sociocultural context. Animal and human inarticulate noises, grunts, and groans indicate feelings or responses, but none is a word (16a26f.). The Stoics believed that language was composed of matter (sound) and natural processes but that language as an expressive system was organized conventionally. Ancient philosophers and grammarians argued that meaningful language, articulate utterance, was ‘writable,’ so writing became coterminous with language itself. Grammar and literacy were intertwined.

In pre-1500 ce Asia, the shift from ideogrammatic to alphabetic writing followed other trajectories. Written Arabic developed late from Phoenician-Aramaic script (fourth c. ce). By the seventh century ce, scribes were using dots with consonants to create closer one-word–one-sound correlations and avoid ambiguity. About the same time, Hebrew scribes relying on the Massoretes’ versions of the Torah also incorporated diacritics (niqqud) to distinguish consonantal pronunciations. Chinese and Indian writing systems derive from two different visual archives. At first, the Chinese used pictograms to represent objects and convey pronunciation information. Beginning in the second century ce, under western influences, Chinese scholars combined two or more pictograms to create new signs, ideograms, using a tool called fănqiè (literally, ‘reverse cutting’). These new characters represented the pronunciation of one character by combining two other characters. When the new sign composed from two pictograms was pronounced as a linguistic unit, the newly formed word replaced earlier words in the lexicon. Later, other Chinese characters were devised to indicate semantic functions or specialized usage (Branner Reference Branner and Auroux2000: 36–40). In 601 ce (Sui Dynasty), the first Chinese rhyme dictionary using fănqiè, Cutting Rhymes, was published. Chinese scholars and scribes’ gradual deiconizing of the writing system segmented spoken Mandarin and subordinated speech to new combinations of written characters, which then became canonical in official texts and poetic composition. Indian writing systems, especially Sanskrit (first millennium bce) and Brahmi (c. 300 bce), also became more deiconized and conventional over many centuries. The earliest Sanskrit scribes used iconically motivated pictograms. Eventually, these pictograms became more phoneticized, that is, more conventionally correlated with syllables and concepts and then with individual sounds independent of meaning or concept associations. Special schools were established to train scribes to properly write Sanskrit characters and sacred texts.

The Stoics considered writing to be a part of the larger question – What is language? – and their philosophy of language was fundamental in developing linguistics and disciplinarity in the late Roman and early medieval periods. Stoicism was founded by Zeno of Citium around 300 bce and was centered in the Hellenistic world before being introduced into Rome by Panaetius of Rhodes in the second century bce. According to Diogenes Laertius (early third c. ce), most Stoics divided logic into rhetoric and dialectic, and dialectic into argument (topics of discourse) and language (1925: 7.39, 42–3). Language study was part of philosophy, the “science of statements true, false, and neither true nor false” (7.42). Language study “included written language and the parts of speech, with a discussion of the errors in syntax and in single words, poetical diction, verbal ambiguities, euphony and music, and according to some writers chapters on terms, divisions, and style” (7.44). The Stoics distinguished mere sound or physical voice (phone) from articulate, meaningful utterance (lexis). The principal unit of language was the sentence or discourse, logos, comprised of subject + predicate (7.55–8). Spoken and written language are related to realis through lekta. The lekton (noun, from verb legein ‘to say’) is the incorporeal some-thing and later the incorporeal word-concept which underpins what a speaker means to say, what can be said or written in a given situation, what is sayable. Lekta mediate between an abstract linguistic realm and linguistic usage as situated social behavior.

By locating language study within dialectic and positing a ‘some-thing’ between language and realis, the Stoics continued the Greek idea of philosophy as the premier discipline for understanding language and thinking. They paid special attention to the characteristics of Greek verbs for distinguishing different kinds of semantic modality (hypotheticals, conditionals, etc.) and representations of temporality. The Stoics associated articulate utterance and grammaticality with what is writable and sharply criticized grammarians’ declensions and paradigms as inadequate. Moreover, they argued that because alphabetic literacy can be made to inscribe many kinds of utterances, written forms interrupt grammatical conventions with representations of non-referential utterances which convey affect (cf. Blank & Atherton Reference Blank, Atherton and Inwood2003).

The Stoics were joined by the Skeptics in their criticism of the technical grammarians who did not consider meaning, situated speech, or the relations between language and reality. According to Sextus Empiricus (second c. ce), technical grammarians “plume themselves and give themselves great airs” as the presumptive guardians of linguistic purity and literate knowledge (Sextus 1949/1997: Against Professors 1.5.97). Sextus and the Stoics argued that meaningful speech is something more than the sum of its constituents (parts). However, Sextus went further to criticize some Stoic assumptions, arguing that the immaterial lekta cannot really exist because their existence is asserted and then justified with speech, which is then based on other lekta, and so on in an infinite regression (Sextus 1933/1993: Outlines of Pyrrhonism 1.156ff.). Sextus’ critique called into question the very ground of a discipline or science of linguistics. The Stoics, on the other hand, regarded language as a dialectical tool which can be disciplined to investigate all aspects of human experience, including language itself.

Stoic philosophy of language, concept formation, and ethics proved to be fertile ground for later critiques of data-rich, theory-poor technical grammar and for Christian thinkers concerned with linguistic reference as well as linguistic form. After his conversion to Christianity, Augustine redeployed Stoic concepts of conventional signs and immaterial (conceptual) meaning to create a philosophical framework for Christian belief and sacred reading. Like the Stoics, Augustine resituated language study within a general analysis of natural and conventional signs. Words can signify themselves, other signs, or material or immaterial realis, depending on the context and intention. In De dialectica (composed c. 387 ce), Augustine described the relations among speech, writing, and understanding in semio-Stoic terms: “a sign (signum) is something which is itself sensed (apprehended in itself) and which indicates/shows to the mind (animo ostendit) something beyond the sign itself” (1975: ch. 5). Language’s referential flexibility and conceptual elasticity enable people to access the immaterial Inner Word, the Christian Logos, which would otherwise remain inaccessible or even unrecognized unless understood as a some-thing.

Augustine’s account of speech and writing in Stoic terms was influential in later linguistic thought. In his De doctrina Christiana (1962: Bks. 1–3 [397 ce]; Bk. 4 [426 ce]) and elsewhere, Augustine criticized Roman technical grammarians and teachers as superficial formalists or purveyors of affectively seductive texts such as the Aeneid. Virgil’s scenes of emotional turmoil or edgy behavior, as in Aeneas’ loss of his father or his affair with Dido, were dangerous texts for Christians to read without gloss, but the technical grammarians only discussed the formal elements of Virgil’s poetic language. Christian readers should take what they can from pagan intellectual culture so long as it serves a Christian literacy. Augustine claimed languages or dialects are all equally capable for accessing the Inner Word because no single language can account for all meanings and because individuals understand truth through their own mental connections. No individual variety can be set up as the exemplar of latinitas (correct and elegant Latin expression) or absolute meaning (1962; De doctrina Christiana 2.13). He adopted a reception hermeneutic and distinguished hearing speech and other noises from seeing and interpreting written (visible) language and seeing and interpreting gestures or non-written objects. Meaning is located in the dicibile, what can be said about the world (1975; De dialectica, ch. 5). For the Stoics and Augustine, the dicibile bridges socially and discursively constructed meaning with metaphysical or transcendental meaning.

According to the Stoics and Augustine, the science (ars) of grammar is properly concerned with vox, and vox as articulate utterance is bound up with writing. Speaking and reading hold equal status in Augustine’s semiotics of language in that both rely on immaterial understanding of concepts of reality. Here, Augustine’s theory is very close to Saussure’s notion of the ‘sound concept.’ Individual readers may pronounce letters and words differently, so the sound images readers associate with conventional groups of letters are not stable or fixed. As material linguistic elements (stoicheia, elementa), sounds and letters are temporal, transient, and vary depending on the speech community’s habits, usage, and attitudes. But within the linguistic system as a system and apart from any individual speaker, the sounds and letters of a language are relatively stable. Pronounced words may vary from speaker to speaker and over time, but the ‘proper’ word is the mental entity, the Inner Word, which as part of a linguistic system pre-exists any utterance.

Augustine modifies the ancient concept of the letter as a linguistic unit as described by Latin grammarians. In De dialectica and elsewhere, he says both speaking and reading create signs of words, which the perceiver reconstructs as mental images. Spoken sounds are heard through the ears. Letters (litterae) are the minimum elements of articulate sound, with three characteristics: vocality (potestas), shape (figura), and a metalinguistic label, which places it in the phonetic and scribal inventory (nomen). Letters and thus writing achieve their power by combining vocal values to produce or trigger shared meanings. But Augustine then sets writing on a higher plane: “written sounds/letters” are “completely silent” (litterae scriptae, quamvis omnino tacita sit [littera]; 1975: ch. 5). Augustine’s apparent redundancy, “litterae scriptae,” indicates the complexity of late ancient and medieval theories of writing and language. Like other philosophers and grammarians (and not a few classroom teachers and language mavens), Augustine sometimes considers writing to be equivalent to language itself, prior to utterance. Elsewhere he worries about the ambiguity of using the term littera to denote both linguistic sounds and inscriptions, but he doesn’t offer an alternative. Grammatical metalanguage had not caught up with linguistic perception. Later, Isidore and some Carolingian grammarians did propose an alternative explanation for littera, based on the word’s derivation from legitera (‘reading’) or iter (‘path’). Letters were imagined as indices or tracks of writing and reading, traces of the author’s voice and scribe’s hand, guides for oral or silent construing, and textual pathways readers follow to gain information and wisdom. But the relation between sounds and written representations remained a problem for medieval linguistic thinking.

In Augustinian semiotics, the ambiguity of the concept of littera opens a space for considering the elements of language as not just uttered or seen signs but as signs cognized in memory. Augustine privileges silent reading over conventional reading aloud to oneself or others in public performances. His theory of the Inner Word destabilized latinitas as a norm by declaring that all languages and usage can be means for accessing a singular truth if the speaker or reader can connect with the Inner Word, the divine lekton or dicibile. Augustine subordinates the science of language (grammatica) to an ethical standard of faith and understanding. Still, his theory of literacy and textuality creates a place for delimited linguistic reasoning within an investigation of discursive practices.

Medieval Multilingualism and Linguistics

Ancient and medieval linguistics developed within various multilingual contexts. When versions of Greek, Latin, or Arabic emerged as language norms or the basis for grammatical description, literacy education, and social order, grammarians were effectively shaping language ideology. When linguistic thinking took issue with accepted norms for polite, learned, or divine speech, grammarians marked language variation in multilingual situations and challenged language ideology. Grammarians and scribes contributed much to the development of linguistics with language comparisons.

In the Hellenistic and Roman worlds, the literacy teacher (grammatikos) was an important cultural figure, despite the Skeptics’ criticisms. Donatus (fourth c. ce) taught Latin grammar in Rome, Priscian (c. 500 ce) in Caesarea, and their respective grammars constituted the core knowledge of Latin grammar for centuries. Fifth- and sixth-century grammarians wrote extensive commentaries on Donatus’ text and on the Aeneid and Terence’s plays, core texts in the Latin curriculum. Grammarians were not only teachers and arbiters of cultural traditions and values, many were also close observers of the changes in Latin usage from what appeared in canonical school texts and especially the differences between spoken and written forms. In Constantinople, the grammarian Cledonius (fifth c. ce) noted the Latin of the antiqui (‘ancient writers’) sometimes differed from that of present-day speakers and writers: “We should now say iugerorum; the ancients used to say iugerum, like tuberum” (Keil Reference Keil1961: v, 47).

As Christianity expanded throughout Romanitas (‘Romanness’), grammar became a master discipline structuring monastic education and literacy. Grammatical models were applied to exegesis and other areas of knowledge. Grammarians used glosses and commentary to explain meanings, etymologies, allusions, and historical references in literary or scriptural texts. Teaching reading and writing in monasteries and using texts to guide liturgical pronunciation and other public performances required that scholars produce linguistic handbooks and analyses for students and teachers from different language backgrounds. After 400 ce, Latin became more differentiated and regionalized. By 700 ce, Latin was no one’s native language. The focus of Latin grammatica shifted from teaching literacy within a speech community to introducing students to a second or third language which commanded special and elite status as the language of faith and learning. In a parallel movement at about the same time, the expansion of the Muslim empire by conquest around the Mediterranean depended partly on the deployment of grammar and religious literacies as means for maintaining religious orthodoxy and a reading knowledge of Classical Arabic among new Muslims who were not speakers of traditional Arabic.

In multilingual western Christian culture, Latin grammar and texts were positioned as a culturally unifying practice and a learned field for knowing what Language is. Definitions, principles, and Latin examples from the Roman artes grammaticae (‘technical grammars’) were adapted to the needs of Christian literacy and monastic text production and to account for the changed usage and lexicon of contemporary Latin and sometimes vernaculars. Latin grammar also formed the basis for investigating other languages in contact situations, although not without changes and challenges to prevailing linguistic assumptions. In the eighth century, when Pope Gregory II summoned Boniface to be examined in Rome on his orthodoxy, the ‘Apostle of Germany’ asked that he be allowed to write his responses in Latin and send them to the Pope’s advisers, “so that the silent letter (muta tantum littera) alone may reveal my beliefs in a well-reasoned manner (rationabiliter)” (Willibald Reference Levison1905: ch. 6). As a non-native Latin speaker and writer, Boniface (born in England as Wynfrid) expressed more confidence in his ability to write “reasoned” Latin than to speak the southern European variety pronounced in Rome. When Alcuin oversaw the Carolingian revision of the Latin Bible, he effectively installed an eighth-century Latin norm and stratified other varieties as new Romance vernaculars (Wright Reference Wright and Wright1996). In this context, written Latin became a shared vocabulary and discourse which floated above vernacular situations as a transregional, almost transcendental norm.

From the second century bce on, throughout Asia and the wider Mediterranean world, scribes and grammar teachers of Greek, Latin, Hebrew, and Arabic became increasingly important for producing and commenting on sacred texts and for schooling new readers in the intricacies of reading complex pagan texts and scripture. Scribes and readers noticed how when reading aloud, the voice lags behind the eyes moving across the page, how the eyes might skip a line or word or transpose characters, or how individual characters might correlate ambiguously with more than one sound. The scribal fetish of correctness was keyed to how the text, especially the sacred text (Bible, Torah, Qur’ān), governed proper reading aloud or stabilized the received language of scripture (Latin, Hebrew, Arabic) as divine speech.

Medieval scribes, relying on the writings of Roman imperial grammarians, also recognized differences between spelling and pronunciation. In nonnative Latin-speaking regions, scribes adapted the Roman orthography to represent their vernaculars’ sound inventories. Anglo-Saxon scribes introduced characters from Ogham and runic writing to represent Old English sounds that didn’t occur in Latin: æ, þ, ð, 3, p. After 1066, Norman scribes copying manuscripts in English used the h character to mark certain English consonants they heard as continuants. Some grammarians took up the challenge of orthographic reform based on the differences between speech and writing.

In monasteries and courts the exegete, the liturgical reader and the scribe held honored positions. In De orthographia, Cassiodorus characterized the scribe as the protector and purveyor of the divine Word through correct spelling (Patrologia latina [PL] 70:1241C–1242). In the seventh century, Isidore interpreted some letter shapes (figurae) as signifying sacred truths iconically: A and Ω signify Christ at the beginning and end of his life; O signifies death; T, the Cross; and Y, life, because “the forked path which is above begins in youth, of which the right side is steep, but leads to blessed life, while the left side is easier, but leads down to ruin and destruction” (1911; Etymologiae 1.3.7). Isidore’s interpretation relied on the scribe’s pen strokes as much as on the conventional shape of the letter Y.

Some writers analyzed letter–sound correlations in language contact contexts. Communicative correctness and disambiguation were matters of faith as well as linguistic performance. Abbo of Fleury (late tenth c.) compared Greek, Roman and Old English orthography in order to teach non-native Latin speakers in monastic settings to read Latin properly and not confuse, for example, actus with agtus or scriptus with scribtus, that is, voiceless/voiced consonants (c/g or p/b) before t (1982: chs. 26–8). Abbo, like Isidore and Hrabanus Maurus in the ninth century, stressed the study of orthography to ensure readers correctly accented written texts during liturgy and avoided mispronouncing and perhaps misleading those whom they were instructing: “there are many words which ought to be discerned only by means of the accent by the pronouncer, so that he might not be led astray in their meaning” (Hrabanus Reference Hrabanus1996: II.52.46–8). In England, the priest Orm (twelfth c.) produced a poorly inscribed manuscript of versified vernacular sermons, Ormulum (1878), and invented his own spelling system to ensure their proper pronunciation. He believed many post-Conquest priests were no longer pronouncing English properly, thus putting their parishioners in spiritual danger. Orm used double consonants to mark the preceding vowel as short and a regular meter to indicate syllable stress. He wrote the older, ‘open’ yogh 3 to represent [ʤ] and [j], and the newer, ‘closed’ yogh g to represent [g].

Some medieval orthographic reforms were based on contrastive phonemic analysis. Around 1150 ce, the anonymous author of the First Icelandic Grammatical Treatise (see Haugen Reference Haugen1972) compared, with sophisticated analysis, Latin and Icelandic phonology and writing. The grammarian analyzed the pronunciation of Old Icelandic (OI) and then proposed an orthographic reform: “Some of the consonants of the Latin alphabet were rejected, and some new ones added. No vowels were rejected, but a good many were added, since our language has the greatest number of vowel sounds” (Haugen Reference Haugen1972: 13). His careful analysis of OI pronunciation revised the framework for describing Latin speech as received from Roman technical grammarians. The anonymous grammarian considered the Roman alphabet to be an expandable cognitive tool for making visible in text the elementa of OI utterances. Latin literacy was the starting point for inscription, and the Roman alphabet was the model for sound–letter correlations. But as vernacular speakers and Anglo-Saxon and Icelandic scribes understood, the Roman alphabet did not account for all possible linguistic sounds. Writing the ‘excess’ vernacular sounds required new characters. Latin–vernacular language contacts produced new understandings about writing, linguistic difference, and the needs for literacy education.

Scribes also contributed to linguistic knowledge and practices in the later medieval multilingual world. After 1100, more people had access to Latin and vernacular literacies. The growing university culture depended on multiple copies of key texts. Expanding royal and ecclesiastical bureaucracies demanded archives and retrieval systems as well as more standardized text formulas and scripts. Scribes, authors, and compilers developed new technologies for organizing written work and facilitating reading: tables of contents, indices, running heads, chapter divisions, text highlighted in contrasting colors or engrossed scripts, and extensive use of diagrams and charts to present information. These technologies enabled grammar writers and language philosophers to develop more complex strategies for analyzing words and sentences with contextual material to guide readers. For example, some twelfth- and thirteenth-century Priscian manuscripts included marginal diagrams presenting his analysis of pronouns in schematic form, showing the subsets of nouns and verbs, and tables representing verb paradigms. Grammar texts which might otherwise be written as undivided lines of prose were separated into sections, paradigms, and charts for easier visual access. The anonymous writer of one such formatted grammar, the Flemish Latin grammar Exercitium puerorum grammaticale per dietas distributum 1485 (Children’s Grammar Exercises for Daily Use), published just after the invention of printing, wrote:

Page layout, a key feature of packaging linguistic information, was becoming more individual reader oriented in the later Middle Ages.

But even as later medieval grammarians took advantage of new ways to represent linguistic information, the formalization of linguistics as a ‘science’ often meant that the importance of sound–writing correlations were downplayed in favor of a rationalized, systematic approach to syntax and meaning based on Aristotelian categories. The Modistic grammarians and other philosophers of language reoriented grammatical discourse to the connections between Latin utterances and mental representations.

Commentary Discourse

Much linguistic thinking in premodern Euro-Christian, Islamic, and Indian traditions was produced in commentary threads. Commentary discourse is based on textual collaborations and disseminations. Commentary is an interpretive practice which locates and appropriates a sacred or canonical text within an authoritative discourse accounting for the text’s meaning, significance, and value. Commentary as a supplementary discourse appropriates or resituates a core text in another context. Eastern and western grammatical discourses were codified as teaching grammars or as commentaries on teaching grammars or on scripture (Bible, Qur’ān), canonical poems (Aeneid), or learned texts (Donatus, Priscian, Galen). Secondary grammatical commentaries included word lists, glossaries, pronouncing guides, and composition exercises. These grammatical intertexts existed alongside philosophical discourses on language composed by philosophers and theologians. While premodern linguistic thinking was bound up with literacy education and language ideology, grammarians often clashed with other teachers, scholars, priests, and rulers about the extent to which grammar should be able to account for textual meaning or sacred wisdom (Amsler Reference Amsler1989; Irvine Reference Irvine1994).

Commentaries travel with core texts. In Hellenistic schools, the grammar attributed to Dionysius Thrax and Homeric poems were the canon for literacy education and the basis for numerous commentaries on language forms and textual meanings. In the Roman Imperial schools, Palaemon’s grammar, Aeneid and Terence’s plays were texts of reference for both Latin form and ethics. By 500 ce, Donatus’ Ars minor (a compendium of Latin grammar) and Ars maior (a longer Latin grammar) had themselves become standard teaching texts and also the basis for numerous pagan and Christian commentaries. Donatus’ reputation derived partly from the fact that his grammars were classroom-friendly and that he had been Jerome’s teacher. Priscian’s more elaborate grammar (c. 500 ce) would only achieve prominence in later, more philosophical contexts. From 100 bce on, grammatical commentaries in the West were a necessary part of the pedagogical discourse on language. Most fourth- to seventh-century commentaries on Donatus were composed by North African or provincial grammarians and scholars, teaching outside the Roman metropole. Grammatical commentaries often used the same exegetical strategies which were applied to sacred texts, especially the Bible: etymology, definition, gloss, and narrative explanation. Sacred text and grammar were structural mirrors in commentary discourses.

Grammatical commentaries served many purposes. Commentaries on Donatus solidified doctrine and descriptions by explaining or recasting his definitions, metalanguage, and paradigms for later generations. But his authority was not fixed in commentary discourse. Some grammarians took issue with his descriptions or examples and revised his grammatical descriptions or updated his accounts of fourth-century Latin. For example, the North African Pompeius (fifth c. ce) embedded snippets (lemmata) of Donatus’ grammar in his commentary and often corrected or amended Donatus’ account. The commentator repeats and supplements the Donatus base text. Pompeius strongly implied that ideals of elegant or correct Latin (latinitas) should be regarded as socially constructed and temporally bound. He criticized contemporary grammarians for maintaining some ancient writers (antiqui) as exemplars. Linguistic authority, he argued, derives from current custom and convention as much as from ancient authority or rules. Despite Cicero’s great auctoritas, his use of syllable repetition (homoteleuton) does not accord with contemporary usage: “This is completely archaic, no one does this now; if anyone were to do it, he would be laughed at” (Keil Reference Keil1961: v, 304).

Christian grammarians mounted similar arguments against Donatus’ authority, based on current usage and Latin scripture, but their commentaries had a different purpose. In the fourth century Roman education and literacy programs were in crisis. The inherited system of Roman schooling relied on grammars and commentaries keyed to earlier Latin usage and the writings of pagan Roman authors, especially Virgil, Cicero, and Terence. For many, the Church represented the continuity of the late Empire, and Roman and western bishops were sometimes seen as defenders of Romanitas, that is, the ideals of the republic and early empire, especially humility, duty, piety, and civic responsibility. The Church had a clear stake in the future of grammatical and literacy education. But Christian literacy required a different set of base and commentary texts. In the sixth century, Cassiodorus and Pope Gregory I argued that the words of scripture and other holy writing should not be made to conform to Donatus’ rules and ordinary human language usage (cf. Chin Reference Chin2008). Augustine, however, argued differently. Christians should adapt secular Latin literacy to the goals and beliefs of the Church, taking the “gold out of Egypt” and carrying it to the Promised Land of Christian literacy. Donatus’ grammatical writings were recontextualized for new Christian Latin readers, especially in monastic situations. Most Christian grammarians recognized the need for ‘good’ Latin usage, but in light of Augustine’s program they modified received linguistic ideas to accommodate the received language of scripture and new vocabulary and grammatical forms (e.g., the infinitive) used by provincial Latin speakers. Language ideals and concepts were remade in new Christian cultural practices and reflected in commentary discourse.

From Augustine to Bede (eighth c. ce) to Smaragdus (ninth c. ce), grammarians Christianized and updated Latin grammar in various ways. Bede produced a new account of schemes and tropes using Latin biblical examples of rhetorical figures. Smaragdus’ commentary on Donatus substituted scriptural for classical citations to exemplify Latin grammatical rules: “I have adorned its pages with verses from the Holy Scripture, with the intention of pouring out for my reader a pleasant draught of the Liberal Arts and the Scriptures, so that he may come to grasp the discipline of grammar and the sense of the Holy Scripture side by side” (1986: 1). Boniface, Smaragdus, and other grammarians expanded Donatus by describing new Latin syntax and vocabulary drawn from ecclesiastical organization or Hebrew: ecclesia ‘church,’ episcopus ‘bishop,’ cherub (“For Cherub in the book of Ezekiel is considered to be in the neuter gender”; Smaragdus Reference Löfstedt1986: 61). These commentaries constructed an intertextual grammatical discourse combining ancient and contemporary cultures. Donatus’ grammars remained the texts of reference for grammatical discourse until the Carolingian period, when an abbreviated version of Priscian’s Institutiones grammaticae (1961) began to replace them. The grammarian as adaptive commentator was central to Latin literacy as a sacred and cultural practice.

Ninth-century grammarians such as Alcuin began to supplement their Donatus commentaries with philosophical definitions and examples from Priscian’s Institutiones, and thus extended the range of linguistic discourse, paving the way for later medieval philosophers of language. After 1100, university scholars began applying new philosophical terms and definitions to the study of the language arts and theology. In the twelfth and thirteenth centuries Peter Helias, Peter Abelard, Roger Bacon, and Robert Kilwardby wrote commentaries on Donatus and Priscian or philosophies of language using Aristotelian categories and logical analysis to probe further into the theory of linguistic forms, word classes, and syntactic government.

Commentary discourse was also foundational in Arabic and Indian grammatical traditions, but with different emphases. Sībawayhi’s grammar (eighth c. ce) quickly came to dominate Arabic scholarship on language, presenting a tripartite account of syntax (noun, verb, particle) based on the language of the Qur’ān and pre-Islamic poetry and emphasizing linguistic rather than extralinguistic features. Subsequent commentaries on Sībawayhi’s Kitāb (Book) introduced semantic, pragmatic, and variationist approaches to grammar. Given that the Qur’ān formed the linguistic basis for Arabic grammar, commentaries on Sībawayhi often adopted a strong disciplinary stance emphasizing the autonomy of grammatical discourse.

In Khandehar, Punjab, Pāṇini’s detailed and innovative Aṣṭādhyāyī (The Eight Chapters, c. late fourth c. bce) superseded earlier Sanskrit grammars to become the authoritative account of classical Sanskrit and the basis for subsequent commentaries. Just as Latin grammarians noted differences between ancient and contemporary usage, Pāṇini (1962) noted differences between earlier Vedic usage (Brahman) and contemporary speech (Prakrit). Similar to the linguistic procedures used by Arabic, Roman, and Christian grammarians, he regarded sacred and poetic language, this time from the Vedas, as exemplary. For Pāṇini, literacy and grammatical knowledge helped maintain the continuity of the Vedic textual tradition against the forgetfulness of presentism. The term ‘Sanskrit’ was derived from samskrta (‘polished, purified’), and Pāṇini was keen to explain and foster correct Vedic forms in speech and writing. If a reader used correct speech forms (sabda) and had a general knowledge of Sanskrit grammar, he had a better chance of attaining religious merit (dharma).

Early Hindu study of language was closely connected with Sanskrit literacy, specifically, reading and interpreting the Vedas, properly pronouncing ritual formulas, and standardizing ancient Sanskrit. By Pāṇini’s time and certainly by 300 bce, Sanskrit had become a second language for most of South Asia, except maybe some members of the Brahman caste and Vedic seers. The more standardized the Vedic texts in ancient Sanskrit became, the more written Sanskrit differed from the many Prakrit (lower-caste) varieties and regional dialects of everyday Indian speech. Sanskrit grammar was implicated in religious epistemology and ethics while contributing to the stratification of speech in South Asia. By systematizing the linguistic structure and ideal performance of sacred Sanskrit texts, Pāṇini’s grammar effectively created a ‘Classic Sanskrit’ located in the historical past but realized in a textual present. Pāṇini’s Aṣṭādhyāyī formalized grammar for the performance and understanding of written Vedic Sanskrit, while also being used to teach ancient Sanskrit as a second language (see Cardona Reference Cardona and Auroux2000).

Pāṇini’s earliest extant commentators, Kātyāyana and Patañjali (second c. bce), explained how best to apply Pāṇini’s rules for pronunciation, word formation and constructions. Patañjali expanded on Pāṇini’s grammatical system by introducing the concepts of the word and the phoneme as independent and eternal linguistic entities. Patañjali’s commentary also introduced the concept of sphoṭa, which later ignited a debate in Indian grammar about semantics, ontology, and linguistic form. Bhartṛhari (seventh c. ce) in his commentary on Pāṇini refined sphoṭa theory. The eternal sphoṭa (meaning, lit. ‘bursting forth’) accessed through speech is identical with the permanent elements of language, that ‘something’ in language’s ontology which causes articulated sounds and expresses meaning, similar to the Stoics’ lekton. Sphoṭa identifies something more than intention and is connected to the eternal and indivisible Brahman. But Sanskrit grammarians were teachers of language and literacy, not religious leaders. They were interested in how language served ritual and expressed belief. They divided sacred discourse into discrete verbal units (words, alphabetic symbols) and produced linguistic objects in grammatical discourse. Some later commentators followed the order of Pāṇini’s text; others rearranged his rules into thematic groups.

Some Sanskrit scholars took issue with the grammatical model for religious purity. Mīmāṃsāka philosophers (c. sixth–seventh cc. ce), adopting a strong argument for linguistic realism and linguistic universalism, claimed that one can understand the Sanskrit of the Vedas without necessarily studying grammar. The language of the Vedic texts is eternal. Language as such and properly pronounced ritual language in the Vedas have the power to bring into being the states of affairs of which they speak. According to Kumārila (late seventh c. ce), in his Ślokavārttika Pratyakasūtra, “as the words are uttered in a sentence, each word performs its task of expressing its meaning, and the sentence is the summation of these meanings.” Sentences do not exist as such but are mediating constructions projected by grammarians onto Vedic language (Brough Reference Brough and Staal1972: 415; cf. Arnold Reference Arnold2005). The cognitions or understandings arising from perceptions of linguistic or material phenomena do not in themselves have linguistic form.

Because Pāṇini’s theory of morphophonology and syntax drew on religious epistemology, his commentators struggled with whether his analysis and rules were consistent with religious principles or relied on an autonomous language system. Pāṇini’s concept of the Zero element (lopa, the ‘non-perception’ [a darśanam] of an element in speech production) was especially influential. The Zero element is an unfilled slot in the word or sentence. In Bhuddist and Hindu epistemology and ontology, what is absent (or silent) in a system is said to exist as much as what is present in the system, a position not unlike the Stoics’ and one which resonates with Jakobson’s theory of the ‘zero sign’ (1939a/1984). The concept of the Zero element reflects the highly formal nature of Pāṇini’s grammatical theory. Because Pāṇini’s account of correct speech was keyed to reading ancient Sanskrit aloud, written diacritics were important for guiding the reader’s vocalization of the text. Based on semantic context or phonology, the Zero element can replace a word or affix in specific contexts and thus has a positive function in Pāṇinian grammar. The Zero element indicates where in the word string the derivational context deletes an expected or systemically predicted element from the surface representation. Some commentators believed the Zero element in Pāṇini’s grammatical theory reflected Bhuddist spirituality.

Traditional Chinese linguistics presents a very different history from that in Western Europe or South Asia. Lexicography, glosses, and pronunciation texts dominated Chinese scholarship on language until the publication in 1311/1324 ce of the first grammar of Chinese, Lu Yiwei’s Yuzhu (Grammatical Particles). Rhyme and dialect dictionaries on classic and sacred Chinese texts were compiled not by grammarians but by poets, bureaucrats, and Bhuddist monks. Nonetheless, these texts reveal a growing understanding of Chinese syllable and word structure and preserve the variety of pronunciations among ancient Chinese speakers.

Linguistic Purity

The quest for a pure, correct, or original language permeated ancient and medieval grammatical traditions. Pāṇini’s grammar and subsequent commentaries sought to maintain the correctness of the Vedic texts and their oral recitation by imagining a semi-autonomous written sacred language as superior to ordinary speech. Arab grammarians were concerned to explain the linguistic correctness of the language of the Qur’ān. With great linguistic descriptive sophistication, these commentators effectively contributed to conservative and elitist cultural approaches to language policy. In Western Europe, Varro (early second c. bce) and early medieval Christian grammarians modulated between linguistic correctness and regularity, on the one hand, and legitimizing variation in spoken and written usage, on the other. Grammatica became a master discourse, the foundation of literacy education, and the arbiter of linguistic and culture values.

Much of Varro’s De lingua Latina (1977–9) has been lost, but apparently the text outlined Latin declensions, conjugations, and semantic networks, many of which became the core of later Latin grammatical theory and practice. Varro theorized the radix (‘root’) as the basis for morphology and distinguished, for the first time in Hellenistic–Roman grammar, between inflectional and derivational morphology. Varro borrowed the Stoics’ analysis of verb aspect (assertive, hypothetical, conditional) and Pythagorean models of mathematics and augmented the Latin verb paradigm to include the future perfect indicative. He rejected the absolute opposition between grammatical analogy and anomaly in favor of degrees of regularity. Varro’s etymological analyses gave equal weight to Greek and Latin word origins, and he collected obsolete or archaic words from old texts to explain contemporary forms, thus embracing diachronic understanding of the Latin lexicon and usage. The ‘grammarian,’ unlike the antiquarian or the dialectician, compares words, the key elements of language, within classes to determine the degree of regularity in language usage. Etymological analysis focuses on declinatio voluntaris (‘derivational morphology’) and reveals the limits of linguistic systematicity and regularity marked by declinatio naturalis (‘inflectional morphology’). What Varro’s model grammarian does not do is provide many literary examples or illustrations from Latin grammar. The grammarian was primarily an investigator of word formation, and Latin word formation was as regular as other natural processes.

Quintilian (first c. bce) extended the role of the grammarian to include Ciceronian rhetoric and the model of the ‘good’ (bonus) person speaking and writing well. His Institutio oratoria (Oratorical Training, 1920–2) combined technical grammar with linguistic standards and interpretation of literary texts in a comprehensive pedagogy which formed the basis for later Roman grammar education. Quintilian’s literacy pedagogy foregrounded grammar as a normative and normalizing discourse (latinitas) and an ethics. Although he recognized that individual utterances occur in specific contexts and intentionally, latinitas overrides rhetorical contexts or discursive conventions which might motivate nonstandard usage. Quintilian’s grammatical norms are written norms, encoded in correct spelling, morphology, and syntax and exemplified by the best Latin authors.

Throughout the premodern world, grammarians developed the ideals of hellenismos, latinitas, and „arabiyya (Classical Arabic) as counterstrategies to the perceived or assumed decline in standards of usage, especially written usage, and to anxieties about language mixing and borrowing as a result of conquest, trade, intermarriage, evangelization, migration, and other contexts of language contact. Aristotle had acknowledged that people pronounce and write words differently, but he regarded these differences as epiphenomena masking underlying cognitive and affective universals among human language users. Others treated language differences in more parochial or lethal ways. In Judges 12:6, the dialectal difference between sibboleth and shibboleth among Semitic speakers had life and death consequences. In the early Middle Ages, some monastic grammarians and literacy experts wondered whether the newly Christianized peoples in the non-Roman world were corrupting Latin norms. Alcuin’s revision of the Biblical Latin text prescribed one-letter–one-sound pronunciation as a governing linguistic principle of the Church universal in liturgical contexts. Throughout the Middle Ages, Church leaders regularly worried whether different pronunciations of Latin words in ritual discourse invalidated a rite’s efficacy.

Arab grammarians and rulers were concerned that dialect differences were corrupting the originary, sacred language. According to Arabic grammarians, Bedouin speech was a continuation of „arabiyya, the original Arabic given by Allah. In the early history of Arabic grammar from Abu l-Aswad (mythic founder of Arab grammar) to the early Basrans to Sībawayhi (eighth c. bce) social and intellectual elites lamented that ‘correct’ Arabic grammar and usage were declining as the Arabic empire spread across the Mediterranean. Some attributed the decline to newly Islamicized non-Arabs, a linguistic consequence of missionary or military success. As in the wider Roman world, the expansion of Islamic culture and Arab political and religious order created mixed language situations and linguistic anxiety. What is ‘correct’ Arabic? Who are proper Arabic speakers? By the eighth century ce, Arab grammarians were compiling dictionaries and word lists and composing grammatical commentaries and dialogues to codify the Arabic language as part of religious literacy and language policy.

Unlike most Christian traditions, Islam does not believe the Qur’ān can be truly or effectively translated. Instead, ‘renderings’ of the Qur’ān for Muslims who do not read or speak Arabic are bilingual texts, with juxtaposed versions of the Classical Arabic text and explanatory ‘paraphrases’ in the local target language. Qur’ānic renderings and commentaries use similar page layouts, so grammatical and exegetical discourses are textually co-equal while the sacred text remains inviolable. Nonetheless, Arabic script was and is considered an aesthetic object which embellishes the language and the page. Today, the Qur’ān text is still used to decorate mosques, markets, and Islamic homes. al-Kitāb (The Book) is the alternate title of the Qur’ān and the title of Sībawayhi’s grammar, suggesting the link between sacred and intellectual discourses in early Arabic culture.

The Latin/Christian and Arabic/Muslim traditions were embedded within religious structures of literacy, ritual, and intellectual discourse. The imposition of a literate standard based on a sacred text created multilingual and di- or triglossic situations where different versions of Arabic or Latin had sacred status as divinely revealed language or were the normative discourses of ritual and religious literacy or were many peoples’ everyday language. In Germanic regions, the situation was a combination of diglossia and bilingualism. After the Carolingian Bible reforms, the European sociolinguistic situation became looser, less clearly stratified. Latin was taught as an elite and foreign language, and vernaculars were competing with Latin grammatical norms. Similarly, the Arabic spoken and written by Muslims and Jews in post-1100 Andalusia varied from Qur’ānic language and the Arabic described by Sībawayhi.

Disciplinarity and Boundary Work

The conflict between faith and reason was played out in both Christian and Muslim grammatical discourses. Arab scholars had appropriated, translated, and adapted a range of Greek philosophical, medical, and natural science texts, and Greek rationalism had a significant influence on Arabic grammar. Also, thanks to Arab scholars, Aristotle’s writings, newly introduced into western universities after 1100, profoundly influenced later medieval linguistics. The Mu’tazilites, who flourished in Basra and Baghdad in the eighth and ninth centuries ce, believed that human reason is necessary for, and capable of, understanding divine truths. God’s creative speech presented in the Qur’ān represents His divine intentions not directly but in mediated form. The Qur’ān text and individuals’ speech can be investigated in reasoned analysis to determine the speakers’ motives based on free will. The Mu’tazilites also argued that the Qur’ān, as an intentional linguistic artifact, was created rather than revealed. Their position was endorsed by the caliph al-Maʾmūn as official doctrine, but its danger to received clerical orthodoxy soon became apparent. After 847 ce the decree was rescinded by the caliph al-Mutawakkil, and the Mu’tazilites were banned from teaching in theological schools (cf. Versteegh Reference Versteegh and Auroux2000: 301).

Nonetheless, Arab grammarians continued to adapt aspects of Greek philosophy to develop their linguistic ideas, partly due to the prestige of Byzantine schools and Greek discourse in Asia Minor. Most Arab grammarians bracketed or ignored the divine motivation of speech and focused instead on developing theories of linguistic autonomy, chiefly through analogy, logical categories, and causal analysis. az-Zağğāğī (fourth c. ce) and others adopted a tripartite word class division (noun, verb, particle) which perhaps suggests some influence from Aristotle’s tripartite distinction (onoma, rhema, synthesmos) based on propositional logic and predication rather than on a model of six or eight word classes from technical grammar. az-Zağğāğī distinguished between the professional grammarian’s understanding of Arabic and native speakers’ intuitive knowledge which enables them to speak and write the language properly or effectively. Knowing the basis of naḥw (language) does not mean one has a complete knowledge of the lexicon (1995: 152–3). az-Zağğāğī privileged grammatical study over lexical study, construction over reference, both for the native speaker and the grammarian. He also challenged the status of elite speech, asserting that everyday Arabic speech is more representative of the Arabic language than is rare, elite, sacred, or specialized usage (ġarib ‘rare words, glosses’). az-Zağğāğī asserts that studying infrequent words used only by elite readers is not as important as studying the Arabic speech which everyone uses or should use. People of different social status (scholars, social elite, peasants) have different linguistic abilities and different concepts of usage, and no one can know all the words in a language. Not everyone writes poems or studies grammar, but everyone speaks Arabic: “with respect to normal everyday speech, with the exclusion of irregular and rare expressions, they [the Arabs] are all equal” (1995: 153).

az-Zağğāğī identified linguistic causes (’ilal an-nahw) as the products of grammatical discourse and theorizing: “linguistic causes are not obligatory but they are inferred as rules and standards” (1995: 87). While “necessary causes” govern the existence of entities in the world as entities, “linguistic causes” are constructed by grammarians from observed speech data. Unlike native speakers’ intuitive knowledge, grammarians’ causal argumentation presents plausible accounts of why and how particular linguistic constructions and usages come about. For the Mu’tazilites and az-Zağğāğī, grammar is a rational science, separate from theology. In grammar, nothing is accepted until demonstrated with reason and evidence (cf. Versteegh Reference Versteegh1995: 22–3). As reasoned discourse, grammatical analysis can be corrected, improved, and clarified through further research. az-Zağğāğī and others argued for grammar’s disciplinary autonomy and at the same time for measuring grammarians’ accounts of language against perceived everyday usage.

Between 1120 and 1400 ce, European theorists posed similar questions, for the most part in universities as part of grammatica. Grammar was properly concerned with underlying structures common to all languages. Peter Helias, Roger Bacon, and the Modistae based their theories of language on a wider discussion of cognitive, semantic, and ontological universals. Abelard and Ockham argued that universal categories (motion, quantity) were physically real but understood only through the mental concepts we have for them in language. Like some Arab grammarians, some medieval European philosophical grammarians enlarged the domain of autonomous language theory to include sociolinguistic variation and pragmatic contexts, views which clashed with both language purists and guardians of religious orthodoxy. Working closely with Aristotle and Priscian, Abelard’s linguistic analysis of propositions and truth claims (e.g., different kinds of hypotheticals, how clause meanings differ depending on the meaning of the conjunction cum [‘when, while, because, although’]) helped redraw disciplinary boundaries so as to change both grammar and logic (cf. Amsler Reference Amsler and Brown2006).

Translation and Mixed Language

As noted above, the status of the Qur’ān as revealed language prohibited until recently any translation of the original Arabic into another language. In the early Islamic period, scholars ‘rendered’ the Qur’ān by juxtaposing the sacred text with paraphrase, commentary, and exegesis in contemporary Arabic. Any grammarian who described non-Qur’ānic Arabic was explicitly departing from the normative tradition of Arabic grammatical discourse.

Since the fourth century ce, Christian translations of the Bible (from Hebrew to Greek to Latin to modern vernaculars) have been a productive site for exploring linguistic theories and language attitudes. In medieval Europe, Latin was the principal language of scripture, learning, and official record, but Latin existed in a multilingual discursive space along with vernaculars and earlier Greek and Hebrew versions of scripture. Christian scholars and authorities engaged in debates regarding the appropriate language for scripture, sometimes with violent outcomes. In late fourteenth- and early fifteenth-century England, proponents of translating scripture into the vernacular, following the writings and teachings of John Wycliffe, were persecuted, condemned, and even burned as heretics. Most premodern translation debates centered on who should be able or allowed to read scripture or elite texts and in what language and how. Translation practices posed questions about the comparative structures of languages, the merits of literal, paraphrasing, and sense-for-sense translation, and whether syntax and semantics were universal or context-sensitive. Hebrew words in particular often challenged the strategy of appropriating borrowed foreign words with Latin morphology. Bible scholars and translators were also concerned with language status. Vernacular translations challenged the authority of Latin as the language of scripture by historicizing scriptural writing within a chain of linguistic significations.

Of course, Latin itself had once been the vernacular (vulgate) language translating the Hebrew or Greek scriptures for western audiences. Jerome (fourth c. ce) had questioned the status of the existing Christian Latin Bible and proposed a return to veritas Hebraica (‘truth of the Hebrew’) to produce a more accurate text for Christians. Using the Torah and Septuagint Bible, his revised Vulgate translation became the standard Latin text throughout the Middle Ages. Jerome’s retranslation of the Latin Bible created a new originary text and a new linguistic archive from which Christian grammarians drew their examples.

Ælfric (tenth c . ce) compared Latin and Old English to determine whether the vernacular could capture scriptural meaning. In his Preface to Genesis (1922) he worried that the spiritual complexities of Latin scripture might be lost to non-Latin speakers and readers or misunderstood by poorly educated priests who relied on Old English translations. Ælfric and other grammarians argued that borrowed Hebrew or Greek words should be regarded as indeclinables and used in Latin texts without Latin or vernacular morphology. However, the writer of the parsing grammar known as Beatus quid est (eleventh c. bce), a compilation of Donatus, Priscian, and other Latin grammatical authorities, enlarged the forms for the first noun declension in Latin as -a (Maria), -as (Andreas), and -am (Abraham). Since these names appear in the Latin scriptures, they might be taken as indeclinable words in sacred text, but the grammarian instead appropriated their final letters/sounds to the Latin noun first declension, thus rendering the names ‘acceptable’ as first declension nouns, although indeclinable (without productive morphology) beyond their instantiations: -a (nom. sg.), -am (acc. sg.), -as (acc. pl.).

In late fourteenth-century England, scholars openly debated the merits of vernacular scripture translation and connected linguistic and religious discourses in provocative ways. But the growing threat of Lollardy (derived from Wycliffe’s writing) prompted Thomas Arundel, Archbishop of Canterbury, to close down the debate. Arundel’s Constitutions (c. 1407–8), stipulated that a lay person who owned or used vernacular scripture or even a Latin Bible risked being accused of heresy and possibly excommunicated or executed. Despite the hostility toward Lollards and vernacular translation, some scholars used sophisticated linguistic analysis to underwrite religious polemics. The Prologue to the Wycliffite Bible (c. 1395–7, often but erroneously attributed to John Purvey, Wycliffe’s secretary) deployed cross-language clause analysis to defend the translation method “according to the meaning (sentence) and not merely according to the words” (my modernization of the Middle English; Hudson Reference Hudson1978: 68). In the Prologue, sentence refers to the plain, obvious, or idiomatic meaning of the text in the target language, whereas “after [according to] the words” means keeping the word order and vocabulary of the Latin text in English, which the writer argued produced a stilted Latinized English, misleading sentences, and sometimes nonsense. “Translating into English,” he wrote, means “many transformations [from Latin to English] are necessary in order to make the meaning plain” (Hudson Reference Hudson1978: 68). The translator must use different clauses in idiomatic English to overcome obscure or confusing constructions produced when Latin is translated into English “word for word.” The translator is responsible for delivering proper textual meaning to the vernacular reader. Once again, the reader’s spiritual condition is the ground for linguistic argument.

The Prologue writer illustrates how to “resolve” Latin syntax into English constructions with examples drawn from grammar school vulgaria exercises, designed to teach vernacular speakers how to compose Latin prose. Often, the Latin is rendered into English by adding subordinators or implied subjects or objects based on pragmatic context and inference. For example, the Latin ablative absolute construction, a reduced past participle clause, does not explicitly mark the semantic relation of the subordinate construction to the matrix (main) clause. English, however, uses more explicit clause joiners and temporal or causal subordinators to link clauses together. The writer explains, “the Latin ablative absolute construction should usually be transformed (resoluid) into these three conjunctions, with the linked verb, the while, for, if, as the grammarians say,” depending on the meaning of the text as interpreted by the reader. Thus, different interpretations produce different translations of the Latin ablative absolute construction: “while the teacher reads,” “because the teacher reads,” “when the teacher read,” or “after the teacher read,” each adding English words to clarify the Latin sense. “Sometimes,” according to the writer, the ablative absolute “may well be resolved into an independent phrase [i.e., clause] by supplying a [English] verb and conjunction” (Hudson Reference Hudson1978: 68). The writer insists that the Latin “sentence” (textual meaning and interpretive context), construed by pragmatic inferencing, determines which English subordinator the translator uses.

Translation is interpretive, as the Prologue writer implies. He argues that English translation recovers the original scriptural sense which the Latin text sometimes obscures: “Our Latin Bibles often disagree with the Hebrew of the Old Testament, as one may see from the commentaries of Jerome and Nicholas of Lyra and other expositors. In such places I have made a note in the margin, giving the true sense of the Hebrew, and how it is interpreted by these commentators” (Hudson Reference Hudson1978: 69). Lollards and other heretical groups in medieval Europe were often accused by the Church of usurping the clergy’s authority to read, interpret, and preach on scripture and doctrine. The mostly lay Lollards were criticized for not being properly educated in reading scripture or for reading too literally. But the Prologue writer turns literalism on its head. His grammatical analysis of Latin–English translation shows how an overly literal rendering of Latin into English is not only unidiomatic but semantically misleading. Using contrastive grammatical analysis, he suggests that informed, flexible vernacular translations are spiritually equivalent to fourth-century Latin. At different times, each is a source language and each is a target language in a translation chain.

Linguistic thinking emerged in the ancient and medieval worlds amid various multilingualisms and religious and philosophical innovations. Although Greek, Latin, and Classical Arabic were authoritative languages for establishing linguistic and literate norms (hellenismos, latinitas, „arabiiyya), ‘vernacularity’ changed or destabilized dominant language ideology (cf. Versteegh Reference Versteegh1984). Language contact shows how translinguistic consciousness can be creative and transgressive. In the fifteenth-century English morality play Mankynde (c. 1450) language mixing was both performative and disruptive. The vice figure Mischief and his henchmen arrive on stage and immediately enlarge language and grammaticality by mixing Latin and English with their body talk and risqué speech. Mischief combines English roots with Latin noun morphology to create a hybrid language, understandable to a general medieval audience and challenging the clergy’s moral and linguistic authority from a materialist viewpoint: “Corn servit bredibus, chaffe horsibus, straw fyrybusque” (‘corn serves to make bread, chaff to feed horses, straw to start a fire’) (Coldewey Reference Coldewey1993: ll. 56–7). The dramatic characters use this ‘Englysch Laten’ as well as cross-language puns (sic-seek) and scatological speech (holic > hol lic) to disrupt, if only temporarily, the dominant social and religious order. Philosophical grammar based on Aristotle’s categories and realist semiotic theory sought to ground language theory in universals. Lollard vernacular translations and the vices’ hybrid Latin–English reveal how multilingual practices shaped and challenged linguistic consciousness in the social habitus (cf. Amsler Reference Amsler2011: 282–301).

Conclusion

Premodern linguistic thinking was situated within other cultural practices – philosophy, theology, translation, education. The role of the grammarian or linguist was structured within various institutional contexts. But in Christian, Islamic, and Hindu cultures, philosophers, teachers, and cultural critics constructed the grammarian or linguist as a particular kind of intellectual within a multilingual society. Establishing the parameters for ‘grammar’ or ‘linguistics’ was part of a new set of cultural practices leading to the formation of language study as a semi-autonomous field within schooling and other institutions. However, the ‘science of language’ was never entirely separated from sociocultural activities because premodern grammar was always about language as a social practice. The perceived similarities and differences among languages, the tension between standards and usage, the universality of language acquisition, and the cultural power of writing and literacy continued to shape how linguists saw themselves and how others saw them.