Derivational networks in French

Vincent Renner 23 Derivational networks in French 23.1 General notes French commonly uses the morphological process of affixation in the formation of complex words, and derivational affixation is recognized to have three functions – a transpositional, a lexicon-expanding, and an evaluative function, to follow Laurie Bauer’s (2004) terminology. Transpositional affixes aim at changing the lexical class of the base word, as in (1); lexicon-expanding affixes modify the denotational meaning of the base, as in (2); and evaluative affixes alter the connotational meaning of the base, as in (3): (1a) musée ‘museum’ + -al ‘-al’ > muséal ‘museum-related’ (1b) mondial ‘worldwide’ + -ité ‘-ity’ > mondialité ‘worldwideness’ (2a) abricot ‘apricot’ + -ier ‘tree’ > abricotier ‘apricot tree’ (2b) a- ‘a-’ + mitose ‘mitosis’ > amitose ‘amitosis’ (3a) chouette ‘nice’ + -os ‘-y’ > chouettos ‘nicey’ (3b) lapin ‘rabbit’ + -ou ‘-y’ > lapinou ‘bunny’. Establishing a derivational network markedly differs from grouping together the items of a word family. The outputs of affixation were included only if they appeared in one of the two largest standard general language dictionaries of Hexagonal French – the Grand Robert de la Langue Française (GRLF) and the Trésor de la Langue Française – or if they had been attested at least twice in reliable contexts returned by online search engine queries. A second limitation is that they were retained only if at least one 20th/21st-century output illustrating the same derivational pattern was listed in the GRLF. The present overview of the affixal capacity of French is thus slightly conservative given that the 30 simplex items of the core lexicon which constitute our study sample are centuries-old (most of them date back to the 11th and 12th centuries according to the GRLF) and that a number of their derivatives were institutionalized at a time when some affixes were still available, but which are not in the present-day state of the language. This is, for instance, the case for the deadjectival nominalizing suffixes -eur and -esse, which are both semantically equivalent to the English ‘-ness’ and appear in derivatives like chaleur ‘hotness’, longueur ‘length’, minceur ‘thinness’, épaisseur ‘thickness’, vieillesse ‘oldness’, and étroitesse ‘narrowness’. https://doi.org/10.1515/9783110686630-023 230 Vincent Renner Due to limitations of space, only one full derivational network is illustratively provided below (Figure 23.1). The network is structured around the simplex base form. All the 1st order derivatives stemming from the original base form are listed in the second column and each 2nd order derivative is listed on the same line as its 1st order base. Base form 1st order derivatives (with semantic category) coup(er)a coupage; coupement (ACTION) 2nd order derivatives (with semantic category) coupage; coupement (RESULTATIVE) coupeur; coupeuse (AGENT) coupeuse; coupoir (INSTRUMENT) coupailler; coupasser (PEJORATIVE) recouper (ITERATIVE) recoupage; recoupement (ACTION) précouper (TEMPORAL) précoupage (ACTION) surcouper (AUGMENTATIVE) surcoupage (ACTION) coupable (ABILITY) incoupable (PRIVATIVE); recoupable (ITERATIVE) Figure 23.1: Derivational network for the verb couper ‘to cut’. a The -er infinitive suffix is inflectional and is thus disregarded in the derivational analysis. 23.2 Maximum derivational networks To measure the structural richness of affixation, a virtual maximum derivational network can be computed for each order of derivation in each wordclass, as shown in Table 23.1. 23.3 Saturation values The degree of saturation, or relative saturation value, is presented in Tables 23.2–23.4 for each item of the three word-classes. In the class of nouns, there is a very wide gap between the highest and the lowest total saturation value (60), with a remarkable 0 value for two items: feu ‘fire’ and eau ‘water’. This dramatically illustrates the frequent character of suppletion in French, 231 23 Derivational networks in French Table 23.1: Maximum derivational networks per order of derivation for all three word-classes. Word-class st order nd order rd order Σ Nouns     Verbs             Adjectives TOTAL Table 23.2: Saturation values per order of derivation, nouns. Noun Saturation value (%) bone os eye œil tooth dent day jour . dog chien  louse pou fire feu stone pierre water eau  name nom   .  .  . st order (%) . . nd order (%) rd order (%) .    . .  . .    . .  .  . .    .      which has been repeatedly stressed in the linguistic literature (see e.g. Meillet 1913: 389). The adjective related to feu ‘fire’ is igné ‘igneous’ (from the Latin igneus); those related to eau ‘water’ are aqueux ‘aqueous’ (from the Latin aquosus) and hydrique ‘hydric’ (from the Greek hudōr). A majority of base nouns have a low 1st and 2nd order degree of saturation (under 20%), highlighting a somewhat limited general paradigmatic capacity. In the verb category, there is a narrower gap between the extreme total saturation values (about 43%), but such a number still bears witness to a considerable heterogeneity of behaviours. 232 Vincent Renner Table 23.3: Saturation values per order of derivation, verbs. Verb Saturation value (%) st order (%) nd order (%) rd order (%) cut couper . . .  dig creuser . .   pull tirer  . .  throw lancer . . .  give donner . . .  hold tenir . . . sew coudre . .   burn brûler . . .  drink boire . . . . know savoir . . . . . Table 23.4: Saturation values per order of derivation, adjectives. Adjective Saturation value (%) st order (%) nd order (%) bad mauvais . .  new nouveau . .  black noir . .  straight droit . .  warm chaud . .  old vieux . .  long long . .  thin mince . .  thick épais . .  narrow étroit . .  23 Derivational networks in French 233 In the class of adjectives, the gap between the extreme total saturation values is again remarkably large at about 67%, but there is less overall heterogeneity than in the case of nouns and verbs, as 7 of the 10 adjectives have a value of 44.44% or 55.56%. In Table 23.5, a bird’s-eye view of the average saturation values per order of derivation is displayed, and it can again be underlined that the various percentages are fairly low, strikingly so in the case of nouns, mainly because of the commonplaceness of suppletion (e.g. œil ‘eye’ ~ oculaire ‘ocular’; chien ‘dog’ ~ canin ‘canine’; nom ‘noun’ ~ nominal ‘nominal’). Table 23.5: Average saturation values per order of derivation for all three word-classes. Word-class st order (%) nd order (%) rd order (%) Nouns . .  Verbs . .  Adjectives .   23.4 Orders of derivation In Table 23.6, the variation in the number of attested orders of derivation is featured. Only verbs and nouns display three orders of derivation. Derivational richness is, however, somewhat limited in the case of verbs – nine items reach the 2nd order of derivation, but only three allow 3rd order derivation (e.g. boire ‘to drink’ > buvable ‘drinkable’ > imbuvable ‘undrinkable’ > imbuvabilité ‘undrinkability’) – and it is minimal for nouns, as it affects only one item, os ‘bone’, whose 3rd order derivatives are all technical terms from the domains of biology and medicine Table 23.6: Maximum and average number of orders of derivation for all three word-classes. Word-class Maximum Average Nouns  . Verbs  . Adjectives  . 234 Vincent Renner stemming from ossification (e.g. os ‘bone’ > ossifier ‘ossify’ > ossification ‘ossification’ > surossiffication ‘overossification’). As for the class of adjectives, only two orders of derivation are attested and only half of the base words allow 2nd degree derivation (e.g. long ‘long’ > longuet ‘longish’ > longuettement ‘longishly’). 23.5 Derivational capacity In Table 23.7, 1st order derivatives are examined cross-categorially. The class of adjectives exhibits the most homogeneity, with variation only between 2 and 6 outputs (respectively for the base adjectives droit ‘straight’ and noir ‘black’) and the maximum proportional deviation from the average is measured for nouns. In contrast to feu ‘fire’ and eau ‘water’, which have no derivatives, dent ‘tooth’, for instance, generates 5 nominal outputs (dentée ‘bite’, dentier ‘dentures’, dentine ‘dentine’, dentiste ‘dentist’, and surdent ‘supernumerary tooth’), as well as 3 adjectival and 2 verbal outputs (respectively denté ‘toothed’, dentaire ‘dental’, dental ‘dental’, édenter ‘to deprive of teeth’, and endenter ‘to tooth’). Table 23.7: Maximum and average derivational capacity for all three word-classes. Word-class Maximum Average Nouns  . Verbs  . Adjectives   In Table 23.8, a comparison of the numbers of derivatives in the different orders of derivation is drawn. If null and quasi-null average values are set aside, the most striking contrast opposes the class of nouns, which counts more 2nd order than 1st order derivatives, to that of verbs, which counts far fewer 2nd order than 1st order items. For denominal 2nd order derivatives, the distribution can be explained by the remarkable profitability of prefixation with the 2 adjectives osseux ‘osseous’ and dentaire ‘dental’ (e.g. LOCATIVE interosseux ‘interosseous’, endodentaire ‘endodental’, QUANTITIVE uniosseux ‘uniosseous’, bidentaire ‘two-tooth’, SIMILATIVE pseudo-osseux ‘pseudo-osseous’, pseudo-dentaire ‘pseudo-dental’, AUGMENTATIVE hyperosseux ‘hyperosseous’) in the specific context of scientific (biological/medical) terminology. For deverbal 2nd order derivation, profitability 23 Derivational networks in French 235 Table 23.8: Average number of derivatives per order of derivation for all three word-classes. Word-class st order nd order rd order Nouns . . . Verbs . . . Adjectives  .  is mostly restricted to the addition of the ITERATIVE re- ‘re-’, AUGMENTATIVE sur‘over-’ and PRIVATIVE in- ‘un-’ (e.g. relancement ‘relaunch’, surcreusement ‘overdeepening’, indonnable ‘ungivable’), which may partly explain the comparatively low average value. 23.6 Correlation between semantic categories and orders of derivation Turning to the distribution of individual affixes and semantic categories, it appears that their relative profitability varies substantially from one word-class to another. For nouns, it is striking to note that no single affix is used with half of the 10 base words. In contrast, a majority of verbs take the 1st order ITERATIVE re- ‘re-’, AGENT eur/-euse ‘-er’ and ABILITY -able ‘-able’. In the 2nd order of derivation, the PRIVATIVE prefixation in- ‘un-’ is remarkably profitable as well as it applies to all the ABILITY deverbal adjectives (e.g. lancer ‘to throw’ > lançable ‘throwable’ > inlançable ‘unthrowable’). As for the class of adjectives, it stands out in that it exemplifies the only cases of full 1st order profitability – the 10 adjectival bases realize AUGMENTATIVE -issime ‘extremely’ and MANNER -ment ‘-ly’ – and also through the fact that all of its 8 2nd order derivatives realize MANNER -ment (e.g. long ‘long’ > longuissime ‘extremely long’ > longuissimement ‘extremely longly’). Affixal rivalry may lead to the co-presence of full synonyms, as in (4): (4a) creusage ~ creusement ‘digging’; empierrage ~ empierrement ‘stone surfacing’ (ACTION and RESULTATIVE); (4b) édentement ~ édentation ‘lack of teeth’ (RESULTATIVE); (4c) coupasser ~ coupailler ‘to cut in an irregular fashion’ (PEJORATIVE); 236 Vincent Renner (4d) chiennerie ~ chiennaille ‘kennel of dogs’ (COLLECTIVE); (4e) demi-long ~ semi-long ‘half-long’ (SIMILATIVE). It may also lead to some degree of specialization of the competing affixes, as illustrated by the INSTRUMENT suffixes and derivatives in (5): (5a) -ette, in tirette ‘bellpull’, tenette ‘lithotomy forceps’; (5b) -oir, in coupoir ‘cutter’, creusoir ‘luthier’s digger’, cousoir ‘sewing press’, brûloir ‘roasting machine’; (5c) -eur, in lanceur ‘launch vehicle’, brûleur ‘burner’; (5d) -euse, in coupeuse ‘cutting machine’, tireuse ‘photographic printing machine’. INSTRUMENT -ette and -oir are used in nouns that typically denote implements, while -eur and -euse appear in nouns that typically denote machines. 23.7 Semantic categories with blocking effects One remarkable limitation has been identified: MANNER (i.e. adverbialization in -ment) is the only semantic category with a blocking effect on deadjectival derivation. 23.8 Typical combinations of semantic categories The co-presence of ABILITY, PRIVATIVE and STATIVE constitutes the only remarkably recurrent combination of semantic categories cross-categorially. Each of the 8 attested ABILITY deverbal adjectives generates a 2nd order PRIVATIVE output, and 5 of these items then allow 3rd order STATIVE derivation (see imbuvabilité ‘undrinkability’ in section 23.4 above). 23.9 Multiple occurrence of semantic categories No remarkable multiple occurrences of the same semantic category have been noted in the derivational networks of French. 23 Derivational networks in French 237 23.10 Reversibility of semantic categories No remarkable reversibility of the ordering of semantic categories has been noted in the derivational networks of French. 23.11 Reasons for structurally poor derivational networks As pointed out in section 23.3, a primary reason that explains why some derivational networks are relatively poor is suppletion. It can also be stressed that French commonly resorts to other lexicogenetic strategies: conversion is used for transpositional purposes to nominalize verbs (6) and adjectives (7); compounding is employed for lexicon-expanding purposes (8); and replication is utilized for evaluative purposes (9): (6) coup(er) ‘to cut’ > coupe ‘a cut’; (7) mauvais ‘bad > (the) bad’, nouveau ‘new > (the) new’; (8a) lance-satellites (lit. ‘launchV-satellites’) = lanceur ‘launch vehicle’; (8b) odontologie ‘odontology’ = dentisterie ‘dentistry’; (9) chien ‘dog’ > chien-chien ‘doggy’; os ‘bone’ > nonos ‘bone [+ evaluative].1 More broadly, French also frequently has recourse to non-morphological means to form new lexical units. This can, for example, be captured in the following synonymous pairs, which contrast an affixed form with an adjective-noun or nounadjective construct (10), a noun-preposition-noun construct (11), and a simplex form (12): (10a) surnom ‘nickname’ ~ petit nom (lit. ‘little name’); (10b) dentisterie ‘dentistry’ ~ médecine dentaire (lit. ‘dental medicine’); 1 For os, the partial left duplication takes place only after adding a prothetic syllable onset to the VC base. 238 Vincent Renner (11a) tenette ‘lithotomy forceps’ ~ pince à lithotomie (lit. ‘forceps to lithotomy’); (11b) tirette ‘bellpull’ ~ cordon de sonnette (lit. ‘cord of bell’); (12) dentine ‘dentine’ ~ ivoire ‘lit. ivory’. 23.12 Conclusions French can be characterized by three main features: a limited derivational capacity overall (3rd order derivation is only marginally attested), fairly low saturation values in the different word-classes (under 60% for adjectives, under 40% for verbs, and under 20% for nouns), and wide gaps between individual saturation values in all word-classes and all orders of derivation. References Bauer, Laurie. 2004. The function of word-formation and the inflection-derivation distinction. In Henk Aertsen, Mike Hannay & Rod Lyall (eds.), Words in their Places: A Festschrift for J. Lachlan Mackenzie, 283–292. Amsterdam: Vrije Universiteit. Grand Robert de la Langue Française, digital edition (v. 4.1). https://grandrobert.lerobert.com/ (accessed 9 March 2020). Meillet, Antoine. 1913. La crise de la langue française. Revue Politique et Littéraire – Revue Bleue 51 (2). 385–390. Trésor de la Langue Française, digital edition (v. 4). http://atilf.atilf.fr/dendien/scripts/tlfiv4/ showps.exe?p=combi.htm (accessed 9 March 2020).