Grice as modista

Chapter 8 ‘General’ or ‘universal’ grammar: from the modistae to Chomsky The background to general grammar Just as all men do not have the same orthography, so all men do not have the same speech sounds; but the mental experiences, which these directly symbolize, are the same for all, as also are those things of which our experiences are images. (Aristotle On Interpretation 16a 3) The quote above from Aristotle’s On Interpretation is the seed for ‘general’ (? Latin genus, generis) grammar, which takes the view that the fundamental rules of grammar are common to the genus language, a logically necessary component of all languages. Particular languages manifest species in their accidentia (secondary grammatical categories typically expressed by inflections). It is most unlikely that Aristotle himself entertained a notion of general grammar, but what he wrote gives grounds for it. The two facts that (i) Donatus seems to have adapted a grammatical description of Greek from the Techn? Grammatik? attributed to Dionysius Thrax and (ii) Priscian explicitly adapted the Suntaxis of Apollonius Dyscolus to Latin, are not evidence that either Donatus or Priscian were general grammarians. They believed that Latin was the daughter of Greek, an idea that lived on into the nineteenth century. It is a view compatible and consistent with the notion of general grammar. The Christian grammarians of the early middle ages firmly believed that Hebrew, the original language of the Old Testament, was the mother of Greek – the language of the New Testament;1 and they knew that the Romance vernaculars derived from Latin. So the notion of a linguistic genus, a family of languages, was already well established. Against this background a theoretical interest in language structure began to develop in the carolingian period. Definitions were the starting point in describing parts of speech, and the carolingians identified up to six types: substance, sound, species, number, properties, and etymology. They also looked at what a term has in common with other things characteristic of the genus, its unique characteristics (differentia), and the concatenations it enters into. There were attempts to map the eight parts of speech onto Aristotle’s ten categories (Table 3.2). For instance, a collective noun such as crowd falls within two Aristotelian categories: being a noun it denotes substance; and being collective it is relational. Isidore of Seville had offered one sentence exemplifying all ten Aristotelian categories, and it was copied by Alcuin and Sedulius among others. A complete sentence using all these [categories] is Augustine [substance], a great orator [quantity and quality], the son of that person [relation], standing in the temple [posture and place], today [time], adorned with a headband [state], having a dispute [action], gets tired [affection]. (Etym. II.xxvi.11) 1. It is now generally agreed that Jesus spoke Aramaic, a Semitic language. 162 The Western Classical Tradition in Linguistics Dialectic is prior to grammar; a child needs to know what homo denotes before learning its accidence. A general kind of subjection takes place in the names of all things. No creature can come to human knowledge without a name. It can be either a proper name or an appellative. Hence there is no creature, either corporeal or incorporeal, which is not subjected to its name. That is why Priscian said ‘each subjected body or thing’; that is, to their names. Whatever is named by a nomen, is subjected to that nomen. Thus, general subjection concerns the subjection of all things, corporeal and incorporeal, to their names. And it is this kind of general subjection that Priscian is now talking about. (Attributed to Johannes Scottus Eriugena, ninth century; quoted in Luhtala 1996: 66) [T]he noun is a conception of the mind for signifying the substance of things. Similarly the verb is a conception of the mind for signifying action and undergoing the action of substances. (ibid. 69) Speculations about the ontology of parts of speech began during the carolingian renaissance and persisted throughout the rest of the middle ages. Remigius of Auxerre (d. 908) wrote: Why was the verb invented? But this is the difference between the signification of the noun and of the verb, that the noun signifies that which exists and is permanent, that is substance, whereas the verb signifies the unstable, transient motion of the substance, which is in no way permanent. This motion is understood as twofold, as pertaining to the agent and to the patient. When I say amo [“I love”], I express the motion of a substance as well as its undergoing action. But when I say the noun amor, amoris I understand the cause of the motion, since love [amor] is the cause that makes man love [ut homo amet]. (Remigius of Auxerre 1962-65, IV.184.12) According to the medieval mind, motion and affect/effect are implicit in the concept of transitivity: the nominative causes an affect/effect that goes to the oblique. Religious interest in the concept of love caused much consideration of it from Alcuin on, provoked by sentences like (1). (1) Amo Deum. “I love God.” (1) creates a paradox because, if the object of love is acted upon, then humans can act on God by loving God, yet this was inconceivable to the religious mind. The priority of the noun over other parts of speech goes way back through the Western Classical Tradition. Priscian (Inst. XVII: 14) had got it from Apollonius (Synt. II: 4) and the medievals got it from Priscian. A consequence is the developing interest in sentence (clause) structure that was born in the work of Apollonius, adopted by Priscian, and nurtured by medieval grammarians. Supposedly, verbs always joined to a nominative; therefore, the verb presupposes a subject (see Priscian 527, XIII: 28; XIV: 15; XVII: 14). Apollonius had described this as an independent construction marked by concord. According to the Stoics, Apollonius, Priscian, and the medievals, the relation is intrinsically intransitive. Where the intransitive construction is what the Stoics called ‘complete’, Priscian referred to it being ‘absolute’. Where there is also an oblique case, the action expressed by the verb proceeds from nominative to oblique (Priscian 527, XIII: 23): it was not the verb that was seen to be transitive, but the relationship between the nominals. Thus, in (2), Socrates percutit is described after the ninth century as intransitio actus “active intransitive”. ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 163 (2) Socrates percutit Platonem. “Socrates struck Plato.” According to the Apollonian criterion, because there is no concord between verb and oblique they do not compose an independent construction; however, percutit Platonem is transitive because the verb implies a third person nominative or subject. We shall see more of these structures when we look at speculative grammars. Latin is a ‘free word order language’ because intraclausal relations are morphologically marked. This, along with the frequent heavy participial constructions, makes it a difficult language to read without careful parsing. As a result, medieval readers frequently marked semantic-syntactic relations on manuscripts, often in terms of preferred word order. Adverbs were always located sentence finally, and adjectives preceded nouns; in southern Europe the preference was for SVO sequence but, strangely, in northern Europe VSO was preferred (Luhtala 1993: 185). The VSO sequence violates the standard notion that SV is ‘natural’ for all the reasons given earlier. The VO sequence, regarded as natural because the action is naturally prior to the undergoer of the action, was maintained in both traditions. The modistae or speculative grammarians Grammar seems to be permanently split into two camps: one of theoretical, philosophically oriented grammarians; the other of pedagogically oriented, and hence prescriptive, grammarians. The two camps are maintained today in theoretical and applied linguistics. In the middle ages a tradition of pedagogical grammar was set by Alexander de Villa Dei’s Doctrinale of 1199 in rhyming couplets (see Chapter 7). Throughout the period there was heavy reliance on memorization: rules were given, but there was no philosophizing about them. Then came the modistae, who focused on a theory of language structure instead of language instruction. In the twelfth century, William of Conches (c. 1080–1150) and his pupil Petrus Helias (fl. 1130–50) gathered together and systematized existing commentary on Priscian, and sought to define the nature of grammatical enquiry (Petrus Helias 1978; William of Conches 1965). Petrus Helias mentions grammars written for Greek, Hebrew, and Syriac (a branch of Aramaic); and, unlike his successors, he seemed to believe it necessary for there to be as many grammars as languages. He identified four parts to grammar: knowledge of letters, syllables, words, and sentences (orationes). Closer to Apollonius than to Priscian (though he probably did not have access to Apollonius), he defines mood as varia animi inclinatio “the changing inclination of the mind”, and resolves indicative, imperative, optative, and conjunctive (subjunctive) moods into an indicative and infinitive, cf. lege “read.IMP” has the meaning impero te legere “I.command.INDIC you read.INF” (Thomas of Erfurt 1972: 224 [55]2 ). Priscian had written: ‘The parts of speech cannot be correctly known without knowing their proper signification’ (Priscian 527, II: 17) and the grammarians of the later 2. In reference to Thomas of Erfurt, the section of his manuscript is given in square brackets; e.g. ‘Thomas of Erfurt 1972: 134 [1]’ refers to page 134 of the 1972 book and Thomas’ section 1. Thomas’ life dates are unknown but he wrote in the first decade of the fourteenth century. 164 The Western Classical Tradition in Linguistics middle ages took this particular guidance to its logical conclusion. The modistae got their name from seeking to explain the modi significandi “modes of signification” of the parts of speech, of grammatical constructions, concord, and government. [W]ishing to know about grammar [volentes habere scientiae grammaticae] we insist that it is first of all necessary to know its principles, which are the modes of signifying [modi significandi]. (Thomas of Erfurt 1972: 134 [1]) William of Conches complained in Glosae super Priscianum that Priscian’s definitions were obscure without exposition and lack functional explanations for parts of speech and their accidence (Covington 1984: 9; Law 2003: 172). Priscian did not concern himself with abstract grammatical relations but with the correct surface structure of sentences, i.e. the correct sequence of letters, the correct inflection, the correct syntactic structures. When Priscian identified the meanings of the parts of speech it was an aid to learning. The modistae, on the other hand, were much less concerned with surface grammar; they were only concerned with forms when form indicates a contrast in meaning, so they paid no attention to ‘letters’ or to regular and irregular morphology. Consequently, they ignore the morphology of case when defining nouns and verbs. They were students of grammar, not of Latin. They were interested in what, after 1964, was called ‘underlying structure’ – though, as we shall see, their conception of it was very different. The modistae flourished in northern Europe where grammar ceased to be studied solely as a key to Latin classical literature or the Bible, and became ‘speculative’. The term speculative is based on Latin speculum “mirror; image” because speculative grammars sought to mirror reality. Grammarians adopted the Aristotelian belief (quoted at the head of this chapter) that the world is the same for all human beings and that language reflects that world. Grammar was looked upon by the modistae as dependent on the structure of reality and so the rules of grammar are independent of the language in which they are expressed (Bursill-Hall 1971: 35, 331). The basis for grammar is God’s world as it is filtered through the human mind, so that grammar becomes study of the formulation of concepts and their expressibility in well-formed sentences and component structures. The modistic view of grammar led the speculative grammarians to concentrate on the universal properties of grammar instead of on grammars of individual languages. The exception is Roger Bacon (c. 1220–92). Bacon wrote grammars of Greek and Hebrew, as well as knowing some Arabic; he was interested in their practical instruction, and paid more attention to phonology than most of his contemporaries (Hovdhaugen 1990). He observed: In the Latin language, which is one, there are many dialects […] since there is a number of nations using this language. For Italians in many cases speak and write in one way, the Spanish in another, the French in a third way, the Germans in a fourth, the English in a fifth, and so forth. (Bacon 1902: 26) Bacon’s ‘grammatica vna et eadem est secundum substanciam in omnibus linguis licet accidentaliter varietur’ was much quoted as the basic principle of universal grammar; it occurs in the following context – which is obviously no universalist manifesto. Since I want to describe Greek grammar for the benefit of Latin speakers it is necessary to compare it with Latin grammar both because I speak Latin for the most part as it is necessary since the great mass does not know how to speak Greek, and because grammar is one and the ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 165 same in all languages although there are accidental variations [grammatica vna et eadem est secundum substanciam in omnibus linguis licet accidentaliter varietur], and because Latin grammar in a certain particular way is derived from Greek grammar, as Priscian testifies, and as authoritative writers on grammar openly teach. (Bacon 1902: 27) There is a quote that has been wrongly attributed to Bacon by Lyons 1968: 15f; Fromkin and Rodman 1998; and many others in cyberspace: ‘He that understands grammar in one language understands it in another, as far as are concerned the essential properties of grammar. The fact that he cannot speak or comprehend another language is due to the diversity of words and their different forms, but these are the accidental properties of grammar.’ In fact, the author is the anonymous thirteenth century modista quoted below. Bacon would not have shared his view. The fact that Bacon was not a universalist can be seen from: no Latin will be able to understand as he should the wisdom of the sacred Scripture and of [Greek] philosophy, unless he understands the languages from which they were translated. (Bacon 1928: 76) Bacon’s contemporary Robert Kilwardby (c. 1215–79) is more in line with the run of speculative grammars: Combinations of parts of speech in grammar and their consignifications are based on natural properties of the things signified and so follow the rules of nature. But differences in the vocalization of suffixes and differences in their pronunciation are not by nature, but by human design. (Quoted in Hovdhaugen 1990: 120 from Kilwardby 1976: 218) There is also the following from Boethius of Dacia3 (d. c. 1285): There is but one logic for all tongues and therefore just one grammar. […] All dialects are characterized by one grammar. The reason for this is that the whole of grammar is based on things – it cannot be a figment of the intellect, because a figment of the intellect is something to which no reality corresponds outside the mind – and because the natural properties of things are similar among all, accordingly the modes of being and the modes of understanding are similar among all those who have different dialects and accordingly the modes of signifying are similar and accordingly also the modes of construction or speaking are similar. And so the whole grammar which is in one dialect is similar to that one which is another dialect. For which mode of being and understanding and signifying and constructing or speaking can be in one dialect and not in another? This does not seem possible. [… T]he whole grammar does not differ among different dialects. It is necessarily one in species and differs only as a result of different phonetic realizations which are accidental aspects of grammar. (Adapted from Hovdhaugen 1990: 120f, quoting Boethius of Dacia 1969: 11ff) The anonymous thirteenth century modista mentioned earlier wrote: And thus all of the grammar of one language is similar to that of another, and alike in conception to it, differing only according to different ways of speaking, which are accidents of grammar. Consequently knowing the grammar of one language is knowing that of another, so far as are concerned all the essentials of grammar. The fact that one cannot speak or understand another language, is because of its different phonology and morphology [diversitatem vocum et diversas figurationes], and these are accidents of grammar. (Thurot 1964: 125 [Bibl. Nat. de Paris lat. 16297 f.131]) 3. Denmark. 166 The Western Classical Tradition in Linguistics According to Boethius, the ordinary language speaker knows the language on the basis of experience and usage but this is no qualification to teach the language; for that, an artifex “theorist” is needed who can explain the rationale behind the grammar (Boethius of Dacia 1969: 15). Apollonius would have approved. Boethius distinguishes the roles of philosopher and grammarian: speculating on things themselves, [the grammarian] is a philosopher. When, however, he joins the thing with a word [vox] making it signify and making modi intelligendi from modi essendi and voices modi significandi from modi intelligendi he becomes a grammarian. (ibid. 7) However, the anonymous modista quoted earlier thinks that grammar comes out of philosophy: It is not the grammarian but the philosopher, diligently reflecting on the proper nature of things (from which appropriate essential modes are learned from different things), who discovers grammar. (Thurot 1964: 124 [Bibl. Nat. de Paris lat. 16297 f.131]) Aristotle’s Categories, On Interpretation and Topics were already known in the west from the early sixth century on; but Metaphysics and other works seem to have been lost until the crusaders visited Constantinople in 1204.4 These ancient manuscripts came to be known as the ‘new’ logic and were taught along with the ‘old’ logic at the University of Paris and elsewhere. Aristotle’s logic, his rigour in analysis, and his categorization of the universe were adopted by the modistae in their linguistic analyses. But they pushed this method of analysis much further than he had, linking Aristotelian categories with all linguistic expressions. They looked at the world and the way it was spoken about, then speculated on the set of binary relations between language expressions and the things that they denote. The most basic binary distinction, which they attributed to Aristotle, is between things that are permanent and stable in the world, which were said to have a modus ens or entis “mode of being”, as against things that are in a state of flux, variously described as modus esse “mode of succession of states”, modus fieri “mode of becoming”, modus fluxus “mode of flux”, modus motus “mode of motion or mutation”. Following the Western Classical Tradition of Plato and Aristotle, who had split the sentence into two (onoma and rhema) on logical grounds, and yoking this to the grammatical split between substance and either action or passion made by Apollonius and Priscian, the medievals reanalysed them into the modus ens¸which was the mode of the nomen, noun or NP, and the modus esse, which was the mode of the verb (not today’s VP).5 4. This may not be so. St Anselm may have had access to Metaphysics in the eleventh century, see below. 5. This split derives from the philosophies of Parmenides and Heracleitus, respectively. Parmenides fl. 500 BCE held that the multiplicity of existing things, their changing forms and motion, are but an appearance of a single eternal reality (Being), such that ‘all is one’; thus, all claims of change (non-Being) are illogical. Heracleitus (c. 540–480 BCE) asserted that the world exists as a coherent system in which a change in one direction is ultimately balanced by a corresponding change in another. ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 167 In speculative grammar the parts of speech are motivated by (idealized) language speakers and not just found in the metalanguage of the grammarian. If modistic reasoning is accepted, this gives speculative grammar greater explanatory adequacy than any earlier grammars and most later ones as well. A pars orationis (part of speech, lexical class/ category, henceforth PARS, plural PARTES) consists of * the res “thing, referent”, * the modus intelligendi “mode of understanding”, * the vox “expression, utterance”, and * the modus significandi “mode of signifying”. The properties of the referent constitute a modus essendi “mode of being” (essendi is superordinate to ens and esse so is relevant to all parts of speech). The modus essendi is apprehended by the modus intelligendi, leading to the modus signandi “mode of signalling”, which converts the vox into the form of a dictio “word” combining form and meaning. The modus signandi gives rise to the modus significandi, which converts the vox into a PARS. Going a step towards syntax, the modus consignificandi “mode of consignification” adds syntactic function to the PARS. In speculative grammar, every PARS correlates with a res. The description just given oversimplifies. There are active and passive modes of understanding, signalling, and signifying. The idea is that the mind has to be in an active mode to perceive and apprehend, thus it is the modus intelligendi activus by which mind comprehends the modus essendi; the modus intelligendi passivus is the mental concept arising from the modus intelligendi activus and linking to the next step in the process, the modus signandi activus. The signalling process leaves off unless the resulting modus signandi passivus links to the modus significandi activus. The process is linear: modus essendi ? modus intelligendi activus ? modus intelligendi passivus ? modus signandi activus ? modus signandi passivus ? modus significandi activus ? modus significandi passivus, etc. The modus significandi passivus represents the functional or class meaning of the referent whereas the active mode reveals the property of the formal expression (vox significativa), giving rise to the ratio consignificandi “the means of achieving consignification”. There is also a ratio significandi “potential to signify” to which the modus signandi passivus links; and further back in the chain, a ratio signandi “signalling potential”. The ratio significandi creates lexical meaning; the modi significandi essentiales class meanings; and the modi significandi accidentales respectivi determine the ability of a PARS to be construed with other PARTES. The modistae sought to explain all these terms (and more) along with the relationships among them (Thomas of Erfurt 1972: 134 [1]). A thing can be signified in as many ways as it can be understood. (Boethius of Dacia 1969: 71). Within the system a single res may have different modi essendi; for example, a human may be of either masculine or feminine gender. Some properties present in the res may be ignored; for instance, one can speak of a man without reference to his colour, even though he has colour. Homo and hic homo “this man” can have the same referent, but they still have different modi significandi. Thomas of Erfurt 1972: 212 [46] compares and contrasts 168 The Western Classical Tradition in Linguistics different species6 of nouns having the same modus entis, such as albedo “whiteness” and albus “white”, against the related verb dealbo “whiten, whitewash” and its participle dealbans, which share the same modus fluxus. Each of these words has a distinct modus significandi. Following Aristotle in Categories 10a 29, Thomas claims that whiteness must exist before white or white things can exist; he might have been echoing pre-modist Anselm of Canterbury (1033–1109), too, ‘just as a man is a rational mortal animal, so white is something-having-whiteness, or what-has-whiteness’ (Anselm 2000: 160). The modus ens and modus esse are subdivided into matter versus form, yielding two more parts of speech, as in Table 8.1. Table 8.1. Modi ens and esse, matter and form. MODUS ENS MODUS ESSE MATTER noun verb FORM pronoun participle The noun denotes substance and determinate quality, e.g. lapis “[the substance] stone”. The pronoun signifies substance without quality, because the quality is ignored (Boethius of Dacia 1969: 239f); e.g. ille “it” denotes substance but is not determinate like lapis. Nouns were subclassified in the tradition of Donatus and Priscian with some metaphysical modifications such as that proper nouns (which include the indexicals hic “this” and nunc “now” (Thomas of Erfurt 1972: 156 [17])) cannot be true predicates. Like a noun, a pronoun ‘can demonstrate and refer to something either in the mind or outside the mind, either fictive or true, in reality or hypothetically’ (ibid. 208 [43]). Given that in human languages reference to something unreal is just as possible as reference to something real, the criterion that a nominal must name something with a modus ens raises the question of how the speculative grammar is to account for figmenta “fictions” such as chimaera, hircocervus “goatstag”, negationes “negations” like nihil “nothing”, and privationes “deprivations” such as caecitas “blindness”. The terms chimaera, nihil, and caecitas have active modes of signifying just like concrete nouns do. The explanation for this comes from Aristotle: ‘because contradictories outside the mind are contraries according to the mind, as stated in Metaphysica IV.9’ (Thomas of Erfurt 1972: 208 [43], see also 140 [6], 154 [16]). [I]t is admitted that deprivations are not real entities outside the mind, they are however real entities in the mind. (ibid. 140 [6]) The mind is, of course, in the real world. [B]ecause we do not understand distinct substances except those perceived by our senses, therefore we give names to them according to the properties we perceive in them, and attribute active modes of signifying to their names. […] Similarly with the names of fictions, the active modes of signifying come from the names of the parts, from those components we imagine in a chimera, then, we imagine they are taken from the head of a lion, the tail of a dragon, and so on. (ibid. 138 [5]) 6. A species is made up of genus and differentia specifica, which contrast matter and form. ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 169 There is a partial rephrasing here of Anselm, who says forms of speech allow us to talk about blindness and the like: ‘that which is signified is not something from the point of view of how things are, but only arises from the form of expression used’ (Henry 1967 §6.65).7 [S]ince no deprivations and negations are entities, it is seen that they cannot be classified under some property and therefore the active mode of signifying via the mode of being cannot originate in such cases from a property of the thing signified. […] And because deprivations, negations, and fictions are entities according to the mind, therefore they count as an entity which property is a condition of permanence from which the general mode of signifying for the noun derives. (Thomas of Erfurt 1972: 154 [16]) In speculative grammars, noun subcategories are not significantly different from what we have seen in antiquity; though there are some new twists. For instance, gender was seen as reflecting active and passive forces. Masculine gender was perceived to be active on the ground that men are active in procreation, women are allegedly passive, and this is the essence of feminine gender (ibid. 178 [27]). Martin of Dacia explains masculine lapis~lapidis “stone” as deriving from the active laedens pedem “injuring the foot”; whereas feminine petra “stone” derives from the passive pede trita “rubbed away by the foot”; see Law 2003: 176f. Neuter is neither active nor passive, and omnis generis “common gender” is one criterion for distinguishing adjectives from nouns. Another is that nouns have a modus per se stantis (because they can stand alone) whereas the adjective has a modus adiacentis and so is dependent on the noun for gender and number, e.g. in Plato albus “pale Plato”. The treatment of case is very unsatisfactory; for instance, Siger de Courtrai (c. 1280– 1341) describes the vocative as signifying a modus essendi excitati “an essential mode of excitation” (Siger de Courtrai 1977). Perhaps he saw a similarity with interjections. The modistae ignored the formal morphosyntactic characteristics of Latin case. The nominal category of person was linked to the speaker, de se, addressee, ad alium, or third person, de alio. Thomas derives the term persona from per se sonando “through oneself sounding” (Thomas of Erfurt 1972: 194 [34]). We turn now to the two modus esse PARTES, the verb and participle. The verb has the modus essendi and modus significandi of being/becoming/action, and by the modus distantia it is separated from the nominal with which it is in concord, its suppositum. The modus essendi is material and shared with the participle; the modus distantia is a formal characteristic8 that distinguishes it from the participle, which has nominal properties and so a modus indistantia from the noun and a modus adiectis to the noun. However, to the usual subcategories of the participle, Thomas rather bizarrely adds person, which derives from the associated nominal. In the ninth century Sedulius had defined the verb as A part of speech with tense, that is signifying any time whatever, which can be recognized through the motion of acting or suffering [patiendi]. The motion can either be increase or 7. Supposedly Aristotle’s Metaphysics was not available to Anselm; so either that assumption is false or Anselm came up with much the same idea. Incidentally, Thomas of Erfurt does not once refer to Aristotle by name, though he does twice refer to Philosophus “the Philosopher” (as one might refer to the Pope). 8. ‘verbum de se significat per modum distantis’ (Thomas of Erfurt 1972: 220 [52]). 170 The Western Classical Tradition in Linguistics diminution, through local or temporal motion or through states of mind. (Sedulius Scottus 1977, In Donatum Minorem 34: 4–7) The modus esse implies the passing of time, successio; but for the modistae, tense does not derive from the essence of the verb; it is an accidental mode either signifying or consignifying tense. This raises the problem of eternal propositions such as Deus est “God is/exists”. Thomas says Not every ens has successive being [esse successivum] for the being of God and of intelligence is not in flux and succession and yet we say Deus est and intelligentia est. […] They are however successive in terms of the succession of eternity [successivum successione aeternitatis]. (Thomas of Erfurt 1972: 210 [45]) Otherwise the verb subcategories are the same as those identified by Donatus and Priscian. Mood is an accidental mode of the verb that consignifies dependence of the verb on the suppositum “subject”. Thomas quotes Petrus Helius’ observation, already mentioned, that each mood can be resolved into an indicative that spells out the primary illocution together with an infinitive: lego, id est indico me legere; lege, id est, impero te legere, et sic de aliis. “I.read, that is I.say me to.read; Read! that is I.command you to.read, and similarly for others.” (ibid. 224 [55]) All subcategories of grammatical voice are semantically defined, e.g. active amo is verbum adiectivum activum; passive amor is verbum adiectivum passivum. Thomas writes, This mode of signifying is called genus [=voice], said to derive from generando because the form of one voice is generated from another; for instance the form of the passive [e.g. amor] from the active [amo]. (ibid. 230 [59]) Thomas innovatively defines mood in terms of the relation between suppositum and verb; and voice (active, passive, deponent) in terms of the relation between verb and the oblique – his term for which was significatio (in today’s terms V?NP – not VP, because the NP is obligatory). The modists’ account of conjugation, like their account of gender, is very unsatisfactory because the category is formal and not semantically based. The declinable PARTES – noun, pronoun, verb, and participle – were referred to by Siger de Courtrai 1977 as ‘magis principales’. Indeclinables are ‘partes minus principales’ whose modus essendi is the modus disponentis “mode of ordered distribution”. The PARTES therefore have one of three basic essences: being (ens), becoming (esse), or ordered distribution, i.e. entering a certain syntactic relationship with other PARTES. Hence indeclinables are classed as grammatical words defined on their syntactic functions; and they have fewer modes of meaning than declinables. Otherwise there are few innovations in the modistic discussion of adverbs, conjunctions, prepositions, and interjections. The different PARTES are differentiated by modi speciales. The adverb has a modus adiecentis which determines the modus esse. The conjunctio per vim links two unifiables (curiously described by Thomas as duo extrema “two polarities”; op.cit. 256 [76]); the coniunctio per ordinem (e.g. ‘ergo’ in Socrates currit, ergo movetur “Socrates runs, therefore he moves”) is a subordinating conjunction. The modistae distinguished prepositions from prefixes on the basis that a preposition signifies a relationship between the modus ens and modus esse: ‘the preposition was in fact invented on behalf of case forms, not just any, only accusative and ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 171 ablative’ (ibid. 262 [80]). In fact, a preposition is ‘the part of speech signifying by means of the mode of adjacency to some case form, linking it and referring it back to the act’ (ibid. 264 [81]). Speculative grammars were like their forebears in the very extensive discussion of what the modistae called ‘etymologia’ (corresponding to Priscianus Major), which means something like “word classes” – a use that filtered through to the general and rationalist grammars of the seventeenth and eighteenth centuries, and into prescriptive grammars like those of Fisher 1753 and Murray 1795. According to Covington 1984: 127, about a quarter of the manuscripts of Thomas’ treatise lack the section on syntax. Nonetheless, the modistae are far more interesting for their account of ‘diasynthetica’, which relates PARTES and their modi significandi to one another in syntax. Grammar was regarded as scientia organica (Thomas of Erfurt 1972: 312 [116]) and grammatical constructions should likewise be organic and be dealt with as a series of stages (passiones sermonis). In modistic syntax the basic syntagm is a constructio which holds between two PARTES and never more than two. A constructio is a change in accidence of the constructible; therefore constructions result from processes. In every constructible there is a modus dependendi and a modus dependentiam terminans; in other words modistic grammar is a series of binary dependency relations holding between a dependent and a terminant. Because it can stand alone, the nominal, denoting substance, is a terminant; it is also the suppositum “subject”. The verb is characteristically dependent on the suppositum (this particular idea recurs in today’s categorical grammars where the verb is “dependent on an entity to establish a truth value”). The dependent may acquire properties from the terminant: for instance, the verb gets person and number from the suppositum; an adjective gets gender, number, and case from its noun terminant. The construction that results from the application of the consignification of a verb to the suppositum is a compositio (named from Boethius’ translation of Aristotle’s sunthesis9 ). Within the compositio the verb functions as the appositum that says something about the suppositum, see Plato’s Sophist 262d. The compositio of noun and verb varies by the quality of mood. A compositio is one kind of constructio. Figure 8.1 summarizes the syntax so far. The suppositum does not have to be in the nominative case, see Table 8.2 (next page). constructio ? compositio suppositum appositum terminant dependent Plato currit modus ens modus esse Figure 8.1. Compositio. 9. This is the Roman Anicius Manlius Severinus Boethius (c. 480–524), see Boethius 1860; 1998. 172 The Western Classical Tradition in Linguistics Table 8.2. The case of the suppositum. suppositum appositum Socrates.NOM Socratis.GEN Socrati.DAT Socratem.ACC A Socrate.ABL currit interest accidit legere oportet legitur “S runs” “It is in S’s interest” “It happened to S” “It behoves S to read” “It was read by S” Figure 8.2. Types of transitives. Doubtless influenced by the fact that Aristotle’s On Interpretation 21b 9 asserts propositional equivalence between the man walks and the man is walking (see also Metaphysics 1017a 28), Thomas Aquinas (1225–74) wrote ‘nihil enim differt dicere homo convalescens est et homo convalescit’ “there is no difference between saying the man is convalescing and the man convalesces” (Sententia Metaphysicae V.ix.9, Aquinas 2005). On these grounds ‘percutit’ in Figure 8.3 and ‘currit’ in Figure 8.4 resolve into a constructio, with the terminant percussens “striking” in the former and currens “running” in the latter and a dependent est “is” in both. The example sentences in Figures 8.1 and 8.2 are all intransitive. The modistae adopted the carolingian account of transitivity, based on Priscian, constructio compositio significatio verbum suppositum a p p o s i t u m terminant dependent terminant i n t r a n s i t i v e t r a n s i t i v e terminant dependent percussens est i n t r a n s i t i v e Socrates percutit Platonem Figure 8.3. Modistic analysis of Socrates percutit Platonem. transitiva dependent terminant misereor faveo lego utor Socratis.GEN Socrati.DAT librum.ACC toga.ABL “I pity S” “I help S” “I read a book” “I wear a toga” transitiva personarum dependent terminant filius similis celer Socratis.GEN Socrati.DAT pedibus.ABL “son of S” “like S” “fleet of foot” ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 173 constructio suppositum appositum intransitive terminant intransitive dependent terminant dependent terminant dependent terminant dependent currens est Socrates albus currit bene Figure 8.5. Modistic analysis of Video legentem librum. and discussed in respect of sentences (1) and (2) (pp.155f). In intransitives a pair of constructibles refer to the same thing; in transitives constructibles refer to different things in which there is a transition from one to another (Thomas of Erfurt 1972: 284 [93]). It follows that the compositio is always intransitive and a reflexive (reciprocatio) is also intransitive. It is notable that ‘video’ in Figure 8.5 is not resolved into a first person modus ens, despite this being incorporated as the person of the verb. Modistic syntactic structures rely on Aristotle’s four ‘causes’ (see p. 39) as criteria for well-formedness. The ‘material’ from which a constructio is constituted are PARTES or combinations of PARTES of only two kinds, dependent and terminant, and that is why Socrates percutit Platonem has several dependencies (see Figure 8.3 and Thomas of Erfurt 1972: 274 [89]). The combination of constructibles gives rise to the ‘form’ of the constructio: ‘constructabilium unio est forma constructionis’ (ibid.). The constructio is brought into being (the ‘efficient cause’) by a proper congruent combination of constructibles. And ‘the expression of a compound concept of the mind is the goal of the construction’, i.e. the ‘final cause’ (ibid. 276 [89]). A well-formed constructio needs to have congruitas and the well-formed sentence be perfectio – a concept that is heir to the Stoic notion of completeness. There are two types of congruity, ‘concord’ and ‘propriety’ (nonanomaly). Cappa nigra “black cape (hooded clerical garb)” displays concord and propriety; constructio dependent terminant dependent terminant Video legentem librum “I see someone reading a book” Figure 8.4. Modistic analysis of Socrates albus currit bene. 174 The Western Classical Tradition in Linguistics *cappa nigri lacks concord; *cappa categorica “categorical cape” has concord but no propriety. In isolation a dictio or PARS is powerless without a ratio and modus consignificandi which identify its syntactic function and lead to a modus construendi which permits it to combine with other PARTES. There are two aims in every perfectio construction: first is the expression of a complete concept by means of a construction containing a verb; second is to create perfect understanding in the mind of the hearer by means of a congruent combination of constructibles containing the verb. Notice that the occurrence of the verb is pivotal (a fact remarked by Siger de Courtrai 1913: 139). Priscian had allowed for sentence fragments to occur in normal use; but the modistae were interested in the ideal constructio rather than a pragmatically viable one. Incomplete constructions like si Socrates currit “if Socrates runs” require a terminant such as laedit pedem “he gets sore feet” (Thomas of Erfurt 1972: 316 [118]). Homo albus (see the suppositum in Figure 8.4) is incomplete because it cannot be assigned a truth value, whereas homo est albus is a sermo congruus et perfectus “a well-formed sentence” because ‘circa quam compositionem consistit veritas et falsitas’ (ibid. 314 [116]; the idea is found in Aristotle’s On Interpretation). ‘It is required of the construction that no dependent should be left unterminated which might hold it back from its final purpose which is to express a compound concept of the mind and to generate perfect sense in the mind of the hearer’ (ibid.). I have quoted extensively from Thomas of Erfurt’s Grammatica Speculativa, written c. 1300–10 which is one of few extant treatises to be complete. Thomas was one of the last of the modistae, and his work was much copied and commented upon in the following century. Nonetheless, speculative grammar was a theoretical discipline and was not so much used in the teaching of Latin as the Doctrinale of Alexander de Villa Dei and similar pedagogical works. The realism of speculative theories gave way to the nominalism of William of Ockham (c. 1285–1349). William says in Summa Logicae (c. 1323) that both language and cognition use terms like noun, verb, pronoun, adverb, conjunction, and preposition, but cognition is more parsimonious than language: one might wonder whether, among intentions [of the mind], participles constitute a separate part of speech over and above verbs in the way that they do in spoken and written language; for the participle of any verb, with the appropriate form of to be, signifies precisely what the corresponding form of that verb by itself signifies. [… T]he relevant multiplicity has no place at the conceptual level. But since the distinction which spoken language exhibits between verbs and their participial forms does not enable us to express anything we could not express without the distinction, there is no need to postulate mental participles to correspond to spoken participles. A similar doubt is possible in the case of pronouns. (Ockham 1974: 52f [I.3]) Mental language has no need to distinguish verbs from participles nor pronouns from nouns. And although differences in number and case are both linguistically and mentally relevant, solecisms like homo est alba “the man is pale.F” are no impediment to understanding because (an utterance of) this sentence has the same truth condition as homo est albus[M] (ibid.). In other words, the study of syntax is no way to study cognition. What is important is reason itself. So grammar once again diverged from logic; though not for long. The true ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 175 grammatical heirs to the modistae were the rationalists of the late sixteenth to late eighteenth centuries. The recognition of vernacular languages From the end of the middle ages, vernacular languages came to be acknowledged as important vehicles of communication within the lay community, and were therefore worthy of study in their own right. The literary works in Italian of Dante Alighieri (1265–1321), Petrarch (Francesco Petrarca, 1304–75), and Giovanni Boccaccio (1313–75) led Leon Battista Alberti (1404–72) to defend the use of Italian and write a grammatical sketch of the language. Dante championed the vernacular in De Vulgari Eloquentia, written in Latin between 1302 and 1305 but not published until 1529. What I call ‘vernacular’ is the language which we learn without any formal instruction by imitating a nurse. There also exists another, secondary language, which the Romans call grammatical [gramatica]. This secondary language the Greeks and others also have, but not all [peoples]; indeed, few achieve complete fluency in it, since knowledge of its rules and theory can only be developed through dedication to a lengthy course of study. And of these two, the nobler is the vernacular: in part because it was the language originally used by mankind; in part because the whole world makes use of it – though with different pronunciations, accidence and words; and partly because it is natural for us, whereas the other is, instead, artificial. (Dante Alighieri 1981, I.i.2–4) The first grammar of Spanish, Gramática de la Lengua Castellana, was published in 1492 by Antonio de Nebrija (1444–1522); it included a section on Spanish for foreigners and developed Spanish terminology for grammar. In his Reglas de Orthographia en la Lengua Castellana of 1517, Nebrija recognized that languages use only a small number of the sounds that humans can make (Nebrija 1926). In Germany Christoph Helwig (1581– 1617) wrote the very short Allgemeine Sprachkunst, published in 1619 by his widow. Motivated by the belief that German Christians should read the Scriptures in Hebrew and Greek, he argued that students should learn the universal properties of grammars before studying a particular language (Juntune 1985: 98). I briefly surveyed the rise of English grammars in Chapter 7. The earliest French grammars were written by Englishmen in the fifteenth and sixteenth centuries; the best of these was Lesclarcissement de la langue françoyse (Palsgrave 1530). Petrus Ramus (Pierre de la Ramée, 1515–72) wrote grammars of Latin and Greek as well as a theoretical work Scholae Grammaticae 1559. In his Gramere10 of French (Ramus 1562), he divided grammar, modistic style, into Etymology (letters and morphology) and Syntax. In 1559–60 Abel Mathieu established that many French words once thought to show the Greek origins of French were in fact comparatively recent borrowings from Greek, arising from translations of Greek texts. His work was followed by the development of a more systematic comparative method for establishing etymologies, leading to Gilles Ménage’s Les Origines de la Langue Françoise (Ménage 10. The title is in Ramus’ revised spelling for French grammaire; it was normalized in later editions. 176 The Western Classical Tradition in Linguistics 1650). The book begins with a survey of letter conversions, for example, from V to E and V to F (bear in mind that V corresponds to both our V and our U): LAT. ???ò?, remus. ???o???ò?, remulcus. ??????, mattea. Priscien liure I. V transit in E, pondus, ponderis; dejerat, pejerat, pro dejurat, pejurat; labrum, labellum; sacrum, sacellum. Antiqui auger & augeratus pro augur & auguratus dicebant. En F. FRANC. vices, vezes, fois. Vara, la Fere (Ville). nauis, nef. clauis, clef. boue, boeuf. nouum, neuf. ouum, oeuf. (Ménage 1650: xxxv) All etymologies look plausible and systematic, e.g. HIVER. De hibernum, qu’on a dit pour hiemps, comme vernum pour ver; æstiuum pour æstas, & diurnum pour dies <& quernus pour quercus [p. 801]>. Solin chap. v. parlant de la Sicile: Principem urbium Syracusas habet, in qua etiam cum hiberno conduntur serena, nullo non die sol est. Tertullien: Hiberna, verna & autumni. Voyez M. de Saumaise sur Solin pag.105. 891. & 1258. Les Italiens on fait de mesme verno de hybernum. (ibid. 385) Ménage (1613–92) is among the earliest of the ‘modern’ more rigorous and systematic etymologists. In Diatriba de Europaeorum linguis (Scaliger 1610) Joseph Scaliger (1540– 1609) suggested that there were several parents for European languages. Johannes Becanus of Antwerp in 1569 claimed that all languages derive from his mother-tongue, Brabantic (a dialect of Dutch): Adam derives from Hath-Dam “dam against hate”, Eve from Eu-Vat, “barrel from which people originated” or from Eet-Vat “oath-barrel”, Noah from nood “need”, Latin quercus “oak” from werd-cou “keeps out cold”. Becanus convinced himself that Egyptian hieroglyphics represented Brabantic. But he also launched the idea that many European languages derive from Scythian, spoken during the eighth and seventh centuries BCE by a powerful warlike nation of Iranian stock in what is now the Crimea. Wilkins 1668: 3f says Irish reputedly derives from Scythian; and Gottfried Wilhelm Leibniz (1646–1716) believed Scythian to be what is now called Proto-Indo-European (Waterman 1978: 63) – an idea that survived into the nineteenth century. The Greek historian Herodotus (c. 484–425 BCE) noted a few Scythian words that have been identified with Indo-European cognates.11 With the reduced hegemony of the Roman Catholic Church, by the seventeenth century Latin was dying as a lingua franca. Classical instead of Church Latin became the object of study once again, as interest in classical authors was renewed during the renaissance; medieval Latin came to be despised as degenerate. With the change in the status of Latin the association between grammar and Latin was loosened. In the thirteenth century appeared grammars of Hebrew and Arabic; in the fourteenth, of Provençal and Catalan; in the fifteenth, Italian and Spanish grammars. ‘In the course of the sixteenth century grammars of 21 more languages were printed, and in the seventeenth century at least an additional 41’ (Rowe 1974: 361; for a list of these grammars see ibid. 372f). However, grammars of Native American languages, for instance, tended not to be available in Europe. So grammar was, to some extent at least, separated from Latin; nonetheless all learned men in Europe knew Latin until the twentieth century (and many scholars do today). However, there was no systematic contrast made between languages from different language families. Materials 11. There is more on language genealogy in Chapter 9. ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 177 were assembling for the growth of comparative linguistics, but it did not come about for another two hundred years following, on the one hand, the flowering and decline of general and rationalist grammar and, on the other, the search for an international language as a replacement for Latin. These two movements were closely intertwined, as we shall see. The search for a ‘philosophical language’ and a ‘real character’ The search for an international language as a replacement for Latin led to proposals by, among others,12 the Czech Comenius (Jan Amos Komenský, 1592–1670), Bishop John Wilkins (1614–72), George Dalgarno (c. 1626–87) and Leibniz. They were in touch with each other and with the publisher-philanthropist Samuel Hartlib (1600–62), a German who had settled in London. A strong motivation was the concurrent interest in a general or universal grammar based on rationalist principles. The language to be created was therefore dubbed ‘philosophical’ or ‘real’. It was to be a clear representation of thought that would make learning easier and the transmission of knowledge more precise; it was also to be a nonsectarian, non-nationalistic means to reduce religious and political conflict. Babel was a curse that could be mitigated by a universal language. [The Universal Character and Philosophical Language is judged] to be of singular use, for facilitating the matter of Communication and Intercourse between People of different Languages, and consequently a proper and effectual Means, for advancing all the parts of Real and Useful Knowledge, Civilizing barbarous Nations, Propagating the Gospel, and encreasing Traffique and Commerce. (Foreword by William Morice to Dalgarno 1661) In a broadsheet, Dalgarno writes, ‘This Character shall immediately represent things, and not sounds of Words, and therefore Universal [sic], and equally applicable to all Languages’ (Cram and Maat 2001: 109). It was, therefore, ideographic in conception. However, the language had to be transmitted through some medium, and in the written form was a ‘real character’ (i.e. one that directly represents the denotatum); there was a ‘vocal character’ (phonological representation) too. The real character was in part motivated by existing interests in the universality of mathematics, in shorthand, cryptography (because of bloody civil unrest in Britain between 1642 and 1651), and spelling reform (see Chapter 9). Francis Lodwick or Lodowick (1619–94) was, like others, motivated by * the partly true but over-simplistic belief that Chinese logographs consist of ‘real’ basic roots and modifications on them; * the universal recognition of Arabic numerals (0, 1, 2, …), despite almost every language having different names for them; and 12. Such as Tommaso Campanella Philosophiae Rationalis 1638; Jan Comenius Via Lucis 1668 (in circulation from 1641); Pedro Bermudo Arithmeticus Nomenclator c. 1653; Thomas Urquhart Eksubalauron, or the discovery of a most exquisite jewel 1652; Seth Ward Vindiciae Academiarum 1654; Cave Beck The Universal Character 1657; Johann Becher Character, pro notitia linguarum universali 1661; Athanasius Kircher Polygraphia Nova et Universalis 1663 (Cram 2000). 178 The Western Classical Tradition in Linguistics * the possibility of exploiting abbreviation such as the Cabbalistic use of Hebrew letters as symbols for whole words (e.g. ? for “God”). Lodwick’s characters were constructed from signs for a large number of basic roots and diacritics to signify negation and secondary grammatical categories like tense and mood. His conceptual analysis was ad hoc: e.g. king is composed from “rule + England”, emperor from “rule + Germany” (Lodwick 1647: 11) – which makes the ‘common writing’ logographic. There were better examples of ‘philosophical language’. Dalgarno’s Ars Signorum claimed on its title-page: The Art of Signs or a Universal Character and Philosophical Language. By means of which speakers of the most diverse languages will in the space of two weeks be able to communicate to each other all the notions of the mind (in every day matters), whether in writing or in speech, no less intelligibly than in their mother tongues. Furthermore, by this means also the young will be able to imbibe the principles of philosophy and the true practice of logic far more quickly and easily than from the common writings of philosophers. (Cram and Maat 2001: 139) Nevertheless Dalgarno wrote in Latin and ‘only for the learned’ (Dalgarno 1661: 36). He analysed all knowledge into 17 irreducible categories indicated by letters from the Roman and Greek alphabets. A second letter indicates a sub-class, the third a sub-sub-class, etc. Further specification distinguished plant species and the like. Certain letters are prefixed to denote contrary meaning, or the mean between extremes; suffixes indicate secondary grammatical categories like tense and mood, cf. ponesi “I love”, ponese “I loved”, ponesa “I have loved”, pones? “I had loved”, pones? “to love”, ponos? “to be loved”. The categorial system was probably intended to be a Cartesian methodical arrangement of ideas (Arnauld and Nicole 1965; 1996; Descartes 1637; 1968), e.g. Ens, res Substance Accident Concrete Corporeal Mathematical Physical Artificial Spiritual Soul Angel Composite [man] Examples of the philosophical language (see Cram and Maat 2001; Dalgarno 1661) are tim “affirm”, trim “deny” (note the contrary infix -r-), tin “speak”, trin “write” (-r- infix), tif “understand”, tib “teach”, trib “learn” (-r- infix), tig “tell, narrate”, tip “rumour”, tit “define”, trit “distinguish” (-r- infix), tik “restrict”, trik “increase” (-r- infix), daf “past”, dlaf “present”, draf “future” (-r- infix); snapg?m “bee” = snap “flying exsanguinous beast” + g?m “sweet”; nefsis “gold” = nef “metal” + sis “perfect”. Radical words are ‘syncategoremata’: pronouns, conjunctions, interjections, and prepositions; they occur without an affix, e.g. sab “with”; an affix converts sab into a substantive sabu “instrument” (Dalgarno 1661: 83). ‘[O]ne and the same thing can be called by several names, by means ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 179 of different combinations of radicals, thus elephant can be N?kbeisap [N?k “whole-footed beast” (vs cloven-footed) beis “mathematical accident” ap “superlative”] or N?ksofpr?k [sof “without” pr?k “to rise”]’ (ibid. 41). Dalgarno’s taxonomy of human notions is not comprehensive, but a minimal set of elements and rules to express the notions. Sanctius 1587 had earlier resolved the impersonal weather verb pluit “it is raining” into the unsatisfactory Deus pluit “God rains” or pluvia pluit “the rain is raining”. Dalgarno 1661: 76 offers the same: ‘nen nenesi, pluvia pluit’. His French contemporaries Lancelot and Arnauld 1660: 125f did something similar, interpreting pluit and il pluie in terms of pluvia fit, la pluie est “the rain is”. Also like Lancelot and Arnaud and a tradition that extends back to Aristotle 21b 9, Dalgarno resolves verbs into a copula + participle, e.g. amamus into sumus amantes: ‘the word amamus [“we love”] contains within itself four distinct notions; i.e. we, present tense, are (or rather yes), and loving’ (op.cit. 65 [sic]). Impressive though Dalgarno’s work was, far more comprehensive is Wilkins 1668, Essay toward a Real Character and a Philosophical Language. Wilkins also set out to schematize human knowledge: The second Part shall contein that which is the great foundation of the thing here designed, namely a regular enumeration and description of all those things and notions, to which marks or names ought to be assigned according to their respective natures, which may be styled the scientifical Part, comprehending Universal Philosophy. It being the proper end and design of the several branches of Philosophy to reduce all things and notions unto such a frame, as may express their natural order, dependence, and relations. (ibid. 1) Like Dalgarno, Wilkins thought that ‘a Real Universal Character, […] should not signifie words, but things and notions, and consequently might be legible by any Nation in their own Tongue’ (ibid. 13). He was motivated by the countering of Babel: ‘in almost every valley of Peru, the Inhabitants have a distinct Language. And [in North America it is reported that there are] more than a thousand different Languages’ (ibid. 3). He also sought to offset dialect variation: Whereas the inhabitants about London would say, I would eat more cheese if I had it. A Northern man would speak it thus, Ay sud eat mare cheese gyn ay had et. And a Western man thus, Chud eat more cheese an chad it. (ibid. 4) And it would obviate the change of language over time: ‘Every change is a gradual corruption’ (ibid. 7–8). It would reform spelling and remove all irregularities, homonyms and other ambiguities from language. Wilkins analysed nearly all knowledge13 into the 40 categories reproduced in Figure 8.6. It was thesaurus-like (ibid. 22–288), for example, herbs are categorized according to their leaves (70–80), their flowers (81–95), and their seed-vessels (96–106). It was not a natural classification. After the original manuscript was lost in the Great Fire of London, Wilkins was assisted by John Ray (1625–1705), who prepared the botanical tables; Ray wrote in a letter to Martin Lister (May 7, 1669): 13. Wilkins’ list of exclusions (ibid. 295f) includes names (of places, nations, sects), times, titles, degrees of professions, legal terms, garments, stuffs, games, drinks, meats, tunes, tools. 180 The Western Classical Tradition in Linguistics Things TRANSCENDENTAL GENERAL RELATION MIXED RELATION OF ACTION Words DISCOURSE CREATOR Collectively WORLD Distributive Substance Inanimate ELEMENT Species Minerals STONE METAL Plant HERB LEAF FLOWER SEED-VESSEL SHRUB TREE Sensitive EXANGUIOUS Sanguineous FISH BEAST BIRD Parts PECULIAR GENERAL Accident Quantity MAGNITUDE SPACE MEASURE Quality NATURAL POWER HABIT MANNERS SENSIBLE QUALITY SICKNESS Action SPIRITUAL CORPOREAL MOTION OPERATION Relation Private OECONOMICAL POSSESSIONS PROVISIONS Publick CIVIL JUDICIAL MILITARY NAVAL ECCLESIASTICAL Figure 8.6. Wilkins’ 40 categories (from Wilkins 1668: 23 [sic]). ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 181 I was constrained in arranging the tables not to follow the lead of nature, but to accommodate the plants to the author’s prescribed system. […] What possible hope was there that a method of that sort would be satisfactory, and not manifestly imperfect and ridiculous? (Quoted by Aarsleff 1992: 27f) The categorization was driven by (now obscure) mnemonics into groups of nine, six, and pairs. The character consists of about 3000 radicals. They include no verbs. [W]hatsoever hath an Essence, must likewise have an Act; either of Being or becoming: or of Doing or being done: or of making or being made: to be, or do. And consequently every Radical Substantive which is capable of Action, should have an Active or Passive formed from it, which is commonly called a Verb. (Wilkins 1668: 300) Instead Wilkins lists the copula among essential particles. That part of speech, which by our Common Grammarians is stiled a Verb, (whether Neuter, Active or Passive) ought to have no distinct place amongst Integrals in a Philosophical Grammar; because it is really no other then an Adjective, and the Copula sum affixed to it or conteined in it: So Caleo, Calefacio, Calefio, is the same with sum Calidus, Calefaciens, Calefactus. (ibid. 303) Dalgarno had done much the same in resolving amamus into sumus amantes. The most necessary [particle] amongst all the rest, which is essential and perpetual in every compleat sentence, is stiled the Copula; which serves for uniting of the Subject and Predicate in every Proposition. The word Subject I use, as the Logicians do, for all that which goes before the Copula; which if it consist only of one word, then it is the same which Grammarians call the Nominative case. By the word Predicate, I mean likewise all that which follows the Copula in the same sentence, whereof the Adjective (if any such there be) immediately next after the Copula, is commonly incorporated with it in instituted Languages, and both together make up that which Grammarians call a Verb. (ibid. 304) Particles may belong 1. To the Integral alone, as Articles. 2. To the Copula alone, as the Modes. or 3. Both to Integral and Copula as the Tenses. (ibid. 315) As in other schemes for a ‘philosophical language’ words are formed on the set of their semantic components; e.g. the word for father is constructed from the symbols for ‘relation’ + ‘consanguity’ + ‘direct ascendant’ + ‘male’ (ibid. 249, 396), which is more finely semantically decomposed than Dalgarno’s ‘father is pagel [pag “beget” el “person”], the person who begets’ (Dalgarno 1661: 92). ‘World’ is Da + ‘celestial’ (= Dad “heaven”) + ‘globe of sea and land’ (= Dady “Earth”); “planet” is Dade and “comet” Daded (Wilkins 1668: 51f, 398, 416). Initial o marks a contrary: Dad “Heaven” odad “Hell”; Dab “spirit” odab “body”. The vocal form of the Wilkins character is rather similar to that of Dalgarno: Tit? “temperateness”; Tit?la “temperateness + excess” = heat; Tit?lo “temperateness + defect” = cold; Tuit?la “adjective + temperateness + excess” = hot; Tuilt?la “adjective + active + temperateness + excess” = heating; Tuimt?la “adjective + passive + temperateness + excess” = heated (Vickery 1992: 344). To Wilkins’ Essay was attached An Alphabetical Dictionary Wherein all English Words According to their Various Significations, Are either referred to in their Places in the Philosophical Tables, Or explained by such Words as are in those Tables. This dictionary 182 The Western Classical Tradition in Linguistics was compiled by William Lloyd; it has its own title page and, unlike the Essay, has no page numbers. According to Dolezal 1992: 309–13 it is innovative in the following respects. It covers a broad range of vocabulary that includes many multiword lexemes. It is highly systematic with a very methodical construction of entries. Unusually for contemporary dictionaries, the words used in definitions were also defined. [T]he dictionary is largely dependent upon the Tables for complete definition of its lexicon. […] For example, the Tables have a Radical for “Bee” but not for “Hive”; however, “hive” could be expressed in the philosophical language by addition of the Transcendental Particle “House” – i.e., bee + house = hive; ammunition + house = arsenal; fornication + house = brothel. (Dolezal 1992: 316, 321) Lloyd’s dictionary was a model for later lexicographers; but unfortunately Wilkins’ philosophical language and real character proved too abstruse and recondite for practical application, and his highly stimulating, if flawed, masterpiece remains merely a historical curiosity. Leibniz played with the notion of creating a philosophical language that would concomitantly be a medium for international communication, a simplified notation for science, and a method of discovery and demonstration (Cohen 1992). Such an aim was very similar to the aims of all others who worked in this field. However, Leibniz also proposed a logical calculus that had no precursor: he would try to symbolize primitive concepts using prime numbers and complex ones by the appropriate product. Such a synthesis as the latter could, in principle, be decomposed into prime factors. Unfortunately, Leibniz was no more successful than Dalgarno or Wilkins – indeed less so, because he published nothing on the matter. The heirs to the seventeenth century seekers after a philosophical language include: * The late nineteenth and twentieth century philosophers who have developed a variety of logics (e.g. Frege 1980; Russell 1956; Carnap 1937; 1950; 1956; 1959; Church 1941; 1956; ?ukasiewicz 1963; 1998; Montague 1974; Kripke 1963; 1972; Lewis 1969; 1973; 1986; 1998; Groenendijk and Stokhof 1991; Kamp and Reyle 1993. Surveys include Kneale and Kneale 1962; Gamut 199114; McCawley 1993; Benthem and ter Meulen (eds) 1997; Seuren 1998). * The nineteenth century phonetic-based shorthand systems of Isaac Pitman (1837) and John Gregg (1888). Shorthand has been dying out with the development of voice recording and most recently speech recognition programmes that automatically convert speech to print. * The international alphabet of Morse code was developed between 1837 and 1844 for telegraphic communication. It is today mostly obsolete, though it is used in the vibrating alert of mobile phones, and the default SMS alert is ••• ? ? ••• “sms”. * The International Phonetic Alphabet was developed in 1887–8 (see Chapter 9). * The International Radio Operator alphabet (Alpha Bravo Charlie Delta Echo Foxtrot Golf Hotel India Juliet Kilo Lima Mike November Oscar Papa Quebec Romeo Sierra 14. L. T. F. Gamut is a collective pseudonym for Johan F. A. K. van Benthem, Jeroen A. G. Groenendijk, Dick H. J. de Jongh, Martin J. B. Stokhof, and Henk J. Verkuyl. ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 183 Tango Uniform Victor Whiskey X-Ray Yankee Zulu) was first adopted by the International Telecommunication Union in 1927 and updated in 1932. * The componential analysis of meaning was taken up by Kroeber 1909 for analysis of kinship systems and Goodenough 1956; Zellig Harris 1948; Katz and Fodor 1963; Lounsbury 1956; Nida 1951 brought it into modern linguistics, where it thrives within a variety of theories. * The development of Unicode for use on the web began in 1986 and it is still being developed; it is contrary to the spirit of a single alphabet (writing system) for all languages, since it aims to make orthographic symbols from all writing systems for natural and non-natural languages (like logics and mathematics) available in digital form for faithful reproduction on web documents. General or universal and rationalist grammar Rationalist (or rational) grammar goes beyond description to achieve a rational explanation for linguistic phenomena (Chomsky 1972: 14f). As Lakoff 1969 pointed out, the tradition of general and rationalist grammar seems to begin with Sanctius (Francisco Sánchez de las Brozas, 1523–1600), who, in Minerva (Sanctius 1587; 1664), postulates a syntactic rather than semantic rationalist basis for grammar that bears some resemblance to Apollonius’ system of explanatory underlying forms that do not appear on the surface to explain the ‘elliptical’ surface structure of ordinary language. Sanctius set out to explain the logic (ratio) of the Latin language (Sanctius 1664: 2). He sets logic as prior to tradition, authority, and custom, saying that usage is not changed without there being a logical basis, otherwise it is abuse (ibid. 4). Since the subject about which we are speaking is to be supported first by logic, then by attestations and usage, no one should be surprised if now and then we do not follow the great men, for however much prestige a grammarian may have in my eyes, unless he supports what he has said with logic and with examples that are put forward, he will instil in me no confidence, particularly on questions of grammar. ‘Grammarians,’ as Seneca says […] ‘are the guardians of the Latin language’ – not its creators. (ibid. 5, transl. by Lakoff) Language is the product of a rational mind and is therefore open to rational explanation, even if this is not immediately obvious. A grammarian must offer logical analyses and back these up with attestations and usage (ibid. 5). Sanctius allows for correct parsing on the basis of form without recourse to meaning. A grammarian is not prescriptive; instead the goal is to explain syntactic structures (ibid. 7). Language is elliptical and the meaning of underspecified expressions needs explaining; but this should be done carefully and noncontroversially (ibid. 269). It is notable that Sanctius has adopted many of the notions found in Apollonius Dyscolus (see Chapter 6); he differs in suggesting that the expanded versions of elliptical expressions are all possible surface structures that may be dispreferred for stylistic reasons (Lakoff 1969: 362); that was not the case for Apollonius. Perhaps general and rationalist grammar is most famously exemplified by the Port-Royal grammar Grammaire Generale et Raisonée “General and rational grammar” (Lancelot and Arnauld 1660), which is a sort of digest of grammatical ideas found in Lancelot’s Nouvelle 184 The Western Classical Tradition in Linguistics Méthode pour Comprendre, facilement et en peu de temps, la Langue Latine “New method for understanding the Latin language easily and quickly”, first published in 1644 but then greatly revised for the third edition of 1654 under inspiration from Sanctius (compare Lancelot 1644 with Lancelot 1681). The title-page to Grammaire Generale et Raisonée (Lancelot and Arnauld 1660; 1753) states the following: The explanation for what is common to all languages and the principal differences among them. This is the standard view of general grammar dating back to the thirteenth century, and it makes rationalist grammar the proper heir to speculative grammar. Nicolas Beauzée, in his Grammaire Générale of 1767, writes: Grammar, whose object is the description of thought via the spoken or written word, is of two kinds. One kind consists of immutable and universal principles which derive from thought itself and follow its course. The other kind consists of contingent principles, dependent on the fortuitous, arbitrary and mutable conventions which give rise to different languages. The first kind is called ‘general grammar’; the second is the object of ‘particular grammars’, i.e. grammars of particular languages. (Beauzeé 1767: ixf) Beauzée goes on to say that general grammar is ‘rational scientific study of the immutable and universal principles of language’, and grammatical knowledge is prior to all languages because its principles make languages possible. They are the same ones that direct human reasoning in its intellectual activities. In a word, they are eternally and universally valid. (ibid. xf) Particular grammar is the ‘art’ or artifice of applying the principles of general grammar to ‘les institutions arbitraires & usuelles’ of a particular language, and Grammatical artifice, unlike grammatical knowledge, depends on the prior existence of languages so that the general principles of language can be introduced to act on them artificially; because the analogical systems introduced by artifice are the necessary result of observations that have been made of existing usage. (ibid.) Because the principles of general grammar are those of human reason, it follows that the characterization of general grammar is a characterization of the properties of human reason. Hence it was called rationalist grammar, and this is one way in which it differs from modistic grammar. Géraud de Cordemoy (1620–84), in his Discours Physique de la Parole “A philosophical discourse on speech” (Cordemoy 1668; 1970), is backward looking when he writes that grammarians teach parts of speech in the same order that children learn them: substantives, adjectives, verbs, etc. (ibid. 34–36). But he is forward looking when he writes the Reason of Children is entire from the beginning, seeing they learn perfectly the Language of the Countrey where they are born, and that in less time than Men of age need to learn that of a Country, where they should chance to travel, and not find anybody that understood theirs. (ibid. 38) This is a thought echoed by Chomsky and others in the twentieth century. Learning a second language is a matter of assigning a new linguistic expression to ideas associated with an existing linguistic expression in the first language. ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 185 [W]hen we are in company with persons of different Countries, whose Tongues we understand, we easily retain every news, and all what was said upon the matters, that were spoken of, without remembering just the words nor the Language that was made use of to give us those images, which remain of them in us. (ibid. 39) [For some people] when they learn a new Language, they always joyn to the words of that, which they already know, the words of the second, to represent to themselves what they signifie. Others, that have another disposition of the brain, do so easily joyn the sound of a new word in it self to the Idea of the thing, that that Idea is equally represented to them by the two words, and they not obliged to think on the one to understand the other. (ibid. 87) Thus, although the manner of expression is different across languages, the underlying general and rationalist grammar is universal. It is a hypothesis about the nature of human reason. But let’s go back to the Grammaire Generale et Raisonée. It was a product of the Jansenist convent of Port-Royal, outside Paris. Jansenism, often in conflict with the Jesuits, was a seventeenth century form of Augustinianism. It taught that divine grace, rather than good works, is the key to salvation. The best opportunity for children to resist the curse of original sin is education and especially a mastery of language and thought attained through the study of grammar and logic. The Augustinian motto is: We owe then to reason what we understand; to authority what we believe; to error what we have an opinion about. (The Advantage of Believing, Ch. 25, Augustine 1885: 130) Lancelot specialized in grammar and Arnauld in logic. Their colleague Bernard Lamy (1640–1715) was a rhetorician, and published De L’Art de Parler “On the art of speaking” (Lamy 1676). So, Port-Royal was the centre of contemporary studies in language and logic. Claude Lancelot (c. 1616–95) was co-author of hugely successful grammars of Latin, Greek, and Italian (Lancelot 1644; 1681; 1758). Where previous grammars were often in Latin, he used French – arguing that it is common-sense to use the mother tongue to teach a foreign language. Rote learning and memorizing passages from the classical authors were replaced by a thorough grounding in general grammar exemplified mostly from French, but also much Latin and a little Greek and Hebrew. Lancelot believed that by using the rationalist basis for general grammar he could teach a second language using the general grammar implicit within the student’s first language. The rules of general grammar are ‘natural’. A Sanctius-like concern with explanatory underlying structure is found in Lancelot’s grammars. Following Aristotle’s suggestion that ‘occasionally, perhaps, it is necessary to coin words, if no word exists by which a correlation can adequately be explained’ (Categories 7a 5), Lancelot proposes a strict analogical regularity for the derivation of Latin inchoative verbs like calesco “grow warm”: The inchoatives are formed of the second person of the present, as from Labo, as; labasco; from Caleo, es; calesco: […] of Dormio, is, dormisco. It is the same in regard to deponents, which are formed by feigning the active of the primitive. For Fruiscor comes as it were from fruo, is. The impersonals also follow this analogy: Miserescit, from misereo, es, &c. […] 186 The Western Classical Tradition in Linguistics Some of them are even supposed to come from nouns, as Ægresco from æger; Repuerasco from puer: though they may be said to come from the verbs Ægro, repuero, and the like, which are no longer in use: just as Calvesco, which they generally derive from calvus; and Senesco from senex, come from calveo, which we find in Pliny, and from seneo, in Catullus. (Lancelot 1681: 335) Lancelot’s discussion of relative pronouns in Nouvelle Méthode has a significant theoretical consequence that is discussed below. The emphasis throughout the book was on good communication rather than good style, and students were encouraged to use naturalistic language in translation. Lancelot wrote user-friendly textbooks that present concise rules with catchy titles; give pronunciation clues (e.g. stress was marked on foreign words); points of special note like case endings are typographically distinct; French text is in a different type-face from Latin or Italian; and comments (notes) are in small print. Many of these innovations were later used by others, including Lowth 1762 and Murray 1795 (see Chapter 7). Antoine Arnauld (1612–94) was principal author of La Logique, ou l’Art de Penser “Logic, or the art of thinking” (Arnauld and Nicole 1965; 1996), the most important book on logic in the seventeenth century. It was promptly translated into Latin and English and underwent many reprintings in the late seventeenth and eighteenth centuries. The Logic is strongly influenced by René Descartes’ (1596–1650) Discours de la Méthode “Discourse on method” (Descartes 1637; 1968), with its emphasis on using reason to think things out for oneself. ‘[L]ogic must examine how ideas are joined to words, and words to ideas’ (Arnauld and Nicole 1996: 24). The Logic is approachably informal and divides the topic into four parts. The first is to have an idea of something, often through the senses but ‘the ideas of being and thought in no way originate in the senses. Instead, the soul has the faculty to form them from itself, although often it is prompted to do so by something striking the senses’ (ibid. 29). The emphasis, then, is on conception rather than perception. The Logic uses Descartes’ example (6th Meditation, Descartes 1968: 150f) that one can comprehend and describe all the properties of both a triangle and a chiliagon (figure of 1000 angles),15 but picture only the former (Arnauld and Nicole 1996: 25f). To gain a proper idea of something, it is best to break it down into manageable parts. Neither Descartes nor Arnauld gives due credit to Aristotle for having proposed exactly this. To comprehend a general term one has to identify its attributes; its extension is the sets (and subsets) of individuals to which it applies. ‘I cannot conceive prudence while denying its relation to some person or other intelligent nature having this virtue’ (ibid. 32). Words are the conventional signs of thoughts (ibid. 37) but we should not try to define all words, because this would often be useless. […] For in order to define a word it is necessary to use other words designating the idea we want to connect to the word being defined. And if we again wished to define the words used to explain that word, we would need still others, and so on to infinity. [… W]e should accommodate ourselves to usage as much as possible. [D]efinitions […] represent the truth of usage rather than the truth of things. (ibid. 64, 66). 15. E.g. its 1000 angles add up to 1996 right angles. ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 187 The discussion of definitions was inspired by the work of the Port-Royal mathematician Blaise Pascal (1623–62). Primitive terms, evident to all, cannot be defined; ‘real’ definitions are descriptive and state the essential properties in the definiendum; ‘nominal’ definitions stipulate the conventions by which a linguistic term is to be used. Bringing ideas together creates a ‘judgment’ consisting of an idea and an attribute (another idea) predicated of it. Affirmation brings two ideas together (the earth is round) whereas negation separates them (ice is not hot). One needs to be as certain as possible of identifying the real truth because ‘he who is falsely persuaded that he knows the truth thereby makes himself incapable of learning about it’ (ibid. 35). ‘Reasoning’ is syllogistic: the act of forming a third judgement from two others. Sorites are a series of interdependent arguments, e.g. Misers are full of desires. Those who are full of desires lack many things because it is impossible for all their desires to be satisfied. Those who lack what they desire are miserable. Therefore, misers are miserable (ibid. 137). Finally there is ‘method’. Never accept anything as true that is not known for sure to be so. Divide up the data into the smallest elements required. In definitions, use only terms that are perfectly known or have already been explained. Leave no term undefined. Be exhaustive, but pay attention to natural order, beginning with the most general and the simplest, and explaining everything belonging to the nature of genus before proceeding to particular species (ibid. 238, 259). The spirit of the Logic is combined with Lancelot’s pedagogical prowess in the Grammaire Generale et Raisonée (hereafter GGR), which embodies that aphorism of the thirteenth century modista who wrote ‘non ergo gramaticus sed philosophus [...] gramaticum invenit’ (Thurot 1964: 124 Bibl. Nat. de Paris lat. 16297 f.131). But the grammars of the modistae and the rational grammar of Port-Royal, although broadly similar in distinguishing universal grammar from particular grammars, are nonetheless very different in the way they approach the problem of universal grammar. Modistic grammars investigate a tripartite relation between reality (res), human understanding of the reality (modus intelligendi), and the linguistic expression that signifies the reality (vox). The speculative grammarians philosophized on the relation between form and meaning (modus significandi), and also the relation between form and thing (modus signandi); but they ignored the actual form of the sign (vox) as being an accidental element of grammar, inessential to general grammar. The rational grammarians, on the assumption that grammar reflects meaning, investigate the link between meaning and form and do discuss linguistic form. They de-emphasize reality. Thus the modistae speculated on nonobservable phenomena, whereas the rational grammarians discuss what would nowadays be called the relationship between deep and surface grammar. The example of the latter made famous by Chomsky 1966 is the sentence Dieu invisible a créé le monde visible. GGR says of this when I say Invisible God has created the visible world through my mind pass three ideas embedded in this proposition. Firstly there is the idea that God is invisible. 2. That he created the world. 3. That the world is visible. And of these three ideas (expressed here as propositions), the second is the principal and essential part of the proposition in the original sentence. The first and third are subordinate; the former being its subject, and the latter its direct object. 188 The Western Classical Tradition in Linguistics These incidental propositions are often in our minds without being expressed in words as in the example given above. But sometimes they are expressly marked, and this is what the relative pronoun is for: so I can reduce my example to the following: God WHO is invisible has created the world WHICH is visible. (Lancelot and Arnauld 1660: 68f) Where two nouns are juxtaposed as in Vrbs Romana [city Roman; sic] “the city of Rome”, or in the N?Adj construction Deus sanctus [God holy] “holy God”, and when the modifier is a participle as in canis currens [dog running] “running dog”: all these are understood to contain the relative pronoun in their meaning, and it can be realized in surface structure Vrbs qua[e] dicitur Roma “the city which is called Rome”; Deus qui est sanctus “God who is holy”; Canis qui currit “the dog which runs”. It depends on the idiom of the particular language which one is used. Latin would use a participle where French uses a relative: Video canem currentem; Je voy un chien qui court (ibid. 69f). English uses both: I see the dog running and I see a dog which is running. When discussing the relative pronoun, GGR points out that Hebrew uses a demonstrative in relative clauses such as I see a dog which is running (I see a dog, that [dog] runs) and a relative (presumably ’asher) as complementizer in sentences like Dico quod tellus est rotunda “I.say that the.earth is round”. It is clear that GGR is as concerned with the surface expression of the underlying meaning as with the meanings of parts of speech. One criticism of the GGR analysis of Invisible God has created the visible world in terms of God who is invisible has created the world which is visible is that no mapping rule is given. There is, however, an explanation in Lancelot 1681, II.i.iv (p. 452) (which appeared in the edition of 1654) to the effect that the relative pronoun indicates a logical antecedent and ‘serves to make an incidental proposition [subordinate clause] form a part of another which may be called the principal’. This is exactly comparable with a similar notion in today’s grammars. Let’s compare the modistic description of verbs with the GGR description of verbs to show just how different they are. In the modistic description of the verbum the metaphysical features of being, becoming, succession, and flux replace the features of action or being acted upon. The characteristic of tempus is relegated to an accidental mode of signifying. Moreover, there is emphasis on the dependence of the verb on the subject. The nominal subject expresses substance, more correctly, the mode of permanence and repose. There must be substance before there can be a predication of it: therefore ens is prior to esse. The verb is distinguished from the participium with which it shares the essential modus motus (motile mode) by means of a specific mode which is the signification of the member of a proposition separate from its subject; in contrast, the participium adheres to its subject. The accidental modes of signifying correspond to the traditional accidentia of tense, mood, person, etc. As it was for Aristotle, every verb in a construction is a compositio of esse + participium (be + participle). Thus Socrates currit ? Socrates est currens (“Socrates runs ? Socrates is running”). As Lancelot and Arnaud say in GGR: [C]’est la mesme chose de dire Pierre vit, que de dire, Pierre est viuant. “To say Peter lives is the same as saying Peter is living.” (ibid. 91) This is true for French, Latin, and Greek but not true for today’s English; however, it was true for very early modern English. ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 189 The GGR description of the verb begins with a description of the proposition that is almost identical with that of Aristotle: A statement about things (as when I say, the earth is round) necessarily entails two terms; one being the subject of the affirmation, e.g. [the] earth; and the attribute, which is what is affirmed, e.g. [being] round. The link between the two terms is the action of the mind which predicates the attribute of the subject. (ibid. 89) Latin sum “I.am” entails the signification of the pronoun ego “I”; thus sum contains both subject and attribute (ibid. 92). A verb is ‘a word whose principal function is to signify predication [affirmation]’ (ibid. 90); they say that this has been overlooked by other grammarians, who have identified only the accidental signification of verb. According to GGR, Aristotle defined the verb as vox significans cum tempore “a meaningful sound with tense marking”; but this short-changes Aristotle. Defining a verb using the accidental features of tense and person is inadequate. Likewise it is inadequate to define a verb according to whether it denotes action or being acted upon; similarly, defining it as the modistae did as denoting “being”, “becoming” or “flux” may fail to account for neque omni, neque soli “all or even one [verb]”. The authors of GGR believe that verbs like existit “it exists”, quiescit “it rests”, friget “it is cold”, alget “it is chilled”, tepet “it is warm”, calet “it is hot”, albet “it is white”, viret “it is green”, claret “it is bright” are counterexamples to the semantic classification of the modistae; but to me these verbs denote either “being” or “becoming” and therefore fall within the modistic definition. Certainly there are parts of speech other than verbs which signify actions and passions, and even movement and change, for instance, the participles of the verbs: fluens is as motional as fluit but it is not a verb. Participles are also sensitive to time in that there are present and past participles. And if person is typically an accidental subcategory of verbs, we should remember that the vocative case is implicitly second person, and case is a nominal subcategory. On the other hand, a participle is not a verb because it makes no affirmation (nor, of course, does the vocative). In Chapter 7 we saw that some proponents of universal and rationalist grammar believe that grammar mirrors nature: THOSE PARTS OF SPEECH UNITE OF THEMSELVES IN GRAMMAR, WHOSE ORIGINAL ARCHETYPES UNITE OF THEMSELVES IN NATURE. (James Harris 1786: 263f [sic]) There was a slightly different take on a similar theme a century earlier: Everything we have said about syntax demonstrates that it has a natural order in it when all parts of a discourse are expressed simply, with not one word too many or too few, and when it conforms to the natural expression of our thoughts. (Lancelot and Arnauld 1660: 145) The universal grammar of the seventeenth century rationalists was given a Romantic twist by some grammarians in the late eighteenth century; a Romanticism which may have contributed to the philological investigations of the Junggrammatiker (neogrammarians, see Chapter 9). But Chomsky’s revival of universal grammar in the mid-twentieth century was 190 The Western Classical Tradition in Linguistics explicitly rationalist and harked back to what he perceived as the Cartesian foundations of the Port-Royal grammarians.16 The central doctrine of Cartesian linguistics is that the general features of grammatical structure are common to all languages and reflect certain fundamental properties of the mind. (Chomsky 1966: 59) Noam Chomsky justifiably claims to be heir to the rationalist tradition. There are, however, doubts about the validity of the label ‘Cartesian’. The principal argument in its favour is that Descartes noted (with little originality, see Chapter 10): it is particularly noteworthy that there are no men so dull-witted and stupid, not even imbeciles, who are not capable of arranging together different words, and of composing discourse by which to make their thoughts understood; and that, on the contrary, there is no other animal, however perfect and whatever excellent dispositions it has at birth, which can do the same. [… I]t is unbelievable that the most perfect monkey or parrot of its species should not be equal in this to the most stupid child, unless their souls were not of an altogether different nature from our own. (Descartes 1968: 74f) Chomsky interprets this to be a declaration of innatism (plausible) and linguistic creativity (questionable). It is also consistent with Descartes being in favour of general and rationalist grammar, although there is no explicit avowal of this in his writings. Nor is there any pronouncement in Descartes that directly bears on deep and surface grammar. Perhaps Descartes’ assertions of the necessary truth of what is rational is what appeals to Chomsky, but note that Descartes attributes this certainty to God – which is not a publicized opinion of Chomsky’s. [O]bserving that this truth: I think, therefore I am, was so certain and so evident that all the most extravagant suppositions of the skeptics were not capable of shaking it, I judged that I could accept it without scruple as the first principle of the philosophy I was seeking. [T]he proposition: I am, I exist, is necessarily true, every time I express it or conceive of it in my mind. (ibid. 53f , 103) [T]he things we grasp very clearly and very distinctly are all true, [are] assured only because God is or exists, and because he is a perfect Being, and because everything that is in us comes from him; whence it follows that our ideas and notions, being real things and coming from God, in so far as they are clear and distinct, cannot to this extent be other than true. (ibid. 58; and cf. 56, 128, 133, 145) It is rather puzzling that Chomsky adopted the title Cartesian rather than Port-Royalist, since he does recognize the gentlemen of Port-Royal as intellectual forebears (unless ‘Royalist’ was a problem). Leaving the so-called ‘Chomsky revolution’ for Chapter 12, here we focus only on Chomskyan universal grammar – henceforth UG, the system of principles, conditions, and rules that are elements or properties of all human languages not merely by accident but by necessity – of course, I mean biological, not logical, necessity. (Chomsky 1975b: 29) 16. On Cartesian linguistics see Aarsleff 1970; 1971; Lakoff 1969; Otero (ed.) 1994; Thomas 2004: 109–19. ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 191 In Aspects of the Theory of Syntax Chomsky wrote Real progress in linguistics consists in the discovery that certain features of given languages can be reduced to universal properties of language and explained in terms of these deeper aspects of linguistic form. (Chomsky 1965: 35) When this was written Chomsky was already seeing himself heir to the Port-Royal tradition. The celebrated distinction in Aspects between ‘competence’, an idealization of what the individual knows about their language, and ‘performance’, the manifestation of competence in the use of language but including such features of spontaneous speech as hesitation and slips of the tongue, was later revised into the distinction between ‘I-language’ (idealized and internalized) and ‘E-language’ (expressed, externalized). The statements of a grammar are statements of the theory of mind about the I-language, hence statements about structures of the brain formulated at a certain level of abstraction from mechanisms. […] UG now is construed as the theory of human I-languages, a system of conditions deriving from the human biological endowment that identifies the I-languages that are humanly accessible under normal conditions. (Chomsky 1986: 23) The medieval distinction between ‘general’ or ‘universal’ grammar common to all languages and the ‘accidents’ peculiar to the grammar of a particular language are reinterpreted in terms of the Chomskyan division between ‘principles’ and ‘parameters’. The grammar of a language can be regarded as a particular set of values for these parameters, while the overall system of rules, principles, and parameters is UG which we may take to be one element of human biological endowment, namely the “language faculty”. (Chomsky 1982: 7) The basic assumption of the P&P model is that languages have no rules at all in anything like the traditional sense, and no grammatical constructions (relative clauses, passives, etc.) except as taxonomic artefacts. There are universal principles and a finite array of options as to how they apply (parameters). (Chomsky 1995: 388) Note the move from the rationalist preoccupation with knowledge of language (in the mind) to the language faculty of the brain, which is a part of the human biological endowment. This incorporates an interest in language acquisition (also found among the seventeenth and eighteenth century rationalists) with the neurological mechanisms involved. So Chomskyan UG pretends to investigate the integration of grammar, mind, and brain. The ‘principles’ of Chomskyan UG are believed to be innate. This view is founded on the belief that all and only human beings have (human!) language and that full linguistic competence cannot be acquired by parroting. Chomsky 1988 presents this in terms of an answer to ‘Plato’s problem’, which we discussed in Chapter 2: how it is that everyone knows more than their direct experience in the world seems to warrant. The solution to Plato’s problem must be based on ascribing the fixed principles of the language faculty to the human organism as part of its biological endowment. (Chomsky 1988: 27) To acquire a particular language, e.g. Maasai, is to know the Maasai version of UG by the ‘switching on’ of certain parameters. For instance, the head parameter is responsible for Maasai speakers learning the conventional VERB?SUBJECT?OBJECT sequence of Maasai; English speakers will learn the SUBJECT?VERB?OBJECT sequence instead; and Japanese 192 The Western Classical Tradition in Linguistics speakers SUBJECT?OBJECT?VERB. Language universals cannot be violated, but do not necessarily manifest themselves in every language. For instance, ‘move ?’ (a UG principle of movement) may not be switched on parametrically: English questions typically involve movement (Where’s the restaurant? Is it close by?), Japanese questions do not (Restoran wa doko desu ka? Chikai desu ka? “Restaurant TOPIC where is Q? Close.by is Q?”). The Chomskyan turn to universal grammar has had the effect that all major grammatical theories from the late twentieth century have pretensions to universal applicability, whether they be Functional Grammars (like Dik 1997); Systemic Functional Grammars (like Halliday 1994; Halliday, Fawcett and Young (eds) 1987-88); Lexical Functional Grammars (like Bresnan 2001; Kroeger 2004); Role and Reference Grammars (like Van Valin 2001; Van Valin and LaPolla 1997); Head-driven Phrase Structure Grammar (like Pollard and Sag 1994); or cognitive grammars (like Langacker 1987; 1991; Fried and Östman (eds) 2004; Goldberg 1995; Goddard and Wierzbicka (eds) 2002). Universal grammar with rationalist underpinnings is the default for all twenty-first century theories of grammar, a legacy from the Western Classical Tradition. From the modistae to Chomsky By the ninth century there was already reawakened interest in the rationale for the parts of speech. This eventually flourished during the late middle ages (1100–1350) in the era of speculative grammars which re-established grammatical theory within the Western Classical Tradition. Unlike the grammars of late antiquity and the early middle ages, speculative grammars sought to explain the ontology of the parts of speech. They were concerned not so much with the vox, the dictiones or partes orationis per se, but with modi significandi; that is, with our understanding of language as human beings and consequently with grammatical universals in general grammar. However, we should not assume that the word-and-paradigm grammar of the early middle ages was simply cast aside; the pars orationis came to be seen as a minimal sentence constituent and the focus (except in pedagogical grammars) was on syntactic construction rather than parsing. A constructio needs to have congruitas (wellformedness); a well-formed sentence is perfectio. Such grammatical constructs came to be reinterpreted in terms of universals. The late middle ages, then, is the first period at which the notion of a general or universal grammar is made explicit. Accordingly, the modistae concerned themselves primarily with modes of being or motion, understanding, and signifying. The essential modes identify what today we would call grammatical or lexical categories, each of which has accidental modes or subcategories such as number, gender, and case for nouns. Morphological form and structure was irrelevant to the semantic definitions of the modistae. The modistae did not aim for descriptive adequacy or pedagogic function, but for what Chomsky would call ‘explanatory adequacy’. Their aim was to produce a theory of syntax that characterized the essential properties of grammar even when describing the grammar of a particular language, because the particular grammar could be more easily learned through the rational explanations derived from general grammar. This was also the aim of the rationalist grammarians of the seventeenth and eighteenth centuries ‘General’ or ‘universal’ grammar: from the modistae to Chomsky 193 and later of Chomskyan UG. One great difference was that these later rationalist theories also sought descriptive adequacy that led them to discuss processes for deriving surface from deep structures, or in more recent terminology to identify not only the principles of UG, but also explain how the parameters get switched on. For the modistae, the modus significandi was based on a concept of the res “thing” and its properties; but unlike the rationalists of the seventeenth and eighteenth centuries, the starting point was the world of experience rather than the constructs in the mind. The mind understands the res and bestows on it a linguistic formulation in its mode of signifying. Thus, a word denotes, but to signify functionally in a construction it must consignify this structural ability. This use of consignificatio shows that full meaning includes the syntacticfunctional meaning of the linguistic correlate of reality. The modistae saw the verb as indicating a change of state in the subject rather than as identifying action or passion. Active and passive came to be seen as subcategories of the change of state. Like Aristotle, the modistae recognize that the verb ‘consignifies’ tense. They regarded the verb as separated from the noun by being active (using terms such as fieri, fluxus, successio) as against stative (habitus, permanens). The participle was seen as adhering to the noun, and sharing some of its features, rather like an adjective. Like Aristotle (Topics 142a 20), Apollonius (Synt. II: 4), and Priscian (Inst. XVII: 14), the modistae believed that the verb is dependent on and succeeds the noun. The priority of the noun justifies their view that the verb is grammatically dependent upon the noun. The syntactic consequence is a dependency hypothesis quite different from any modern dependency theories, in all of which the verb governs nouns (e.g. Hudson 1984; Mel'?uk 1988; Tesnière 1965). Dependency in modistic grammars was a semantic rather than a syntactic category, as one can see from Figure 8.7, in which the head noun of Virgil’s book is dependent: it is dependent because Virgil is the author of the book, and necessarily prior to it. During the renaissance the hegemony of Latin diminished. There was a gradual recognition that the vernacular languages of Europe were not so inferior as to make them unworthy of creative literature or of grammatical study. Also grammars of languages other than the linguae sacrae, Greek, Latin, and Hebrew, came to be written. Concomitantly, interest grew in the development of an international language to replace Latin, but this time an idealized language constructed on logical principles that would somehow directly and unambiguously represent (ideographically) the objects of perception and conception; it would be quick and easy to acquire, and facilitate both thinking and harmony among mankind. The universal understanding of Arabic numerals, despite the fact that they were constructio dependent terminant dependent terminant lego librum Vergilii “I read Virgil’s book” Figure 8.7. Virgil’s book. 194 The Western Classical Tradition in Linguistics pronounced differently in every language, was the ideal model for a ‘character’ to represent this universal or ‘philosophical’ language. Paradoxically, these notions combined with an interest in shorthand, or characterie, also using atomized meaning. To create a ‘philosophical’ language one needed to systematically categorize human knowledge and perhaps the nature of human reason. The most celebrated examples were Dalgarno’s Ars Signorum and Wilkins’ Essay Towards a Real Character and a Philosophical Language, both published in the 1660s. These were magnificent failures in that they were not adopted for general use. Their legacies are dictionaries, thesauri, nineteenth century shorthand, the International Phonetic Alphabet, various systems of logic, and the componential analysis of vocabulary. The interest in a universal ‘philosophical’ language was closely allied to the interests of contemporary rationalist grammarians who were the heirs to the general grammars of the late middle ages. The rationalists took up Aristotle’s observation that human minds have a common understanding, combined it with Descartes’ belief that God would not trick human reason, and sought a rational explanation for linguistic categories which are used to represent the world. They believed in the logical basis for language categories and offered systematic explanations for the structure of propositions and for relations between propositions. Inspired by Sanctius, the ‘gentlemen of Port-Royal’ created the seminal work in their Grammaire Generale et Raisonée. Like the modistae before them, they held that general grammar was a common property of all human languages; particular languages merely differ ‘accidentally’. Echoing the anonymous thirteenth century cleric quoted earlier who wrote that ‘It is not the grammarian but the philosopher […] who discovers grammar’, César Chesnau du Marsais (1676–1756) defined a grammarian as one who needs precise judgment and a philosophical approach enlightened by logic (Marsais 1797, V: 299). It is well-known that Chomsky identifies the Port-Royal rationalists as his own intellectual forebears, and certainly his version of universal grammar is a clear successor to the general grammars of the modistae and the rationalists. Their ‘general’ grammar is reinterpreted as the ‘principles’ of UG, and their ‘accidents of grammar’ as the ‘parameters’ that cause the learner of a particular language to apply certain principles from UG but not others. In pretwentieth century grammars that distinguish deep from surface structure what is lacking is the specification of rigorously stated mapping rules from deep to surface. Also elaborations of simpler surface structures did not use abstract forms in the metalanguage but ordinary words from the object-language; and for everyone except Apollonius, the proposed underlying form was grammatically possible but stylistically (pragmatically) unacceptable. Chomskyan UG moves beyond the mind, beyond the pure rationalism of the seventeenth and eighteenth century grammarians to the ‘human biological endowment’ of the ‘language faculty’. This is a far stronger assertion of the innateness of UG than can be found among Chomsky’s predecessors in the Western Classical Tradition: it constitutes a step from rationalism towards neurological linguistics – a step that is a promise yet to be fulfilled.