Gender of Abstract Noun Suffixes in the Brittonic Languages

The gender of absTracT noun suffixes in The briTTonic languages1 britta irslinger abstract: This article examines the gender of suffixes used for the derivation of abstract nouns in the brittonic languages Welsh, breton and cornish. Their diachronic development is discussed against the background of several functional gender hypotheses. The theory underlying such hypotheses postulates that gender plays a role in the classification of the nominal lexicon, i.e. it classifies nouns according to the feature [±particularized] thus expressing nominal aspect. The effects of this classification can be observed in abstract nouns in Modern high german, but this feature is assumed to be ultimately inherited from Proto-indoeuropean. although all brittonic languages display remarkable similarities with regard to their nominal morphology, the gender of suffixes and the restructuring which each language underwent vary considerably among them. The association of feminine (and neuter) gender with abstractness had probably already been lost in Proto-brittonic. abstract nouns formed through suffixation are predominantly masculine in Welsh and cornish, while breton has a higher proportion of feminine suffixes largely due to french influence. in addition, breton shows the emergence of a functional gender system within some parts of its lexicon. 1 functional gender Theories functional gender theories do not hold that nominal gender is either functionless or that its function is restricted to mere reference-tracking.2 on the contrary, they assume that gender plays a role in the classification of the nominal lexicon. The features determining this classification are still being discussed. unterbeck (2000: xviii, xxvi) proposes the following distinction: systems with noun classes, numeral classification and verbal classification display a classification system based on the features [+count/mass] and [+shape], 1 i owe thanks to claudia letizia, M.a. for the linguistic correction of this text. any remaining mistakes are, of course, my own responsibility. 2 cf. e.g. corbett 2000: 320ff. in the Modhg sentence Maria fotografierte Tobias vor dem Haus als er/sie/es 10 Jahre alt war ‘Maria photographed Tobias in front of the house when he/she/it was ten years old’, gendered third person pronouns allow one to distinguish between three potential antecedents. 58 britta irslinger whereas in gender systems the relevant features are [+count/mass] and [+sex]. The layman tends to regard [+sex] as the main feature and any other classifying feature seems to go largely unnoticed, at least on a conscious level, by speakers of gender languages. Therefore, as of today, these features have not yet been fully recognized and the question of whether they are cross-linguistically comparable or even identical has not been adequately addressed. Weber (2000: 496) suggests that gender has a separate function that cannot be fullfilled by any other grammatical category. only the existence of such separate function could explain the persistence of nominal gender in so many languages. having surveyed a number of examples from languages belonging to different language families, Weber concludes that the function of gender lies in the “perspectivization” of nouns with regard to nominal aspect,3 more precisely with regard to the feature [±particularizing] (Weber 2000: 506, 2001: 113ff.). This feature encompasses the differences implied by more traditional terms such as count nouns, mass nouns, collectives, concrete or abstract nouns, which express conceptual notions regarding [±boundedness], [±internal structure], [±delimitability] and [±divisibility]. Vogel (2000: 468) calls these features “quantifying” and subsumes them under the contrasting concepts of “individuality” and “continuativity”. The feature [+count] is linked to the principle of individuality, whereas the feature [–count] is linked to the principle of continuativity. both concepts, which represent the opposite ends of a scale, are based on cognitive perception and have a strong influence on the nominal lexicon of any language. nouns are conceptualized as individual or continuative nouns and these conceptualizations determine their morphological and syntactic properties. Moreover, their conceptual properties can be altered, i.e. speakers can shift from one conceptualization to another in order to transfer nouns belonging to one category into the other. examples of such processes will be discussed below (1.1.1). according to Weber (2001: 80), gender systems involve the binary opposition [±particularizing], i.e. they classify nouns according to the presence or absence of the feature “particularized”. This is best conveyed by two- 3 The term ‘nominal aspect’ was created in allusion to ‘verbal aspect’: cf. benzing 1955 [1956]: 65, 76, 1955: 28, Pitkin 1984: 202f., rijkhoff 1991, 2002, Weber 2001: 120. for the semantic categories of the noun phrase see Jackendoff 1991: 20, grandi/scalise 1999: 87, corbett 2000: 80. gender of abstract noun suffixes in brittonic languages 59 gender systems, while in systems with three or more genders the semantic differentiation with regard to this feature is not motivated as strongly.4 a growing body of evidence supports the idea that gender is linked with the feature [±particularizing] in indo-european languages.5 as the evidence discussed so far is mostly related to the count-mass distinction, it concerns a rather limited part of the nominal lexicon. in contrast to this, Vogel (1996) and (2000) demonstrates that major parts of the Modhg lexicon, including abstract nouns, are correlated with the genders masculine, feminine and neuter and are thus motivated.6 in the following section, the Vogel’s 2000 findings will be summarized. 1.1 1.1.1 Correlations between Word-Classes and Genders in Modern High German Individuality vs. Continuativity The morphological and syntactic differences between count nouns and mass nouns have been amply demonstrated for various languages. in Modhg, count nouns form a plural, which may be combined with a numeral. They can also be preceeded by an indefinite article or a pronominal adjective such as jed-, viel-, all-, einig-, solch-:7 (1) Häuser 5 Häuser ein Haus jedes Haus ‘houses’ ‘5 houses’ ‘a house’ ‘every house’ When a mass noun is used in such constructions, it undergoes a shift in meaning. The noun becomes individualized as a “kind”, a “sort” or a “part”: 4 Weber 2001: 80 claims that gender systems are originally binary. This implies that systems with three or more genders developed from two-gender systems. Weber does not discuss the alternative hypothesis that such systems might have developed from more complex systems, such as noun classes. 5 cf. irslinger 2010a for literature. 6 classificational effects of gender in other parts of the Modhg lexicon are discussed e.g. by di Meola 2007, Köpcke/Zubin 2009. 7 all examples and their translations are from Vogel 2000: 469ff. as Vogel 2000: 489, fn. 2 points out, a translation of the examples into english is sometimes difficult, since the conceptualization, i.e. the specific combination of formal and semantic properties, varies from language to language. ungrammatical german examples are marked by ‡, ungrammatical english translations are put into square brackets. 60 britta irslinger (2) Wässer/Wasser 5 Wässer/Wasser ein Wasser viele Wässer/Wasser ‘kinds of water / waters’ ‘5 kinds of water / 5 waters’ ‘a (kind of/glass of) water’ ‘many waters’ certain constructions, however, are restricted to either count nouns or mass nouns. While the latter may be used without any article in the singular, the former may not: (3) Wasser ist lebensnotwendig. ‘Water is vital’ ‡ Baum ist teuer. [Tree is expensive] but: Ein Baum ist teuer. ‘a tree is expensive’ if count nouns appear in such constructions, they are recategorized as mass nouns. in example (4), which is highly marked, the count noun U-Bahn undergoes a secondary continuativization or de-individuation. Measure constructions cause continuativization in the case of count nouns, but individuativization in the case of mass nouns (5): (4) Noch ‘even mehr U-Bahn more underground (5) ein Stück Apfel ‘a piece of apple’ 1.1.2 vs. ab 28. Mai. after May 28th’ ein Glas Wasser ‘a glass of water’ Individuality vs. Continuativity with Regard to Abstract Nouns The nominal lexicon of a language consists of a primary and a secondary part. The primary lexicon comprises denotations for entities in the real world like man, dog, water. Words belonging to the primary lexicon are normally not formed with the help of suffixes, i.e. speakers can not analyse them with regard to word-formation. in the lexicon of most indo-european languages, count nouns outnumber mass nouns. Therefore, indo-european languages are “individuality-centered”, i.e. the feature “individuality” is normally inherent and not marked by suffixes (Vogel 2000: 474). The major part of the secondary lexicon consists of nouns denoting abstract concepts. abstract nouns can be subdivided into two different types, as they are either created through conversion or through derivation. conversion involves turning words which originally belong to word-classes other than nouns into nouns, without resorting to any morphological gender of abstract noun suffixes in brittonic languages 61 modification. as the resulting abstract nouns do not contain any suffixes, they are externally unmarked and thus resemble primary lexicon nouns. on the contrary, abstract nouns created by derivation do contain suffixes. as a result, they are morphologically more complex than nouns belonging to the primary lexicon. This morphological complexity iconically mirrors their conceptual complexity. They are overtly marked as members of the secondary lexicon. These two types of abstract nouns have different properties with regard to the continuum individuality vs. continuativity: they differ with regard to their ability to be pluralized and to be used in those syntactic contexts typically associated with count nouns and mass nouns respectively. externally unmarked masculine abstract nouns8 are secondary copies of primary individual nouns which present a verbal action as a “definite whole”. They cannot occur without an article, which is normally indefinite. Their plural is a distributive individual plural: (6) ein Hieb ‘a stroke’ Hiebe ‘strokes’ (7) ein Auftritt ‘an appearance’ Auftritte ‘appearances’ ‡ Auftritt ist gekonnt. [appearance is well-done.] substantivized infinitives and adjectives expressing collective and abstract meanings are located at the opposite end of the scale. like mass nouns, they occur without articles, definite or indefinite (8, 9). The fact that no indefinite article can be used with such nouns and that they cannot be pluralized indicates that they express continuativity (10, 11). (8) Gutes ist angenehm. ‘good things are pleasant.’ (9) Laufen ist anstrengend. ‘running is exhausting.’ (10) ‡ Ein Gutes ist passiert. [something good has happened.] ‡ Die Guten sind passiert. [The good happened.] (11) ‡ Ein Laufen habe ich gesehen. [i saw running.] ‡ Die Laufen habe ich gesehen. [i saw the runnings.] 8 Vogel 2000: 475 calls them ‘conversion nouns’. They are also known as zero-derivatives or back-formations (Rückbildungen), because they look like primary nouns in spite of being derived from verbs: cf. fleischer/barz 1995: 51. 62 britta irslinger abstract nouns which are formally marked as feminine represent points in the middle of the scale, as they have affinities to both ends. deverbal abstract nouns in -ung are closer to [+individuality], while deadjectival abstract nouns in -heit and -keit which denote qualities are closer to [+continuativity]. abstract nouns in -ung can be used without the article when pluralized, but otherwise have to be combined with the indefinite article (12, 13). a use without the indefinite article is however possible in the case of abstract nouns in -heit and -keit (14, 15): (12) Ich habe heute eine Prüfung/Prüfungen ‘i will take an exam / exams today.’ (13) ‡ Prüfung ist unangenehm. [exam is disagreeable] (14) Krankheit ist kein Hinderungsgrund. ‘illness is no excuse.’ (15) Müdigkeit ist unangenehm. ‘Tiredness is unpleasant.’ The examples Krankheit and Müdigkeit illustrate that neither of them can be pluralized and that Müdigkeit also cannot occur with an indefinite article (16). Krankheit can only be used in this way if it refers to ‘a kind of disease’ rather than ‘an event of desease’ (17, 18): (16) ‡ eine Müdigkeit [a tiredness]9 ‡ Müdigkeiten [tirednesses] (17) ich hatte als Kind nur eine Krankheit. ‘as a child i had only one disease.’ (18) Als Kind hatte ich viele Krankheiten. ‘as a child i had many diseases.’ correlations of gender and types of abstract nouns may be summarized in the following table. although the categorization suggested by Vogel (2000) covers only a part of the Modhg system, it can be applied more extensively to the whole of Modhg productive derivational patterns for the formation of abstract nouns. suffixes for the derivation of denominal and deadjectival 9 cf. however expressions like Eine große Müdigkeit überkam sie. gender of abstract noun suffixes in brittonic languages individuality Masculine deverbal drehen ‘to turn’ der Dreh laufen ‘to walk’ der Lauf individual event deadjectival die Drehung die Dreherei die Lauferei die Krankheit müde ‘tired’ die Müdigkeit deverbal continuativity neuter das Drehen das Laufen individual event a kind of repeated/ event continuous event gut ‘good’ krank ‘ill’ das Gute Masculine german feminine 63 Ø / backformation -s -er feminine neuter -ung -e -ei, -erei substantivized infinitives denominal -heit (< heit m.) -keit, -igkeit -ei, -elei, -erei substantivized adjectives collectives -schaft -ei, -erei Ge- . . . -e -icht -werk -wesen -tum abstract nouns and nouns denoting qualities are exclusively feminine, whereas suffixes forming collectives are feminine and neuter. The correlation between gender and nominal aspect in Modhg has thus a twofold function. it differentiates the primary from the secondary lexicon, and with regard to nouns belonging to the latter, it indicates different positions on the individuality-continuativity continuum. french, belonging to the romance languages which have lost neuter as a gender category, displays a similar distribution of suffixes. This supports the hypothesis that the tendencies at work might not be restricted to Modhg alone, but that they are of a more general nature. 64 britta irslinger Masculine french feminine deverbal Ø / backformation substantivized infinitives -age -(e)ment -ation, -cation, -ition denominal substantivized adjectives -esse -eur (orig. m.) -(er)ie -(e)té, -ité -ance, -ence -ion collectives -isme -at 1.2 Diachronic Explanation10 according to Vogel, this correlation is principally inherited from Pie, where masculine gender is associated with individuality, neuter with continuativity and feminine gender with collectivity and abstractness.11 The suffix *-(a)h2- which later became associated with the expression of female sex and neuter plurals originally created deverbal abstract nouns and collectives. The original function(s) of *-(a)h2- as well as the individual steps of its development into a marker of feminine gender are still controversial and will not be discussed here, as they are not relevant with regard to Vogel’s argument. according to Vogel, the correlation of the features individuality/definiteness vs. continuativity/indefiniteness with masculine and feminine/neuter genders respectively are accidental and could just as well be reversed. The crucial point is that in both cases alternative oppositions are contrasted (Vogel 2000: 487). Vogel’s hypothesis concerning the functions of gender is supported by the association between gender and different types of abstract nouns in Modhg. furthermore, these functions are motivated synchronically as well as diachronically. The gender system of Modhg seems to be 10 cf. irslinger 2010a for the discussion of recent research. 11 summaries of previous research starting with the 19th century are given by Vogel 1996: 147ff., 2000: 485ff. and Weber 2001: 15ff. gender of abstract noun suffixes in brittonic languages 65 extremely conservative as it displays at least partially the gender semantics assumed for Pie. but what is the significance of these findings with regard to the gender systems in other ie languages? are the Modhg correlations of gender and perspectivizing features unique and isolated or are they paralleled by similar developments elsewhere? Modern english for example, although in close proximity and closely related to Modhg, contrasts strongly with the latter as it has lost nominal gender altogether. This fact has been adduced to support the assumption that grammatical gender in modern ie languages is largely functionless or opaque (e.g. Trudgill 1999, Trudgill in Kilarski/Trudgill 2000). an evaluation of these extreme positions requires the detailed synchronic and diachronic analysis of a greater number of gender systems. The following section will inquire into whether similar phenomena can be found in brittonic languages. 2 The gender of abstract noun suffixes in the brittonic languages 2.0 The Brittonic Languages brittonic or british celtic12 is a branch of insular celtic, attested in britain from the first century bc onwards. its earliest records consist of personal and tribal names in latin sources. The names contained in the earliest latin charters of the book of llandaf (mid 6th or early 7th cent.) do not show any features typical of Welsh, breton or cornish and may therefore be regarded as identical to the reconstructed late Proto-brittonic. around 800 ad, late Proto-brittonic branched into West-brittonic, which then became old Welsh, and southwest-brittonic, which further sub-divided into breton and cornish from about 1100 onwards. in the 18th century cornish became extinct. 2.0.1 Gender Marking and Agreement in the Brittonic Languages Welsh, breton and cornish all have two grammatical genders: masculine and feminine. in all three languages, there are more masculine than feminine nouns. nouns may be overtly marked for gender through the 12 cf. schrijver 2011: 2ff. 66 britta irslinger use of suffixes. exceptions to this rule occur mainly with unproductive or opaque patterns and are due to language change, particularly in those instances where two originally distinct suffixes became homonymous. each brittonic language possesses a few suffixes whose derivatives adopt the gender of the base noun. These will not be considered here. gender is further indicated by the agreeing forms of pronouns, demonstratives and the cardinal numbers ‘two’, ‘three’ and ‘four’, as well as by the presence or absence of lenition. lenition, i.e. the regular change of a consonant into its lenited counterpart, takes place in the following cases: a feminine singular noun undergoes lenition after the article and one or more attributive adjectives undergo lenition after a feminine singular noun. in Modbr for example, the feminine noun mamm ‘mother’ and the adjective mat ‘good’ are both lenited in the phrase ur vamm vat ‘a good mother’.13 gender is not indicated if the initial of a noun or attributive adjective is not lenitable and not all types of lenition are represented in orthography. furthermore, the indication of lenition in medieval texts is very inconsistent. 2.1 Survey of Derivational Patterns derivational patterns of Modern Welsh and breton, arranged according to gender, function and productivity, are listed in the following tables. This information was collected from the grammars and handbooks indicated in the respective sections. suffixes forming agent nouns as well as homonymous suffixes forming instrument nouns are excluded from the tables, since their grammatical gender is related to natural gender. all other suffixes, including those which form abstract and concrete nouns or concrete nouns only, were however considered.14 The suffixes are grouped according to types of derivational bases, i.e. verbal stems, nouns and adjectives. if a suffix may be used with more than one type of base, it is listed with each of them. Where a cognate verb, an underived noun and an adjective appear side by side, the direction of derivation could not be determined. 13 There are exceptions to this rule: cf. favereau 1997: 151f. in Welsh, lenition of nouns and adjectives in the plural has been abandoned: cf. evans 1964: 14, Thomas 1996: 686. in breton and cornish, it applies only for masculine nouns denoting persons, and thus it may be disregarded here: cf. fleuriot 1964a: 212, hemon 1975: 14, favereau 1997: 151f., lewis 1990: 8, 11. 14 on the semantics of derivatives with W -yn/-en, Mco -en(n) and br -enn see irslinger 2010b. gender of abstract noun suffixes in brittonic languages 67 The tables indicate only the major functions of the suffixes. for a more detailed analysis, the respective grammars and handbooks should be consulted. 2.1.1 Modern Welsh15 in Welsh, masculine suffixes outnumber feminine ones, with regard to all types of bases. Verbal nouns and most suffixes forming action nouns are masculine. in addition, masculine suffixes are very frequent with denominal derivations. There are only three feminine suffixes forming abstract nouns, namely -aeth, -eb and -fa. -fa also forms nouns denoting places. -aeth, with its variant -iaeth, is very productive with any type of base, but its derivatives do not excusively have feminine gender. Morris Jones (1913, 226) gives a list of masculine nouns including claddedigaeth ‘burial, funeral’, darfodeligaeth ‘consumption’, gwasanaeth ‘service, use’, hiraeth ‘longing, nostalgia’, amrywiaeth ‘variety’, gwahaniaeth ‘difference, separation’, lluniaeth ‘food, sustenance’. gender classifications of gPc and gM are sometimes different (feminine or m./f.). The proportion of nonfeminine words is not mentioned in the handbooks. 2.1.1.1 Productive Patterns: Masculine Suffixes deverbal -i -o16 -u -(i)ad -(i)adur -dod/-tod -iant verbal nouns verbal nouns verbal nouns abstract nouns instruments abstract nouns abstract nouns, concrete nouns denominal -(i)adur books, instruments17 berwi ‘to boil’ : berwcludo ‘to carry’ : cludprynu ‘to buy’ : pryngwanhad ‘weakening’ : gwanhau ‘to weaken’ cariad ‘love, charity’ : caru ‘to love’ oeriadur ‘refrigerator’ : oeri ‘to cool’, oer ‘cold’ cryndod ‘trembling’ : crynu ‘to tremble’ cofiant ‘memorial’ : cofio ‘to remember’ geiriadur ‘dictionary’ : geir ‘word’ amseriadur ‘chronometer’ : amseriad ‘timing, tempo’ 15 cf. surridge 1989: 190ff., Thomas 1996: 647ff., Zimmer 2000: 271ff. 16 -o is the most productive suffix for the formation of verbal nouns: cf. Thomas 1996: 669f. 17 cf. Thomas 1996: 655, Zimmer 2000: 282ff. The oldest formations in -adur are borrowed from latin and denote persons such as creadur ‘creature, creator’ < lat. creatūra(m), creatōrem, henadur ‘alderman, elder’ < lat. senatōrem, pechadur ‘sinner’ < lat. peccatōrem. There are also a few hybrid formations denoting persons and agent nouns from celtic stems: cof(i)adur ‘secretary, chronicler’ : cof ‘memory’, cofio ‘to remember’, cysgadur 68 britta irslinger -dod/-tod abstract nouns -edd 18 abstract nouns morwyndod ‘virginity’ : morwyn ‘maid, virgin’ pwylledd ‘discretion, steadiness’ : pwyll ‘discretion, steadiness’ blaeniant ‘priority’ : blaen m. ‘lead etc.’, blaenu ‘to lead’ gwydryn ‘drinking-glass’ : gwydr ‘glass, mirror, etc.’ -iant19 -yn abstract nouns concrete nouns concrete nouns deadjectival -deb -oldeb -dod/-tod -dra -edd -rwydd -wch -yn abstract nouns abstract nouns abstract nouns abstract nouns abstract nouns abstract nouns concrete nouns cywirdeb ‘correctness’ : cywir ‘correct’ gwroldeb ‘manliness’ : gwrol ‘manly, virile’ segurdod ‘idleness’ : segur ‘idle’ caethdra ‘captivity’ : caeth ‘captive’ llonedd ‘cheerfulness’ : llon ‘cheerful’ addasrwydd ‘suitability’ : addas ‘suitable’ tristwch ‘sadness’ : trist ‘sad’ glesyn, glesin ‘common borage, woad’ : glas ‘blue, green’ Productive patterns: feminine suffixes deverbal -(i)aeth (few m.) -eb (few m.) abstract nouns -ell (few m.) abstr. n.: utterances, documents etc. instruments -fa action nouns, places darbodaeth ‘provision’ : darbod ‘to prepare’ anfoneb ‘invoice’ : anfon ‘to send’ estynnell ‘stretcher’ : estyn ‘to reach, stretch’ cnofa ‘gnawing, ache’ : cnoi ‘to bite, to ache’ glanfa ‘landing-place (for ship or aircraft)’ : glanio ‘to land’, glan ‘riverbank, shore’ ‘sleeper, hibernating animal’ : cwsg ‘sleep’, cysgy ‘to sleep’. The modern productive pattern for the formation of deverbal instrument nouns, e.g. awyriadur ‘ventilator’ : awyr ‘air’, awyro, awyru ‘to air, to ventilate’, continues this latter function which is not relevant to the present context. 18 cf. also the feminine noun dannoedd ‘toothache’ : dant ‘tooth’, which Thomas 1996: 651 lists as the only example of a suffix -oedd. 19 according to Zimmer 2000: 394ff. all nouns are derived from verbal bases, while Thomas 1996: 663 assumes also denominal patterns. gender of abstract noun suffixes in brittonic languages 69 denominal -(i)aeth (few m.) functions, sciences -eb (few m.) -eg -ell (few m.) -en -fa tywysogaeth20 ‘principality’ : tywysog ‘prince’ abstract nouns: cofeb ‘memorandum’ : utterances, documents cof ‘memory’, cofio ‘to remember’ collections, various terms telyneg ‘lyrics, poetry belonging to the harp’ : telyn ‘harp’ language terms Almaeneg ‘german language’ : yr Almaen ‘germany’ linguistic terms areitheg ‘speech, meaning of an utterance’ : araith ‘language, speech, utterance’ branches of science bywydeg ‘biology’ : bywyd ‘life’ instruments haenell ‘plate’ : haen ‘layer, seam’ concrete nouns duren ‘steel for striking fire with a flint’ : dur ‘steel’ tywoden ‘sandbank’ : tywod ‘sand’ places gweithfa ‘workshop’ : gwaith ‘work’ deadjectival -(i)aeth (few m.) abstract nouns -eb abstract nouns -en concrete nouns hudoliaeth ‘enchantment’ : hudol ‘enchanting’ fraetheb ‘joke, witticism’ : fraeth ‘fluent, witty’ caleden ‘hardness, corn, callosity, thickened part of skin’ : caled ‘hard’ 2.1.1.2 Non-Productive or Weakly Productive Patterns: Masculine Suffixes Patterns with non-productive or only weakly productive suffixes display an even higher number of masculine suffixes which form abstract nouns from adjectives. in this respect Modern Welsh contrasts sharply with german or french in which such suffixes tend to be feminine. deverbal -ain(t) -ed (also f.)21 verbal nouns, action nouns action nouns verbal nouns diodef(i)aint ‘suffering, pain’ : dioddef ‘to suffer’ syched ‘thirst’ (< ‘dryness’) : sychu ‘to dry’ clywed vn. ‘to hear’ 20 exceptionally masculine: cf. gPc s.v. 21 Zimmer 2000: 345 claims that words in -ed are usually feminine. This is of course not true for verbal nouns in -ed. The majority of the other derivatives listed by Zimmer shows masculine or double gender. Thomas 1996: 666 discusses -ed with suffixes forming words of either gender. 70 britta irslinger -od, -awd (few f.)22 -ol23 abstract nouns periods of time -yd verbal nouns treathawd ‘essay’: traethu ‘to speak’ dyfodol ‘future’ : dyfod ‘to become’ gorffenol ‘past’ : gorffen ‘to finish’ dywedyd vn., m. ‘to speak, speech’ : dywed- denominal -ach (also vn., f., coll.)24 abstract nouns, pejoratives -ain(t)25 verbal nouns -ed (also f.) abstract nouns -od, -awd (few f.) learned neologisms -ol23 abstract nouns -red (also f., m./f.)26 abstract nouns -yd (also f.) abstract nouns deadjectival -ain(t) -der/-ter abstract nouns abstract nouns cyfeillach f. ‘friendship’ : cyfaill ‘friend’ cyfrinach f. ‘secret’ : cyfrin adj., m. ‘secret’ bremain ‘to fart’ : bram ‘fart’ llefain vn., m. ‘outcrying, shout’ : llef f. ‘id.’ lludded ‘exhaustion, fatigue’ : lludd ‘hindrance, obstruction’ anianawd m./f. ‘nature, disposition, temperament’ : anian m./f. ‘nature’ unawd ‘solo (music.)’ : un ‘one’ golygyddol ‘editorial’ : golygydd ‘editor’ gweithred f. (formerly m.) ‘act, deed’ : gweith ‘work’ hydred m./f. ‘longitude’ : hyd ‘length’ gwryd m. ‘manhood’ : gwr ‘man’ henaint ‘old age’ : hen ‘old’ braster ‘grossness’ : bras ‘fat’ 22 gender variation is due to the different etymological origins of the suffixes which have became homonymous: cf. Zimmer 2000: 412ff. 23 These are lexicalizations of substantivized adjectives. 24 according to surridge 1989: 191 and Thomas 1996: 651, -ach forms feminine abstract nouns from nouns and adjectives. This claim is not supported by the collection listed in Zimmer 2000: 274ff. derivatives in -ach may also be masculine nouns, verbal nouns and collectives. feminine derivatives, including cases with double gender, constitute only about a quarter of the material. This is hardly surprising considering that W -ach continues among others the suffixes *-akko/ā-, *-akso/ā-, which originally formed adjectives. Therefore, it does not seem appropriate to list this suffix as feminine. in addition, cyfrinach f. ‘secret’ is the only deadjectival example, since its base cyfrin may also be used as an adjective. 25 Words in -ain(t) with feminine gender have different etymological origins: cf. Zimmer 2000: 295ff. 26 according to Thomas 1996: 651, this suffix is feminine, but for three of the four examples, gender is contested in the handbooks. hydred ‘longitude’ and lledred ‘latitude’ are classified as masculine in gM, while Zimmer 2000: 448, following gPc, regards these words as well as cylchred ‘circuit, orbit’ as masculine/feminine. This is due to the fact that the suffix is undergoing a change from masculine to feminine gender which is not yet completed. gender of abstract noun suffixes in brittonic languages -did/-tid abstract nouns -dwr -fel -i -id -ineb -ni27 -(i)oni -yd (also f.) abstract nouns abstract nouns abstract nouns abstract nouns abstract nouns abstract nouns abstract nouns abstract nouns 71 digalondid ‘discouragement’ : digalon ‘disheartened’ sychdwr ‘dryness’ : sych ‘dry’ oerfel ‘cold, curse’ : oer ‘cold’ diogi ‘laziness’ : diog ‘lazy’ rhyddid ‘freedom’ : rhydd ‘free’ clae(a)rineb ‘indifference’ : clae(a)r ‘lukewarm’ glesni ‘blueness, verdure’ : glas ‘blue, grey, green’ daioni ‘goodness’ : da ‘good’ iechyd m./f. ‘health’ : iach ‘healthy’ non-productive or weakly productive patterns: feminine suffixes deverbal -as (also m.) -ed (also m.) abstract nouns action nouns gwanas ‘wooden peck, hook’ : gwanu ‘to pierce’ adduned ‘vow’ : adduno ‘to desire’ denominal -as (also m.) -ed (also m.) -es -yd (also m.) abstract nouns abstract nouns collective nouns abstract nouns teyrnas ‘kingdom’ : teyrn ‘lord’ oged ‘harrow’ : og ‘id.’ branes ‘flight of ravens’ : brân ‘raven’ celfyddyd ‘art, craft’ : celfydd adj. ‘skillful’, m. ‘expert’ deadjectival -as (also m.) abstract nouns -fa places cymwynas m./f. ‘kindly act, favor’28: cymwyn ‘tender, gentle’ uchelfa ‘high place’ : uchel ‘high’ 2.1.2 Modern Breton29 breton shows significant differences from Welsh, in that the proportions of masculine and feminine suffixes are much more balanced. Masculine suffixes are more numerous with regard to deverbal derivations, whereas the number of masculine and feminine suffixes is equal in the case of denominal derivations. suffixes forming deadjectival abstract nouns are predominantly feminine. 27 cf. Watkins 1961: 100, Williams 1980: 25, surridge 1989: 197, Zimmer 2000: 408ff. not treated by Thomas 1996. 28 feminine according to gM, Thomas 1996: 651; masculine/feminine according to gPc, Zimmer 2000: 302. 29 cf. favereau 1997: 73ff., hemon 1975: 26ff. 72 2.1.2.1 deverbal -Ø -añ -iñ -a -aat -adur -aj, -ach -amant -erezh31 britta irslinger Productive Patterns: Masculine Suffixes infinitives infinitives infinitives denominal infinitives30 deadjectival infinitives actions, concrete results abstr. nouns (pejorative) loan words, popular neologisms actions, their results denominal -eg language terms -erezh actions, their results collective nouns -va places chom ‘to stay’ : chomgwerzhañ ‘to sell’ : gwerzhgwalc'hiñ ‘to wash’ : gwalchlogota ‘to hunt mice’ : logod ‘mice’ ledanaat ‘to enlarge’ : ledan ‘broad, large’ aozadur ‘arrangement’ : aozañ ‘to arrange’ meskach ‘mixture’ : meskañ ‘to mix’ gwiskamant ‘garment, piece of clothing’ : gwiskañ ‘to dress’ kelennerezh ‘teaching, instruction’ : kelenn ‘to teach’ broderezh ‘embroidery’ : brodiñ ‘embroider’ Alamaneg ‘german language’ : alaman ‘german’ peskerezh ‘fishing, catch’ : pesk ‘fish’ peizanterezh ‘farming community’ : peizant ‘farmer’ c'hoariva ‘theater’ : c'hoari ‘play’, c'hoari ‘to play’ deadjectival -der/-ter abstract nouns -erezh collective action nouns donder ‘deepness’ : don ‘deep’ drocherezh ‘stupidities, stupid things’ : droch ‘stupid, foolish’ Productive patterns: feminine suffixes deverbal -adenn single/punctual action -adeg repeated or collectively executed action(s) -añs abstract nouns denominal -adurezh abstract nouns -eg -ell collections, other instruments yudadenn ‘roaring’ : yudal ‘to roar’ c’hoarzhadeg ‘roars of laughter’ : c’hoarzhin ‘to laugh’ nec’hañs ‘embarrassment’ : nec’hiñ ‘to embarrass’ kelennadurezh ‘instruction, education, doctrine’ : kelennadur ‘education’ segaleg ‘place planted with rye’ : segal ‘rye’ fornigell ‘cooking stove’ : forn ‘oven’ 30 cf. favereau 1997: 174: to hunt or collect the item(s) denoted by the base. 31 examples mainly from Kersulec 2010b. gender of abstract noun suffixes in brittonic languages -enn singulatives, concrete nouns -eri, -iri -erezh -iezh -oniezh -ouriezh places places functions, sciences deadjectival -ded/-ted abstract nouns -eri, -iri abstract nouns -idigezh abstr. n. in -ezh to adjectives32 in -idig -egezh -eg -elezh -ijenn -ision (V) -enn -el abstract nouns concrete nouns 73 logodenn ‘mouse’ : logod ‘mice’ gwerenn ‘(drinking) glass, window pane’ : gwer ‘glass, drinking glass, window pane’ bouloñjeri ‘bakery’ kigerezh ‘butchery’ : kiger ‘butcher’ belegiezh ‘priesthood’ : beleg ‘priest’ steredoniezh ‘astronomy’ : stered ‘stars’ prederouriezh ‘philosophy’ : prederour ‘philosopher’ izelded ‘lowness’ : izel ‘low’ koantiri ‘beauty, kindness’ : koant ‘beautiful’ pinvidigezh ‘wealth, riches’ : pinvidik ‘rich’ laouenidigezh ‘joy’ : laouen ‘joyful’ talvoudegezh ‘value’ : talvoudek ‘valuable’, talvout ‘to be worth’ santelezh ‘holiness’ : santel ‘holy’, sant m. ‘saint’ dousijenn ‘sweetness’ : dous ‘sweet’ kaledenn ‘hardness, corn, callosity, thickened part of skin’ : kalet ‘hard’ 2.1.2.2 Non-Productive or Weakly Productive Patterns: Masculine Suffixes deverbal -o (< obr -om)33 -al -at -ed deadjectival -ed (also f.) infinitives infinitives infinitives abstract nouns abstract nouns pouezo ‘to weigh’ (central breton) : pouezhuchal ‘to shout’ : huchlabourat ‘to work’ : labour-, labour m. ‘work’ sec’hed ‘thirst’ : sec’hañ ‘to dry’, sec’h ‘dry’ kleñved ‘illness’ : klañv ‘ill’ non-productive or weakly productive patterns: feminine suffixes deverbal -(i)ant abstract nouns badeziant ‘baptism’ : badez ‘baptism’, badeziñ ‘to baptize’ 32 as in most cases a suffixed adjective is attested, the suffix of the abstract noun is -ezh. rare pairs like laouenidigezh ‘joy’ : laouen ‘joyful’, however, justify the assumption of extended suffixes. 33 cf. favereau 1997: 176ff. for further rare infinitive suffixes. 74 britta irslinger denominal -enti, -inti abstract nouns merzherenti ‘martyrdom’ : merzher ‘martyr’ -od, -id, plantations, places Avalod ‘apple orchard’ (place-name) : aval ‘apple’ -ed covered with plants spernid ‘place covered with thorns’ : spern ‘thorns’ Faouet ‘place covered with beeches’ (place-name) : fao ‘beech’ deadjectival -entez abstract nouns -enti, -inti -ez abstract nouns -ez(h) -nez(h) -iz abstract nouns -ni -oni abstract nouns abstract nouns (negative qualities) kozhentez ‘ancientness’ : kozh ‘old’ mezventi ‘alcoholism’ : mezv ‘drunk’ levenez ‘joy’ : laouen ‘joyful’ hir(a)ezh ‘impatience, nostalgia’ : hir ‘long’ furnezh ‘wisdom’ : fur ‘wise’ yaouankiz ‘youth’, Pl. ‘teenagers’ : yaouank ‘young’ kozhni ‘old age’ : kozh ‘old’ c’hwervoni ‘bitterness’ : c’hwerv ‘bitter’ 2.1.2.3 Opaque Patterns The following patterns are listed because they have cognates in Welsh and thus go back to Pbr. although from a synchronic perspective many lexemes can still be analyzed as derivatives, their function is opaque. -ac’h m. abstract nouns -id m. (also f.?) -red ‘course’34 concrete nouns 2.1.3 ankounac’h m. ‘oblivion’ (only ex.) : ankouaat ‘to forget’, koun ‘memory’ broutac’h ‘scorching heat, heat wave’ : broud ‘embers’ kellid ‘germ’ : kell ‘cell’ bered f. ‘cemetery’ < Mbr bezret *‘grave-course’ gwazhred m./coll. ‘brook, water-course’ gwered (?) ‘action’ < obr gueidret *‘course of action’35 Cornish as cornish is linguistically close to breton and also an extinct language, which thus remained frozen at an earlier stage, the analysis of its derivation patterns could prove essential for the understanding and dating of certain developments. 34 cf. fleuriot 1964a: 370. The noun red ‘course’ is used as second member in composition, which has not fully developed into a suffix yet. 35 gender is not attested in old and Mbr. in Modbr, the word seems to have become obsolete, and still it is classified as feminine in most dictionaries. The Welsh cognate gweithred changed from masculine to feminine. gender of abstract noun suffixes in brittonic languages 75 unfortunately, cornish word formation is extremely underresearched. even the most comprehensive works of lewis (1990) and Williams (2011) fail to treat the topic. Wmffre (1998: 28) includes a short section on derivational suffixes in late cornish, but he does not discuss them with regard to gender. The most extensive and coherent treatment to date is Pedersen (1909–1913, i: 15ff). a first list of derivational patterns compiled on the basis of brown’s Grammar of Modern Cornish (1993) and george’s Gerlyver Kernewk Kemmyn: an Gerlyver Meur (1993)36 revealed quite a few problems with regard to the gender of suffixes. according to brown (1993: 25ff). formations in -eth, -neth and -oleth are feminine in revived cornish. The survey of actual lexemes attested in traditional cornish37 contained in george’s dictionary and Bewnans Ke (bK) presents a different picture:38 masculine -eth brastereth ‘greatness’ bredereth ‘brotherhood’ eskobeth ‘bishopric’ hireth ‘longing’ huneth ‘ancestry, lineage’ (bK) kammhynseth ‘injustice, wrong’ kemmynneth ‘commendation’ konteth ‘county’ natureth ‘natural affection’ obereth ‘mayor work, deed’ sevureth ‘seriousness’ feminine -(y)eth genesigeth ‘time of birth’ linyeth ‘lineage’ -neth boghosogneth ‘poverty’ folneth ‘foolishness’ (bK) furneth ‘wisdom’ (bK) gokkineth ‘folly’ gowegneth ‘falsehood’39 hirneth ‘long time’ pithneth ‘greed’ rywvaneth ‘kingdom’ sherewneth ‘roguery’ sleyneth ‘skill’ sotelneth ‘sleight’ sygerneth ‘idleness’ yowynkneth ‘youth’ -neth muskogneth ‘stupidity’ -oleth drokkoleth ‘ill-deed’ flogholeth ‘childhood’ gorholeth ‘demand’ kosoleth ‘tranquility’ sansoleth ‘saintliness’ skentoleth ‘wisdom’ -ogeth marghogeth vn. ‘to ride’ -ni krefni ‘greed’ -oleth sansolath ‘saintliness’ (bK) 36 cornish words are cited according to george 1993. Textual evidence is given according to the respective editions. 37 cf. also george 1991: 211. 38 The same is true for revived cornish furneth ‘wisdom’ and glasneth ‘verdure’ which are feminine according to brown 1993: 26, but masculine according to nance 1955, george 1993. 39 The word is not listed in nance 1955 and george 1993, although given in george 1991: 211. it is masculine according to Williams 1865: 182, but only attested in rd 906 without agreement. 76 britta irslinger for the following nouns, which are only attested in bK, Thomas/Williams (2007) give no gender classification: eredyeth ‘getting ready, preparation’ eryeth ‘defiance’ ehylnith ‘higher power’ = uhelneth nance (1955)40 applies largely the same classifications which appear in Williams’ Lexicon Cornu-Britannicum (1865). however, the Lexicon contains only half of the derivatives listed in modern dictionaries. a survey of the quotations revealed that in most cases the gender of a word could not be retrieved from the text, as there were no agreeing words.41 The words classified as feminine by the dictionaries are attested in the following passages: (1) bM 4387, ed. stokes 1872: Eff o purguir den worthy ay genesygeth defry bM 4387, ed. syed 1996: Ev o pur wir den wordhi a ’y enesigeth devri he man worthy from his mut.birth be.3.sg.ipf. very mut.true really ‘he was right truely a worthy man from his (time of) birth really’ The lenition of genesigeth ‘(time of) birth’ to enesigeth is caused by the masculine possessive pronoun y ‘his’, which indicates the gender of the possessor, but not the gender of the possessed item. This lenition is, however, not represented in the spelling of the original text. syed (1996) (bM) introduced it in his transcription into Kernewek Kemmyn (revived cornish). Muscogneth is classified as masculine by Williams (1865), while george (1993) assumes feminine gender and nance (1955) does not contain this word at all. again, gender cannot be inferred from any of the relevant quotations (ex. 2–4). The same is true for the word gokkineth ‘folly’ which is attested five times and thus can be considered a well documented cornish word (ex. 5–8). given the ambivalent reading of muscogneth and gokkineth it remains unclear why feminine gender is inferred for the former but not for the latter. 40 curiously, nance 1955 (s.vv.) classifies sansoleth ‘saintliness’ as feminine, its compound ansansoleth ‘unsaintliness’ however as masculine. 41 in fact, this is the case for the majority of abstract nouns, not just those in -eth. gender of abstract noun suffixes in brittonic languages (2) Ty you yv muskegys hag yn muscokneth gyllys are mad and in madness gone 77 ‘Thou art mad, and in madness gone’ (rd 1128) (3) Benen woman na govs muscoghneth not talk.imp. folly ‘Woman, don’t talk folly’ (Pc 1283) (4) Anotho dygheth vye, of-him pity be.3.sg.ppf. y wokyneth na age, his mut.folly not leave.vn. and his madness ha’y muscoghneth ‘of him it were a pity his folly not to leave, and his madness’ (Pc 1990) (5) Ty a heuel you pt. seem muskegys hag yn gokyneth gyllys crazed and in folly gone ‘Thou seemest crazed and lost in folly’ (oM 1511–12) (6) mur much a wokyneth yv of mut.folly be.3.sg.prs. go.vn den orto bewe man on-it live.vn mones the lesky peyth a-yl to burn.vn thing pt.rel.mut.can ‘much of folly it is to go to burn a thing on which a man can live’ (oM 473–5) (7) cussyllyough menough ihesu a gase y advise.imp. wokyneth ha treyle the skentuleth frequently Jesus pt. leave his mut.folly and turn to wisdom ‘advise Jesus frequently, that he leave his folly and turn to wisdom’ (Pc 1807–09) (8) Me a grys i bos nycyte ha gokyneth the ’n empowr mos the syrry pt. believe be.vn. ignorance and folly to the emperor go.vn to anger.vn ‘i believe that it was ignorance and foolishness on the part of the emperor to go and anger him.’ (bK 2077) Linyeth ‘lineage’ is classified as masculine by Williams (1865), but as feminine by george (1993). again, there is no positive evidence for the latter. The lenition of the adjective bras ~ vras ‘big’ in (10) is caused by pur ‘clean, pure, very’,42 cf. also (11). (10) Sevys rise.1.sg.aor. a lydnyathe pur vras of lineage very mut.great ‘i am sprung from a mighty lineage’ (cW 2097) (11) Pur very wyryoneth re geusys ahanaf mut.truth pt. speak.2.sg.aor. of-me ‘Very truth thou hast spoken of me’ (Pc 1587) 42 cf. lewis 1990: 8, no. 9. 78 britta irslinger it seems that Williams (1865) classifies nouns as masculine whenever there is no clear evidence of their being feminine. such a method appears to be rather unscientific and not well motivated. it would certainly be more appropriate to classify the gender of nouns which occur without agreeing words as “unknown”. Thomas/Williams (2007: 464) suggest feminine gender for sansolath ‘health, saintliness, sanctity’, although the two attestations in similar phrases occuring in bK 1538 and 1938 are inconclusive. (12) Lowena ha joy sansolath th'agan arluth pub seson, ha and holiness skeantolath ha wisdom to our lord fues ha every season and fortune and governans ha reson, sertan heb and governance and reason, surely without mar doubt ‘Joy and holiness to our lord at every season, and good fortune and wisdom and governance and reason, certainly indeed’ (bK 1538–1542) in the case of obereth, masculine gender is indicated twice by an anaphoric object pronoun. (13) hag a wyth and pt. guard in keladow the obereth casadow. in secrecy your deed detestable Du a ’n aswon byttygyns. god pt. pron.3.sg.m. knowing.vn however ‘and we shall keep secret your shameful deed. god will know it, nonetheless.’ (bK 2993–95) (14) Pub eur oll obereth da, gwynnvys kemmys a every hour all good work, blessed ’n gwrello so much pt.rel. pron.3.sg.m. do.3.sg.prs.sj. ‘blessed are all those who do good work at all times’ (oM 604–5)43 or rather: ‘every hour all good work, blessed so much (who) would do it’ based on the evidence of examples (13) and (14), the question arises whether masculine gender should be assumed for all derivatives in -eth (including -neth and -oleth). if the suffix underwent a change of gender, recent derivatives would be expected to be masculine. The majority of older derivatives, however, would still be feminine with some derivatives oscillating between the two genders. 43 reading and translation follow Keith syed and ray edwards (http://corpus.kernewek .cymru247.net/omkkks.txt, consulted 15 dec. 2011). norris 1859 and Williams 1865 (s.v. ober) give pup vr ol oberet da guyn-vys kymmys an-gvrello ‘good works at every hour, happy the lot of all who do them’. Obereth is obviously interpreted as plural oberedh (besides oberow), but the following pronoun is singular. gender of abstract noun suffixes in brittonic languages 79 according to nance (1955) and george (1993), derivatives containing the suffix -ni like dellni ‘blindness’, klofni ‘lameness’ and kothni ‘old age’ are masculine in revived cornish. The only ni-formation found in traditional cornish is the hapax legomenon krefni (cW 682). however, the passage including the word seems to be corrupted. both dictionaries assume that the lenited noun grefnye is employed instead of the adjective kraf ‘greedy’. lenition can therefore not be considered a clear indicator of feminine gender, especially since the article which triggers it is missing. (15) Ny vannaf not proclaim.1.sg.prs. bos mar grefnye, tha wetha ol ow honyn be.vn. so keep.vn all myself greedy to ‘i proclaim not to be so greedy to keep all myself’ (cW 682) Words in -(V)shyon (spelled -c(y)onn in the manuscripts, -(V)ssyon in Thomas/Williams (2007) are relatively frequent. at least sixteen lemmata borrowed from old french or more probably from Middle english44 can be listed such as benedykshyon ‘benediction’, dampnashyon ‘damnation’, posseshyon ‘possession’. george (1993) (s.vv.) classifies all of them as masculine. his categorization is supported by several instances of unlenited initial consonants following the article.45 however, as there are no derivatives from cornish bases, -(V)shyon cannot be regarded as a true cornish suffix. The existence of -yb, -eb as contained in the suffixes -neb and -enep46 is attested for cornish through its occurrence in the word gorthyb ‘answer, reply’. both -neb and -enep as well as gorthyb itself are unanimously classified as masculine by all dictionaries and by Thomas/Williams (2007: 411). however, within more than a dozen attesteations of the word gorthyb, no instances of masculine agreement could be established. Quite to the contrary, the word appears to be feminine in bK 211.47 (16) ro gorthyb vas give.imp. answer mut.good ‘give a satisfactory answer’ (bK 211) 44 cf. george 1991: 204. 45 cf. an benedyccon ‘the benediction’ (bM 4532), an passyonn ‘the passion’ (bM 122), re’n Passyon ‘by the Passion’ (bK 1447), un conduconn ‘one/the same condition’ (bM 2824). 46 it corresponds to W -eb, which is originally a second member in composition, s. above 2.2.2.1.2. Mco gorthyb could be analyzed as a compound. due to the lack of other words in -yb, it is unclear if the latter has become a suffix as it did in Welsh. 47 note, that Welsh possesses feminine compounds in -eb. 80 britta irslinger in the following section, suffixes forming abstract nouns in traditional cornish are listed according to the gender classification suggested by george (1993).48 as illustrated in the discussion above, nominal gender in cornish has so far not been adequately studied, and indications in dictionaries and handbooks are not to be regarded as reliable. for the majority of nouns the consulted texts did not provide any information on their gender. furthermore, lexicographers seem to have a strong preference for masculine gender, irrespective of etymological or comparative considerations. as double-checking gender indications for each suffix is beyond the scope of the present study, in some cases the gender classification could be incorrect. Where question marks accompany an entry, the suffix was double-checked but the textual evidence was not sufficient to establish a gender classification beyond any doubt. Productivity was not considered as, within the Middle and late cornish corpus, abstract nouns are relatively rare in general. for some of the patterns below, no more than a handful of examples could be found. all examples cited are in traditional cornish (according to george 1993). Masculine suffixes deverbal -a verbal nouns -y/-i verbal nouns (-e) verbal nouns -he verbal nouns -ya verbal nouns -ia verbal nouns -adow abstract nouns sygha ‘to dry’ : sygh-, sygh ‘dry’ leski ‘to burn’ : loskoco cusce ‘to sleep’, Mco kosk glanhe ‘to clean, clear’ : glan-, glan ‘clean’ governya ‘to govern’ : governgolia ‘to wound’ : goldannvonadow ‘message’ : dannvon ‘to send’ -(y)ans/ -nans49 gevyans ‘forgiveness’ : gava ‘to forgive’ bywnans ‘life’: bywa ‘to live’ -es -vann -ynsi abstract nouns from verbal nouns in -he, -a abstract nouns action nouns abstract nouns -ys action nouns syghes ‘thirst’ : sygha ‘to dry’, sygh ‘dry’ loskvann ‘burning’ : leski ‘to burn’ ankombrynsi ‘embarrassment’ : ankombra ‘to embarrass’ kenys ‘singing’ : kana ‘to sing, sound’ 48 cf. also george 1991: 204. 49 cf. george 1991: 207ff. with a list of derivatives. gender of abstract noun suffixes in brittonic languages denominal -ek -edh -ji, -di, -ti -la -ses language terms abstract nouns? houses (compounds) places (comp.) abstract nouns -sys50 -yn(n) abstract nouns concrete nouns deadjectival -ans abstract nouns -der/-ter51 abstract nouns -edh52 abstract nouns ? -neb abstract nouns ? -enep53 abstract nouns -ethes abstract nouns (-edhes)54 -i abstract nouns ? -ni ? -uri ? -duri abstract nouns abstract nouns bases = loan words 81 Sowsnek ‘english’ : Sows ‘englishman’ tiredh ‘land, country’ : tir ‘land, ground’ koskti ‘dormitory’ : kosk ‘sleep’ dornla ‘handle’ (lhuyd) : dorn ‘hand’ ? kowsesyow pl. ‘conscience, conviction’ : kows ‘speech’ mestr-yn-ses ‘dominion’ : mestr ‘master’ arloettes ‘lordship’ : arloedh ‘lord’ dywsys, dywses ‘deity, godhead’ : dyw ‘god’ no clear examples attested (see below 3.2.2 and fn. 122) tristans ‘sadness’ : trist ‘sad’ braster ‘greatness’ : bras ‘big, great’ gwiryonedh ‘truth’ : gwiryon ‘true’ folneb ‘folly, foolishness’ (bK) : fol ‘foolish’ hevelenep ‘likeness’ : haval ‘similar’ podrethes ‘corruption’ : poder ‘corrupt’ mostethes ‘filth’ : mostys ‘filthy’, most ‘filth’ anfeusi ‘disaster’ : anfeus f. ‘ill luck’ : anfeusik ‘unfortunate’ krefni ‘avarice, greediness’ : kraf ‘greedy’ falsuri ‘falseness’ : fals ‘false’ lelduri ‘loyalty’ : lel ‘loyal’, lelder ‘loyalty’ lenduri ‘sincerity’ : len ‘faithful’ trayturi ‘treachery’ : trayta ‘betray’, traytour ‘traitor’ 50 This suffix, having another origin than -ses, appears in traditional cornish, e.g. dewsys (oM 2666, rd 2487, 2504), and in nance 1955. george 1993 lists dwyses, assuming the same development as in breton, where *-did/-tid and *-dod/-tod merged into -ded/-ted. 51 according to george 1991, 205, this suffix was productive throughout the period of traditional cornish. 52 Written -eth by nance 1955, Williams 1865. 53 cf. further medhvenep ‘drunkenness’ (i.e. methewnep Th 16a) : medhow ‘drunk’, kothenep m. ‘antiquity’ (cotheneb Th 34a) : koth ‘old’, gorthenab ‘opposition’ : gorth ‘obstinate’ (bK 996), rowndenab ‘roundness’ : *rownd ‘round’ (bK 283). The latter derivative from an english base proves the productivity of the pattern. The only denominal example hastenep ‘haste’ (hastenab Th 66) : hast ‘haste’ is semantically identical with its derivational base. alternatively, it might be derived from an unattested verb. The masculine gender assumed by george 1993 and Thomas/Williams 2007 is not attested at all. The suffix is not employed in revived cornish. 54 While traditional cornish seems to display -ethes only, the handbooks of revived cornish spell -ethes (nance 1955, Williams 2000) or -edhes (brown 1993). george 1993 lists both variants based on etymological speculations (Mco -edh ~ W -edd ~ br. -ez vs. Mco -eth ~ W -aeth ~ br -ezh) which are not entirely comprehensible. The assumption of masculine gender might be supported by lack of lenition after the article in oM 2808–9: hay vertu a-vydh lyhys dre an mostethes hep fal ‘and its virtue will be lessened by the filth, without fail’ (but see below fn. 58). 82 britta irslinger -yjyon abstract nouns -ys -yn(n) abstract nouns abstract nouns feminine suffixes deverbal -va action nouns places poesyjyon ‘oppression’ : poes ‘heavy’ glesyjyon ‘grass-plot’ : glas ‘green, blue’ henys ‘old age’ : hen ‘old’ dregyn ‘misfortune, damage’ : drog ‘bad’ diberthva ‘separation’ : diberth ‘to separate’ kuntellva ‘gathering place’ : kuntell ‘to gather’ denominal -as -ek -ell -en(n) ? -(y)eth -ieth -(on)ieth57 abstract nouns plantations, places instruments concrete nouns abstract nouns functions, sciences -va places deadjectival -a58 abstract nouns -en(n) concrete nouns ? -eth abstract nouns ? -ogeth from adj. in -ek ? -oleth from adj. in -el ? -neth abstract nouns -va places kowethas ‘association’ : koweth ‘companion’ only in place-names55 gwynsell ‘fan’ (lhuyd) : gwyns ‘wind’ boemmenn ‘blow, stroke’ : boemm ‘blow, bump’56 linyeth ‘lineage’ : lin ‘line’ medhygieth ‘medicine’ : medhyk ‘doctor’ Kristonieth ‘christianity’ : Kriston*, Kristyon ‘christian’ esedhva ‘seat, siege, sitting-room’ : esedh ‘seat’ lowena ‘joy’ : lowen ‘joyful’ oco tenewen ‘side, flank’: Mco tanow ‘thin’ genesigeth ‘time of birth’ : genesik ‘native-born’ marghogeth vn. ‘ride’ : marghek ‘horseman’59 sansoleth ‘saintliness’ : sans ‘holy’, sansel ‘saintly’ gokkineth ‘folly’ : gokki ‘foolish’ genesigva ‘birthplace’ : genesik ‘native born’ 55 These place-names contain no information about gender, e.g. Park-an-Lastrak (*elestrek ‘iris-bed’), Gersick-an-Awn (*corsek ‘reed-bed’): cf. Padel 1985: 66, 90, 92. 56 cf. also hanasenn ‘sigh’ (lhuyd), a derivative in -enn from unattested *hanas ‘sighing’. 57 examples for -onieth exist only in revived cornish. 58 -a from -edh. The suffix -edh is preserved in tregeredh ‘compassion, pity’ : *trugar ‘compassionate’, for which feminine gender is inferred by nance 1955, george 1993 and Williams 2000. however, the lack of lenition after the article as it is evident in bM 1838: ha’n tregeredh warnedha ‘and the mercy upon them’ indicates masculine gender. it is of course possible that lenition is missing because of orthographic inconsistency (cf. also above 2.1.3 ex. (4) where lenition of the noun after the possessive pronoun sg. masc. is present in y wokyneth and missing in y muscoghneth). This might also be the case for other attestations adduced in support of masculine gender interpretations, which renders the textual evidence for gender even more scanty and the classifications suggested in the dictionaries all the more arbitrary. 59 Marghogeth : marghek shows the regular derivational pattern, but it is irregular with regard to the gender of the derivative (masculine as a verbal noun) and the word class of the base (substantivized adjective). regular examples exist in revived cornish. gender of abstract noun suffixes in brittonic languages 83 2.2 Developments The derivational systems of Modern Welsh and Modern breton and their respective gender systems result from independent developments of cognate languages over more than a thousand years. The sections below will address the major changes which have lead to the emergence of the patterns discussed above. The following facts are crucial: – neuter gender of Proto-celtic was already lost in old brittonic (schrijver 2011: 41). The earliest attestations of the latter show no traces of this gender. – The brittonic languages have continuously been exposed to language contact. They were influenced by latin, when britain was a part of the roman empire (43 until ca. 410 ad). later, Welsh and cornish experienced contact with english, and breton was influenced by romance languages, in particular by french. in addition, there was contact with goidelic from prehistorical times onwards.60 The gender of nouns is attested only sporadically in the earlier periods. statements about the gender of suffixes are thus mainly based on modern languages. 2.2.1 Proto-Brittonic Developments as all three languages show the same gender assignment rules, it is likely that the above developments took place before the separation of Welsh, cornish and breton. 2.2.1.1 Substantivizations substantivizations, i.e. adjectives, infinitives, pronouns etc. used as nouns, are always masculine. since they do not constitute prototypical nouns, they are integrated into the mental lexicon only to a limited degree. The degree of integration into the mental lexicon is important for gender assignment. Prototypical nouns can have any gender, whereas there is only one gender possible for non-prototypical nouns.61 since in all ie 60 Matasović 2007 discusses the assumption of insular celtic as a linguistic area. 61 cf. di Meola 2007: 143. in Modhg, a low degree of integration correlates with neuter gender, which is found in recent loan words as well as in occasionally substantivized words such as das Hin und Her ‘the see-saw’, das Warum ‘the why’, das A ‘the a’. 84 britta irslinger languages with a three-gender system substantivizations are neuter,62 the same may be assumed for Proto-brittonic. after the neuter gender was lost, masculine gender took over this function in brittonic. W, br, Mco syntactically transposed words, substantivizations n. > m. 2.2.1.2 Verbal Nouns Verbal nouns are always masculine, while the same word, when used as a common noun, may be feminine (schumacher 2000: 15). Pedersen (1913: 67) assumes that abstract nouns were rarely used in syntactic constellations which displayed initial lenition. This assumption provides one possible reason for the change of feminine to masculine gender. With regard to verbal nouns, one has to take into consideration the degree of integration into the mental lexicon discussed in the previous paragraph. Verbal nouns are a hybrid category. although they possess nominal features, they are also more grammaticalized than common nouns, and in breton they eventually became infinitives.63 in Welsh, breton and cornish those suffixes which form verbal nouns or infinitives and which are productive in the modern languages can be reconstructed as feminine on the basis of their old irish cognates. They contain the insular celtic suffix *-mā preceded by different vowels.64 W -i, br -iñ, Mco -y, -i65 W -aw, oco -e66 br -aff, -añ W -u67 br -o, obr. -om verbal nouns verbal nouns verbal nouns verbal nouns verbal nouns < *-ī-mā < *-ā-mā < *-ă-mā < *-ū-mā < *-o-mā f. > m. f. > m. f. > m. f. > m. f. > m. 2.2.1.3 The Suffix Vowel -e- as a Marker of Feminine Gender The Pcl feminine thematic vowel *-ā caused a-affection, which changed early Pbr *-i- in a penultimate syllable to -e-. The suffix vowel -e- sets itself 62 compare e.g. substantivized adjectives not denoting persons such as latin bonum ‘the good’, oir a n-olc ‘the evil’ (Ml. 33 d 2), Modhg das Gute, as well as letter names old greek o mikrón ‘small o’, latin a longum ‘long a’ (Matasović 2004: 160). 63 however, the majority of oir verbal noun suffixes, especially some of the productive ones, are masculine. 64 The reconstruction of *-mā is assured by the feminine gender of nouns formed with the oir suffix -em, -am, e.g. cretem f. ‘belief’ (schumacher 2000: 130). 65 cf. schumacher 2000: 130f., Zimmer 2000: 390ff., goi 453. 66 cf. schumacher 2000: 133. 67 cf. schumacher 2000: 136. gender of abstract noun suffixes in brittonic languages 85 apart from masculine suffixes containing the original vowel early Pbr *-iwhich was not affected by the Pcl masculine stem vowel *-o- (see below 3.2.2). Thus, the vowel -e- in final syllables developed to some degree into a marker of feminine gender. although the application of this “rule” caused some nouns to change from masculine to feminine gender, there still are masculine words containing original *-e-. Therefore, it is far from being a reliable indication of gender, especially in breton. in Welsh other gender assignment rules tend to override this phonetic rule, but with regard to formations in -ell feminine gender is stable. The suffix -ell is mainly used for the derivation of nouns denoting instruments. it is based on latin -ellus m., -ella f., -ellum n., i.e. on suffixes belonging to different genders. This suffix is predominatly feminine in all brittonic languages. instances of double gender and the change of Welsh formations with the suffix -red (from *reto- m. ‘course’: cf. Zimmer 2000: 446) into feminine gender may result from the developments discussed. W, br, Mco -ell W, (br) -red m. (some f.) instruments < lat. -ellus m., -ella f., -ellum n. action nouns < *reto- m. ‘course’ abstract nouns m./f./n. > f. m. > f. m./f. 2.2.1.4 Productivity of Proto-Celtic *-aktā f. The feminine suffix Pcl *-aktā > icl *-axtā68 continues to be productive for the formation of denominal and deadjectival abstract nouns in all three languages. in addition, W -(i)aeth, br -(i)ezh and Mco -(y)eth developed several variants by extension with suffixes belonging originally to their bases, e.g. the adjective suffix -ol- → W -ol-aeth, Mco -ol-eth, br -el-ezh.69 The explanation of *-aktā as a suffix conglomerate presupposes a similar development. in Pcl, *-tā- (< Pie *-tah2-) formed abstract nouns from adjectives in *-ako/ā. The resulting cluster *-aktā was later reanalyzed as a suffix.70 it is however difficult to prove the individual steps of this development. Whereas suffixes of the structure *-V-ko/ā- are frequent in celtic with regard to the formation of adjectives, the majority of suffixes productive in insular celtic can be traced back to *-(i̯)āko/ā- and *-ī�k̆ o/ā-. There is no evidence for *-ăko/ā-.71 68 69 70 71 cf. oir -acht, -echt (goi 167f., 452), Modir -Vcht (doyle 1992: 31ff.). cf. Pedersen 1909–1913, ii: 32 with further examples of this process, also from irish. cf. Pedersen 1909–1913, ii: 32, de bernardo stempel 1999: 334. cf. russel 1990: 29, de bernardo stempel 1999: 327. 86 britta irslinger alternatively, the suffix could be explained as originating from the noun *axtā ‘act, deed’,72 which constituted the original second member of compound nouns. based on its semantics, the resulting compounds could be assumed to be action nouns like W marchogaeth vn. ‘action of riding’ : marchawc ‘rider, horseman, knight’. however, the brittonic derivatives normally have more abstract semantics such as ‘condition, state, office, property’. Their english translations often mirror these semantics through suffixes like -dom and -ship, e.g. W tywysogaeth ‘principality, kingdom’ : tywysog ‘prince, lord’. The feminine noun marchogaeth is glossed ‘horsemanship’, and according to schumacher (2000: 208) this is its original meaning. although it is impossible to establish the original derivational pattern of formations in *-axtā, they mostly seem to be denominal and deadjectival abstract nouns while they are only rarely used as verbal nouns.73 germanic suffixes like e -ship or -dom may serve as a parallel as they fulfil functions similar to *-axtā and like *-axtā constitute the second element in the formation of compound nouns. -ship or -dom originate from nouns meaning ‘condition, state, dignity, office’ vel sim,74 not from nouns referring to ‘deed’. The alleged noun *axtā ‘deed’ is therefore unlikely to constitute the origin of the brittonic suffix. in spite of the problems discussed above, the analysis of *-aktā as a suffix conglomerate seems to be preferable. irrespective of its etymology, *-aktā contains the Pie suffix *-tah2-, which preserved its feminine gender in all brittonic languages. 2.2.1.5 Brittonic Suffixes 2.2.1.5.1 native suffixes Two brittonic suffixes, -ni and -oni, form abstract nouns from adjectives. The productivity of -ni in Welsh is feeble, but it has not yet ceased. in cornish and breton, the rare occurrence of derivatives in -ni points to an early loss of productivity. several Welsh abstract nouns in -ni correspond 72 The noun *ax-tā ‘act, deed’ contains the suffix *-tā. it is derived from the Pcl verbal root *ag- ‘to drive’ (Pie *h2eǵ-): cf. Morris Jones 1913: 230, 389, lewis/Pedersen 1961: 80, fleuriot 1964a: 338, schumacher 2000: 208, Kersulec 2010a: 153. 73 cf. schumacher 2000: 208 for MW and (hemon 1975: 2000) for Mbr. 74 for the etymology of -dom see fn. 77 below. Mode. -ship, oe -shipe ‘condition, state, dignity, office’ forms denominal and deadjectival abstract nouns and collectives like oe frēond-scipe ‘friendship’, dryht-scipe ‘accompaniment, allegiance’. it developed from a noun used as the second member in compounds. This noun is also found in oe ge-sceap ‘consistence, form’ (Krahe/Meid 1969: 221, faiss 1992: 61, dalton-Puffer 1996: 85ff.). gender of abstract noun suffixes in brittonic languages 87 to breton formations in -oni, e.g. W noethni ~ b noazhoni ‘nakedness’. The etymology of these suffixes is crucial to establish their gender. at the same time, it raises difficulties. The following list encompasses the earliest attestations of the two suffixes as well as parallel formations in more than one language: obr glethni gl. gastrimargia, uentries ingluues (9th)75 MW glythni (14th) : glwth ‘gluttonous, voracious’ (12th–13th), from lat. gluttō ‘glutton’ ? Mco glotni (Pc 52, Th 16a)76 MW brychni ‘spots, freckles, mildew’ (12th–13th) : brych ‘spotted, freckled’ MW glesni ‘blueness, paleness, greenness’ (13th) : glas ‘green, blue, pale, fresh’ MW goleuni ‘light, illumination, brightness’ (14th) : golau ‘clear, bright’ Mbr kozhni ‘old age’ (15th) : kozh ‘old’, kozhoni ‘id.’ Mco krefni ‘greediness’ (cW 682) : kraf ‘greedy’ W poethni ‘great heat, inflammation’ (16th) : poeth ‘hot, burning’ br poazhni ‘burn, burning’ (18th) : poazh ‘cooked’ The suffix -ni is explained as originally being a second member in compounds by Morris Jones (1913: 232), followed by fleuriot (1964a: 355) and Zimmer (2000: 408ff). it is reconstructed as the noun *gnīmu- m. ‘deed’: cf. oir gním, Modir gníomh m. ‘act, deed’. although this etymology would explain form and gender of the suffix, it is not convincing with regard to its function, as it presupposes that meanings like ‘old age’ or ‘greenness’ have developed from compounds like *‘old-deed’ or *‘green-making’.77 This would only be plausible if ni-derivatives were deverbal. a further argument against this etymology is the fact that in obr (and probably also oW) the suffix is always -ni, while -gnim ‘deed’ is attested in obr 75 corrected from glathni (fleuriot 1964b: 176). 76 shows -t- instead of -th- and may thus have been borrowed from or reshaped after Me glotonie (george 1993: 106f.) or it stems directly from ofr glotonie: cf. also gloteny (Pa 13.1). 77 The assumed semantic development is quite improbable. a comparison could be made with the germanic suffixes *-haiđu- and *-dōma-. both developed from second members in compounds into suffixes forming deadjectival abstract nouns. The semantics of the underlying nouns have nothing to do with ‘act, deed’ vel sim. The suffix *-haiđudeveloped from the noun continued by gothic haidus ‘kind, manner’, ohg heit, oe hād, hǣd ‘(personal) kind, manner, class, rank, dignity’. its scope equals that of *-dōma-, a suffix forming abstract nouns from adjectives as well as nouns denoting personal qualities or social rank. amongst others, the noun is attested in gothic dōms ‘kind, destination’, oe, ohg tuom ‘judgement; fame, power, dignity’. examples of semantic developments are: oe frēo-dom, ohg frī-tuam ‘freedom’ < *‘the personal state of being free’, oe wīs-dom, ohg wīs-tuom ‘wisdom’ < *‘wise judgement’, oe rīce-dom ‘kingly power’ < *‘kingly dignity’ (Krahe/Meid 1969: 219ff., dalton-Puffer 1996: 75ff., 77ff.). 88 britta irslinger im-guo-gnim ‘undertaking, construction’, oW guor-gnim ‘great toil, exertion’, MW gnif ‘labour, pain’.78 one should thus reconsider the analysis by Pedersen (1909–1913, ii: 18) who explained -ni and -oni as extensions of -n- or -on- by the masculine abstract suffix -i. The etymology of the latter is not clear: cf. the discussion by Jackson (1953: 351ff). since different reconstructions of -i are possible, the original gender of the suffix remains open to debate. Jackson (1953: 353), followed by fleuriot (1964a: 336), considers the abstract and collective suffix -i to originate from *-ī̆í o̯ - < Pie *-i̯o-. The same origin is assumed for -ydd, which then developed differently due to a shift in syllabic division. according to Zimmer (2000: 444), -oni is a conglomerate combining the plural suffix -(i)on, which originally belongs to the category of old ōn-stems, with the masculine abstract noun suffix -i. like Morris Jones (1913: 231), he considers the latter suffix to be identical with the suffix of verbal nouns and infinitives, i.e. -i from Pbr *-īmā. however, the suffixes have been distinct from their earliest attestations onwards and continue to be so in Modern breton, whereas in Middle Welsh they have become homonymous. Verbal nouns display the suffix -im, e.g. oW erchim ‘to seek, to ask for’, obr guomonim ‘to promise’ (fleuriot 1964a: 311). This ending developed via Mbr -iff into Modbr -iñ, -i, e.g. beviñ ‘to live’.79 in contrast, the abstract noun suffix -i did not undergo any change: cf. obr diochi, Modbr diegi ‘laziness’ (fleuriot 1964b: 336).80 Therefore, it does not seem appropriate to reconstruct the abstract noun suffix -i as Pbr *-īmā. Pbr *-ni Pbr *-(i)oni deadjectival abstract nouns -n+ -i m. < *-ī̆ó ̯ -? m. deadjectival, denominal abstract nouns -on- + -i m. < *-ī̆í o̯ -? m. being attested in more than one language, the following suffixes are also likely to originate in Pbr or, in the case of -ereth, in sWbr. as each of the suffixes became very productive in one language, while in the others they are just preserved in a few words, the suffixes in question will be discussed in the respective sections below. 78 cf. fleuriot 1964b: 219, 195, falileyev 2000: 73. 79 cf. favereau 1997: 171 for this ending in Modbr dialects. 80 There are no examples attested from oW. gender of abstract noun suffixes in brittonic languages 89 Pbr *-d-eb deadjectival abstract nouns see below 2.2.2.1.2 f.? Pbr *-in-eb deadjectival abstract nouns see below 2.2.2.1.2 f.? sWbr *-or-eth deadjectival, denominal abstract nouns see below 2.2.2.1.3.5 f.? 2.2.1.5.2 loan-suffixes from latin latin loanwords, containing feminine -tātem, developed into W -dod/-tod, br -ded/-ted, Mco -ses. The loan suffix, which was and is productive in all three languages, was probably abstracted from the latin ending in Pbr. Zimmer (2000: 321f.) speculates that lat. -tātem might have merged with the native cognate suffix *-tātan. While in breton words in -ded are exclusively feminine,81 there is more variation in Welsh. loanwords in -dod may be feminine like cardod ‘alms, charity’ (< lat. caritātem), masculine like ufylltod ‘humility, modestity’ (< lat. humilitātem) or oscillate between masculine and feminine like ceudod ‘thought, mind, heart, bosom’ (< lat. cavitātem). in revived cornish, the suffix -ses is masculine,82 although george (1993: 324) (s.v. uvelses) infers feminine gender for the loanword huueldot gl. humilitas (Vc 433). Pedersen (1909–1913, ii: 42) assumes that this inconsistency of gender might be due to the influence of the celtic masculine suffix *-tūto-/-tūti-,83 which forms abstract nouns based on adjectives and developed into W -did/-tid, Mco -sys.84 considering the gender development of these suffixes for the individual languages, it could be assumed that the category of abstract nouns as a whole, rather than another individual suffix, has triggered these changes. 81 although deshayes 2003: 377 (s.v. kawded) classifies Modbr kaouded ‘heart, mind’ < lat. cavitātem as masculine, conclusive agreement has not been attested so far: cf. gib ii 1378. 82 Pedersen 1913: 42 points out that this suffix might also continue lat. -tās. 83 also Zimmer 2000: 319. This suffix has cognates in oir, lat. and gothic, which differ with regard to stem-formation and gender: oir *-tūt- m. is common with deadjectival abstract nouns, e.g. oítiu (goi 165f., de bernardo stempel 1999: 396ff.). lat. *-tūt- f. is attested in four abstract nouns derived from personal nouns like iuventūs ‘youth’ (leumann 1977: 375). gothic *-tūti- f. is found in four deadjectival abstract nouns denoting qualities, e.g. mikildūþs ‘greatness’. lat. iuventūs and oir oítiu are regarded as inherited cognates, but the suffix underwent a change of gender in one of the languages. The correlation between feminine gender and abstractness is found both in latin and gothic. 84 obr -tit (> Modbr -ded/-ted) is ambiguous as it may originate from *-tāt- or *-tĭt-, but not from *-tīt- (< *-tūt-) which is the basis of the Welsh suffix (fleuriot 1964a: 349). 90 britta irslinger loanwords m./f. f. not attested W -dod/-tod br -ded/-ted Mco -ses native formations m. (m./f.) f. traditional cornish: not attested revived cornish: m. Terms denoting languages have feminine gender in Welsh, but masculine gender in breton and cornish. according to Zimmer (2000: 497), at least ModW Saesneg ‘english’ is a direct loan from (british-)lat. Saxonica lingua, i.e. a phrase consisting of an attributive adjective and a feminine head noun which was subsequently lost. The phrase could also have evolved from brittonic. in this case, its component parts would be an adjective in celtic *-ĭkā and the feminine noun *i̯eχtV- ‘language’,85 continued by MW ieith, Mco yeth, br yezh. in cornish and breton, language terms may have changed gender from feminine to masculine because they were reanalyzed as substantivized adjectives without feminine head nouns. This process might have been supported by the fact that the gender of language terms is masculine in french and genderless in english. br -eg Mco -ek language terms - substantivized adjectives - french/english influence? f. > m. 2.2.2 Changes in the Individual Languages 2.2.2.1 Welsh 2.2.2.1.1 change of gender recent derivatives of the suffix -dod/-tod are exclusively masculine. The same is true for deadjectival abstract nouns formed with the suffix -edd, like mawredd ‘bigness’, noethedd ‘nakedness’, which underwent a change of gender (s. below 3.2.2). -dod/-tod -edd abstract nouns deadjectival abstract nouns < lat. -tātem f. < Pbr *-ii ̯ā- f. f. > m. f. > m. 2.2.2.1.2 Welsh suffixes Most of the other Welsh abstract noun suffixes are also masculine, with the exception of feminine -eb. according to Zimmer 2000: 338ff., -eb goes 85 cf. schrijver 1995: 106f., irslinger 2002: 205f. gender of abstract noun suffixes in brittonic languages 91 back to Pie *seku̯- ‘to speak’.86 This root is preserved in fossilized literary verb forms (eb, ebe, ebr ‘quoth he’) and in the suffix -eb, an original second member in composition which survives solely in verbal or nominal compounds. The gender of the latter is not uniform: cp. ateb m. ‘answer’ (12th– 13th), ymateb m. ‘reaction, response’, gwrthateb m./f. ‘reply, answer’ (16th), dihareb f. ‘proverb, maxim, parable’ (13th), gwrtheb m. ‘answer, rejoinder, objection’ (12th–13th). Ateb, ymateb and gwrtheb may also be used as verbal nouns, in which case they are masculine. it is impossible to ascertain the original gender of -eb as used in compounds. here the ending may even originate from different stems (with different genders). The productive abstract suffix -eb is feminine. Two important semantic groups of terms and neologisms mainly from the 19th and 20th centuries denote utterances, e.g. croeseb ‘paradox, contradiction’, gwireb ‘aphorism, truism’, and written matters, e.g. anfoneb ‘invoice’, derbynneb ‘receipt, acquittance’. in contrast, the productive extensions -d-eb, -ol-d-eb and -in-eb are masculine.87 Pedersen (1913: 34) assumes that these abstract nouns rarely occurred with agreeing words, causing their gender to became opaque. evidence for these suffixes in breton and cornish is rare.88 in obr, -dep is attested only once in paroldep ‘duration, consistency, material, matter’.89 There are no formations with this suffix in the modern language. Traditional cornish has the suffixes -neb and -enep (~ W -ineb) for which no gender is attested (see above 2.1.3 and fn. 53). The suffix -dra is supposed to be an original second member in composition. it is identical to the noun tra ‘thing’, which is documented for all brittonic languages. it is frequent in Modbr, but it has gone out of use in Welsh. The noun tra is assumed to be originally neuter because of its irregular agreement behaviour. although classified as masculine in dictionaries and agreeing with masculine numerals and pronouns, it undergoes and causes mutations like feminine nouns.90 in Welsh attestations from the 12th century, tra is found to agree with feminine pronouns 86 cf. leia s-62f. for irish words belonging to this root. 87 cf. Zimmer 2000: 325ff. for the etymologies of the extended suffixes. 88 The adjective br heveleb, W cyffelyb ‘similar, the same’ (*ko-hemel-ep) is probably based on another root in -ku̯, namely Pie *h3oku̯- ‘to catch sight of, to behold’: cf. fleuriot 1964a: 344, Pedersen 1909–1913, ii: 33f., Morris Jones 1913: 230. 89 cf. fleuriot 1964a: 344, 1964b, 281. 90 cf. Pedersen 1909–1913, ii: 14, Morris Jones 1913: 232, fleuriot 1964a: 317, favereau 1997: 72, gPc (s.v.), nance 1955: 95. 92 britta irslinger (lewis/Pedersen 1961: 159). it thus seems very likely that the suffix -dra has undergone a change of gender. it is unclear when the noun developed into the suffix, turning from either neuter or feminine into masculine. The suffix -wch is reconstructed as *-isku̯- and its derivatives are invariably masculine.91 -eb action nouns < *-seku̯- m./f./n.? ‘speaking’ ? utterances, demands f. verbal nouns m. deadjectival abstract nouns < *-ti-/*-to-/*-tu- + -eb- -d-eb -ol-d-eb -in-eb < *-īno-/-īni- + -ebf. ? > m. -dra deadjectival abstract nouns < tra f. ‘thing’, orig. neuter n. > (f. ? >) m. -wch deadjectival abstract nouns < *-isku̯m. 2.2.2.1.3 loan-suffixes from latin The suffix -ol (< lat. -ālis, Zimmer 2000: 549) forming abstract nouns is identical to the suffix -ol forming adjectives. nouns in -ol were thus originally substantivized adjectives and are accordingly masculine. The suffix -dwr is a rare variant of -der and -dra. Zimmer (2000: 442) tentatively suggests lat. -tūra as its origin. if this were correct, the feminine gender of the latin suffix would have changed into masculine in Welsh. -ol ? -dwr denominal and deadjectival abstract nouns deadjectival abstract nouns adjective suffix -ol < lat. -ālis substantivized adjectives < lat. -tūra f.? m. f. > m. although in some cases the original gender of the suffix is not entirely clear, Welsh displays a strong drift towards masculine gender. 2.2.2.2 Cornish 2.2.2.2.1 change of gender The same tendencies are at work in cornish. despite the fact that gender is not attested for the majority of abstract nouns in -eth and -neth, some words seem to have changed into masculine. The productivity of -neth is due to its similarity to the oe, Me suffix -ness (george 1991: 211). 91 cf. Zimmer 2000: 427 with reference to Pedersen 1909–1913, i: 373 and Morris Jones 1913: 142, 232, 383. gender of abstract noun suffixes in brittonic languages -eth -neth abstract nouns deadjectival abstract nouns < *-axtā f. -n- + -eth f. cf. e -ness92 93 some f. > m.? some f. > m.? 2.2.2.2.2 cornish suffixes in cornish, new suffixes are exclusively masculine. The masculine nouns le ‘place’ and ti, chi ‘house’ are used as second members of compounds denoting places and houses. although the feminine suffix -va has the potential to fulfil the same function, its influence on the gender of the masculine compounds is not attested. Masculine gender can be expected to hold for derivatives with the plural suffix -adow, as gender differentiation in the plural was largely abandoned. The suffix -ethes is an ambiguous cluster consisting of the feminine suffixes -eth and -es. george (1993: 335) equated the latter with W, br -ed, e.g. in syched, sec'hed ‘thirst’. While this suffix is masculine in Mco, it may occur with feminine nouns in W and br. it is not evident why the abstract noun suffix -eth should have been extended by *-ed, which is semantically less clear-cut. alternatively, one should consider the W feminine collective suffix -es, e.g. in branes ‘flight of ravens’ : brân ‘raven’.93 The suffix conceptualizes instances of a (negative) quality as an entity, which is consistent with the fact that all attested derivatives belong to the same semantic field. Thus, the meaning ‘filth, dirt, corruption’ is conveyed by mostethes, podrethes, plosethes, and by the word lastethes created by nance (1955). -ji, -di, -ti -la -adow -ethes houses, buildings places deverbal abstract nouns deadjectival abstract nouns compounds with ti, chi m. ‘house’ compounds with le m. ‘place’ original plural < -eth- f. + -es f.? m. m. m. f. > m. 2.2.2.2.3 borrowing and change of gender several loan suffixes have masculine gender. it is not entirely clear when they first entered cornish and from which language they have been borrowed. since -(y)ans and -yjyon have parallels in breton, they could be rather old. however, they were strongly influenced by loan words with homonymous suffixes from Me. 92 cf. dalton-Puffer 1996: 81f. 93 The suffix is attested in obr bues ‘enclosure for cows’, naues ‘flow’: cf. fleuriot 1964a: 339f. 94 britta irslinger Mco -(y)ans is assumed to be a fusion of the native masculine suffix oco -ant > Mco -ans with -ans in Me loan words. The english suffix was borrowed from ofr -ance, -ence f. < lat. -antia, -entia f.94 While loan words from Me were genderless and thus needed to be assigned a gender, loan words adopted from latin or ofr underwent a change of gender. This is also likely in the case of -yjyon-.95 The Mco suffix -uri has no equivalents in ofr or latin. as proposed by george (1993: 336), it was probably abstracted from english words like luxury: cf. also trayturi ‘treachery’ : trayta ‘betray’, traytour ‘traitor’ from anglo-french traitourī(e), Me traitour.96 its variant -dury suggests a certain degree of intermixture with native and borrowed words, e.g. lelduri ‘loyalty’ : lel ‘loyal’: cf. lelder ‘loyality’ containing -d-. four words display the suffix -ri as borrowed from Me -(e)ry: harlotri ‘pollution, filth’ : harlot ‘harlot, rogue, rascal’, maystry ‘authority, mastery’ (bK), coyntwyry ‘contrary, cleverness’ (bK) and from the Mco base pystri ‘sorcery, witchcraft, magic’ : *pyst ‘charm’, pystik ‘hurt, injury’. Their gender is not attested, but dictionaries classify them as masculine. if their ending was associated with -uri, -duri and the Mco masculine abstract suffix -i, masculine gender is indeed probable. -(y)ans abstract nouns from borrowed verbs -yjyon deadjectival abstract nouns -ri denominal abstract nouns abstract nouns from borrowed adjectives -uri -duri 1. oco -ant m. 2. Me -a(u)nce, -ence < ofr -ance, -ence f. < lat. -antia, -entia f.97 1. lat. -ītiōnem f. 2. ofr -ison, -ition f.98 3. Me words in -(a)cion, -(a)cioun, -sion97 Me -(e)ry97 < ofr -erie f.98 1. Me words in -ury 2. influence of fr. -ure f. and -erie f.?98 3. association with Mco -i m.? m. ? f. > m. f. > m. (f. ? >) m.? (f. ? >) m.? 94 cf. Watkins 1961: 102, george 1991: 207ff. 95 an unlenited adjective is found in oM 526–7: yma vn posygyon bras war ov wholon ov cothe ‘there is a great heaviness falling on my heart’. 96 cf. also br treitour ‘traitor’. Thomas/Williams 2007: 481 provide a list of french loan words which suggests language contact between cornish and french. not all of the suffixes above were necessarily borrowed via english. 97 cf. Pinsker 1963: 242ff., fisiak 1968: 116ff. 98 cf. nyrop 1979, iii: 81f. gender of abstract noun suffixes in brittonic languages 95 2.2.2.3 Breton 2.2.2.3.1 change of gender 2.2.2.3.1.1 Masculine collectives vs. feminine singulatives in breton, there is a correlation of gender and nominal aspect within the group of collectives and their corresponding singulatives. The former are masculine and the latter feminine due to their feminine suffix -enn. collectives consist partly of former plurals and denote entities which comprise similar items and which normally occur in groups, such as plants, grains or insects and small animals. in addition, substances conceived of as masses are included. The suffixed singulative denotes a single item or an object made out of a substance.99 Welsh uses two suffixes to form singulatives, namely masculine -yn and feminine -en. Their distribution is determined by the gender of the base. This was also the case in obr where both suffixes developed into -enn (see below 3.2.2). breton grammars classify collectives as masculine,100 but due to the loss of gender agreement with non-human plural nouns, the phenomenon is more properly described as non-agreement. br W logod coll. ‘mice’ : logodenn f. ‘mouse’ gwer coll. ‘glass, window pane’ : gwerenn f. ‘(drinking) glass, window pane’ llygod pl. ‘mice’ : llygoden f. ‘mouse’ gwydr m. ‘glass’ : gwydryn m. ‘(drinking) glass, mirror, piece of glass’ collectives underived loss of agreement with non-human pl. nouns m., f. > Ø singulatives -enn phonetic gender assignment m., f. > f. 2.2.2.3.1.2 inherited suffixes in breton the association of feminine gender with abstract nouns, especially with denominal and deadjectival ones, is much stronger than in Welsh and cornish. The brittonic suffixes *-ni and *-oni changed from masculine to feminine gender. also the suffix -der/-ter (< Pcl. *-tero- m.), which forms deadjectival abstract nouns in the northern (= KlT) dialects, was occasionally treated as feminine during the 17th and 18th centuries (hemon 1975: 27). The switch to feminine gender did not occur however, as the suffix is still masculine today. interestingly, Vannetais breton uses 99 cf. favereau 1997: 66ff., irslinger 2009: 22f., irslinger 2010b. 100 e.g. Kervella 1947: 295, favereau 1997: 66: collectives usually show plural agreement, however, singular agreement does occur: cf. favereau 1997: 50. in the latter case collectives trigger masculine agreement. 96 britta irslinger the feminine suffix -ded/-ted in this function. The masculine suffix corresponding to W -did/-tid m., Mco -sys m. < Pcl *-tūto-/-tūti- was abandoned in breton. -der/-ter deadj. abstract nouns in the 17th and 18th c. also fem. m. (> f.) > m. -ni deadj. abstract nouns m. > f. -oni deadj. abstract nouns m. > f. 2.2.2.3.1.3 breton suffixes breton created several complex suffixes through the combination of masculine and feminine suffixes. both -adenn and -adeg101 form action nouns. While nouns in -adenn denote single actions, those in -adeg denote complex or collectively executed actions, e.g. (ober un) neuñviadenn ‘(going) swimming alone’ vs. redadeg ‘running contest’ (favereau 1997: 74). hence, there are suffixes in breton which form action nouns with differing nominal aspects. Within the dialects, where both suffixes are productive, the difference is expressed solely by the morphemes of the suffixes; gender is not involved in perspectivization (but see below, 2.2.2.3.1.6.). The suffix -ad-enn is a combination of the verbal noun suffix -ad m. + -enn f., which probably goes back to sWbr.102 The explanation of -adeg is more difficult.103 on a synchronic level, it may be analyzed as a breton creation composed of the verbal noun suffix -ad m. + -eg f. denoting collections, especially places covered by plants, such as gwerneg ‘place covered with alders’. some words in -adeg denote an action as well as the place where this action is carried out (e.g. marradeg ‘swiddening, slash-and-burn-farming, place cleared by hoe’, nezadeg ‘spinning together, spinning workroom’). The words labouradeg ‘workshop, factory, laboratory’, gwennadeg ‘laundry’, neuiadeg ‘place where swimming is done’ denote only places. in some cases, the corresponding verbs also have or previously had infinitives in -at (marrat ‘to clear by hoe’, labourat ‘to work’, gwennaat ‘to bleach’), but not all of them (nezañ 101 The suffix -adeg is not common in the breton of Vannetais and Pélem, where -erezh is used instead: cf. favereau 1997: 73. 102 cf. irslinger 2010b: 70f. The combination of suffixes is attested once in late co. hanasenn, a derivative in -enn from unattested *hanas ‘sighing’, see above fn. 56. 103 all formations in -adeg mentioned in the following paragraphs as well as their semantics are attested in dictionaries of the 18th and 19th cent.: cf. gib s.vv. The earliest examples, marradec ‘swiddening, slash-and-burn-farming’ and tennadec ‘drawing, pulling, extracting’ can be found in the dictionaries of Julien Maunoir 1659, ed. le Menn 1996. gender of abstract noun suffixes in brittonic languages 97 ‘to spin’, neuial ‘to swim’). however, this pattern may account only for a part of the formations. in fact, the suffix could be older, although there are no traces of -adeg in old or Middle breton.104 russell (1990: 107) points to the complex celtic suffix *-ătĭkā f., which however does not have any cognates in Welsh or cornish. in addition, the Vulgar lat. suffixes -ātĭcā f. and -ātĭcum n. have to be considered, both of which became productive in the romance languages, with ofr -age m. originating from the neuter suffix.105 Many breton words in -adeg, especially terms denoting traditional agricultural activities, have french equivalents in -age. however, the latter may be restricted to regional or older varieties of french, e.g. br marradeg : fr marrage (Vendôme), br foennadeg ‘haying’ : fr fenage, fanage, br tilhadeg ‘seperating hemp fibres from the stalk’ : fr t(e)illage. in some cases, loan words seem to have been integrated into the breton lexicon by replacing -age with -adeg. obviously, this process did not affect the gender of the breton suffix. abstract nouns in -ent-ez are sometimes masculine: cf. favereau (1997: 77). since -ez developed from stems in *-ii̯o/ā- (s. below 3.2.2), words containing simple -ez are both masculine and feminine. This ambiguity may also have influenced the complex suffix. -adenn action nouns: < -ad m. + -enn f. single actions -adeg action nouns: < 1. -ad m. + -eg f. complex actions < 2. celtic *-ătĭkā f.? < 3. Vulgar lat. *-ātĭcā f., -ātĭcum n.? -adurezh abstract nouns < -adur m. + -ezh f. -entez (also m.) abstract nouns < -ent m. (< obr -ant)106 + -ez m./f. -enti/-inti < -ent m. (< obr -ant) + -i m.?107 f. f. f. n. > f. ? f. f. m. > f. 104 favereau 1997: 73 suggests that the suffix originated from obr adac f. ‘cycle, phase of the moon’ used as a second member in composition, e.g. labour-adeg *‘work-period’, *‘period of collective work’ → ‘collective work’. This view is not adopted here. although Moal 1890: 245 (s.v. époque) claims that words in -adeg denote the time when an action is carried out, this is not their original meaning but a metonymic usage occuring frequently with action nouns. 105 cf. fleischmann 1977 for the development of the suffix in the romance languages. although breton may have borrowed some words already from Vulgar latin or from romance languages other than french, no evidence has been adduced for this process so far. 106 cf. fleuriot 1964a: 348. 107 Vgl. Pedersen 1909–1913, ii: 18. 98 britta irslinger 2.2.2.3.1.4 borrowing of feminine suffixes from latin and french feminine suffixes borrowed from latin and/or french preserved their gender in breton. -añs -ijenn, -ision deverbal abstract nouns abstract nouns -iri, -eri deadjectival abstract nouns, places deadjectival abstract nouns -iz < ofr -ance f. < lat. -antia f.98 < 1. lat. -ītiōnem f. < 2. ofr -ison, -ition f.98 < fr -erie f. f. f. f. < 1. lat. -ītia f.108 < 2. ofr -ise98 f. 2.2.2.3.1.5 borrowing and change of gender several neuter suffixes borrowed from latin had to be assigned a new gender. according to fleuriot (1964a: 357), the masculine suffix -adur, forming action nouns and deverbal abstract nouns, is based on latin -atūra f. and -atōrium n. The change of neuter into masculine gender seems natural and may have been supported by the fact that infinitives and action nouns are predominantly masculine in breton. in order to answer the question whether feminine nouns have been extended by -ezh, more detailed research is required. The french masculine suffix -emant normally preserves its gender in breton. hemon (1975: 27) and favereau (1997: 75), however, list several examples of feminine agreement in sources dating from the 16th to the 19th centuries and in modern dialects. The latin suffix -ētum n., which forms nouns denoting plantations, is continued by the breton feminine suffixes -od, -id, -ed (fleuriot 1964a: 358). Their gender is probably influenced by breton formations in -eg, which frequently denote places covered by plants, e.g. gwerneg ‘place covered with alders’. -adur -amant -od, -id, -ed action nouns, deverb. abstr. n. loan words, deverb. abstr. n. plantations 108 cf. Pedersen 1909–1913, ii: 22. < lat. -atūra f. and -atōrium n. f., n. > m. < ofr, fr -emant m. (< lat. -amentum n.)97 < lat. *-ētum m. (> f.) n. > f. gender of abstract noun suffixes in brittonic languages 99 2.2.2.3.1.6 differentiation by gender: -erezh m. vs. -erezh f. according to the handbooks, gender creates a semantic distinction between formations with the suffix -erezh:´ kigerezh f. ‘butchery (place, shop)’ : kig ‘meat’, kiger ‘butcher’ kigerezh m. ‘butchery (handicraft)’ This suffix consists of the agent noun suffix -er < obr. -or and the abstract noun suffix -ezh f. < Pcl *-axtā (Kersulec 2010a: 153). The fact that words with W -(i)aeth and Mco -eth are occasionally masculine suggests that breton continues an older pattern with regard to masculine formations. Pedersen (1909–1913, ii: 32) speculates that the gender of Welsh nouns might go back to *-axtu- m., whereas schumacher (2000: 208) claims that their gender might be the result of a more recent development. Kersulec’s detailed discussion of breton formations in -erezh demonstrates that the correlation between gender and semantics as presented by the grammars does not describe the distributions correctly. his survey of breton dictionaries109 starting with Quiquer’s Nomenclator (1633) reveals that, while authors agree with regard to the feminine gender of locative nouns, there is some variation with regard to abstract nouns. all of the examples in grégoire de rostrenen (1732) are feminine. later and contemporary authors postulate masculine gender next to varying proportions of feminine nouns and/or nouns exhibiting double gender. double gender is also attested in spoken breton. it occurs with nouns ending in -erezh as well as with underived nouns. Previous research110 revealed that these gender differences appear within certain semantic groups e.g. nouns denoting sensory perceptions, appearances, weather and atmospheric phenomena.111 They convey different concepts, e.g. ur bec'h m. ‘a burden’ ~ ar vec'h f. ‘the conflict’, ur brud m. ‘a noise’ ~ ar vrud f. 109 cf. Kersulec 2010a: 216ff. 110 cf. Kervella 1973, favereau 1997: 69f., Plourin 2000: 13ff. see the discussion in Kersulec 2010a: 217ff. 111 cf. Kervella 1973 who also discusses words which were assumed to be the continuation of ancient neuters. The following words show masculine as well as feminine agreement in different breton phrases and constructions: br amzer ‘time, weather’ ~ W amser m., oir, Modir aimser f.; br avel ‘wind’ ~ W. awel f. as these words are purely brittonic, neuter gender cannot be proved. analogical influence is probable in the case of br aer ‘air’, which was attracted by other terms for weather. it is borrowed from latin aer m. like W. aer m., oir, Modir. aer m. (also f.). cf. further br tra ‘thing’ ~ W tra f., br tu ‘side’ ~ W tu m./f., Modir taobh m. The last example is problematic as oir toeb m. o, later f. ā, m. u displays varying inflections. however, none of them indicates neuter gender. 100 britta irslinger ‘the fame, reputation’. in this context, feminine gender correlates with more figurative meanings112 and relates to the speaker’s feelings, thus fulfilling evaluative functions. Plourin (2000: 13ff). introduces an additional difference. according to him, masculine gender correlates with more general and abstract connotations, while feminine correlates with more punctual ones.113 With regard to nouns in -erezh Kersulec’s (2010a: 262ff.) survey of corpora of spoken breton and his own fieldwork yielded several interesting results. action nouns in -erezh are masculine when they refer to the action itself in general or generic terms. in such contexts, they appear either without any article or with the definite article: (1) An tromperezh def.art. deceit.masc. a zo ur pt. is ind.art. bad fault fall defot. ‘(The) deceit is a bad fault.’ (Kerfourn, Kersulec 2010a: 300) (2) Me’, but ar brochennerazh ’veze def.art.knitting.masc. was.hab.prs. rather when was.hab.prs. not kentoc’h p’ veze ket kal’ a dud . . . many of people ‘but, (the) knitting used to be rather done when there were not many people . . .’ (berné, Kersulec 2010a: 287) They are instead feminine when referring to a particular instance of the action, including its result. in these contexts, they typically appear with the indefinite article: (3) Bout be.vn zo broderazh vrav is embroidery.fem mut.beautiful on your jacket à’ ho chilet. ‘There are beautiful embroideries on your jacket.’ (berné, Kersulec 2010a: 288) The following examples created by Kersulec (2010a: 287) include the same noun and demonstrate the functional difference more clearly: (4) *Pa When vije graet an tennerezh pato, . . . was.hab.prs. done def.art. harvest.masc. potatoe ‘When the harvest of the potatoes used to be done, . . .’ 112 cf. favereau 1997: 69. 113 cf. Plourin 2009 on the different strategies of recategorization with regard to nominal aspect in breton. gender of abstract noun suffixes in brittonic languages (5) *Un ind.art. dennerezh pato deus ar gwashañ tout mut.harvest.fem. potato of def.art.worst ever a oa bet an deiz-se pt. was been def.art. day-this 101 ‘This day, there was a potato harvest, which was one of the worst ever.’ Kersulec (2010a: 275) suggests that the indication of double gender in some entries of the dictionary of l’armeyrie (1744) reflects the author’s knowledge of the respective words in different syntactical constructions. This functional difference developed into an aspectual one in the dialect of st-barthélémy (Kersulec 2010a: 265ff.). nouns in -erezh with masculine gender refer to the action itself or to an action in the future, while feminine nouns refer to an action in the past or to its result. in the context of ongoing actions, both genders occur. The following examples include metalinguistic explications by an informant: (6) mard e d’ober if go.2.pl.prs. to do ur farserezh bras, da zonet, ind.art. joke.masc. big to come ’h et d’ober ur farserezh bras pt. go.2.pl.prs to do ind.art. joke.masc. big ‘if you prepare yourself to do a big joke in the future you are going to do a big joke’ (7) ma ’h eus-eñ graet : me’ if you ’h eus graet ur have-it done but you have done farserezh vras ind.art. joke.fem. mut.big ‘if you have done it: but you have done a big joke’ (8) honnezh a zo ur this.fem pt. is farserezh vras, ind.art. joke.fem. pezh zo paset, pezh zo bet mut.big, what is past what is been ‘This is a big joke, something which has passed, which has taken place’ Kersulec bases his conclusions on contradicting material and discusses the different sources as well as their frequent exceptions in detail. The following points have to be taken into consideration: – There are dialects in which action nouns in -erezh are exclusively or almost exclusively feminine. That is, there is some diatopic variation with regard to those action nouns which refer to the action itself. – some speakers show a certain degree of inconsistency. – some contexts allow different interpretations with regard to the function of the noun. – The gender of the equivalent french term may be assigned to the breton word. 102 britta irslinger Kersulec (2010a: 308) speculates that the same development took place in the case of other abstract derivatives such as nouns in -adeg, -der, -ded, -oni, -aj and -adur. however, this assumption has not yet been subjected to research. due to the endangered state of the breton language, it may have become impossible to recover the full scope of the system. despite the uncertainties related to the lack of research, the following hypothesis may be put forward: breton shows signs of a resemantization of grammatical gender according to the feature [±particularizing], in which masculine gender correlates with the notions [+general, abstract, comprehensive, collective] and feminine gender with [+punctual, concrete, individual, singulative]. in addition, feminine gender correlates with the feature [+evaluative].114 The functions assigned to each gender category are detached from derivational suffixes and other formal characteristics of nouns. since those functions can exclusively be inferred within context, they are only incompletely represented or not represented at all in dictionaries, which makes it implossible to date them. a similar type of correlation between gender and function holds for underived nouns and derivatives in -enn. Most typical in this respect are groupings belonging to the collective-singulative-opposition described in 2.2.2.3.1.1. here, bases denoting collections and masses are paired with derivatives in -enn referring to single entities. for a number of pairs, however, this semantic relationship is less clearcut, because the base itself denotes a countable, non-collective entity. according to Trépos (1982: 268), underived bases correlate with the notion [+general, abstract], whereas derivatives in -enn correspond to [+punctual, concrete]. Thus, lod ‘part/share’ is employed before the division has taken place, lodenn afterwards: Pep hini ’no e lod. ‘everyone will get their part/share.’ Brasoc'h eo e lodenn. ‘his part/share is bigger.’ derivatives in -enn may also convey evaluative and hypocoristic notions.115 as the semantic notions occur in conjunction with masculine underived bases and derivatives which are overtly marked by the feminine suffix -enn, this pattern could have served as a model for the functional reinterpretation of gender with regard to nouns in -erezh. if this assumption is 114 This notion is considered a pragmatic function: cf. dressler/Merlini barbaresi 1994: 153ff. some of its features, notably “augmentative” and “diminutive”, show semantic affinities to “particularization”. 115 cf. Trépos 1982: 268, Kersulec (forthcoming). gender of abstract noun suffixes in brittonic languages 103 right, this process could only have started after the sound change -in(n) > -en(n) (see 3.2.2) was completed. 3 Problems of etymology and reconstruction it is difficult to decide upon the original gender of those suffixes whose gender varies across individual languages. such cases typically include ancient neuters116 and nouns based on adjectival patterns. 3.1 Suffix Originating from a Neuter Common Noun The suffix oW, obr -ma, W -fa,117 Mco -va, br -va forming locative nouns is feminine in Welsh and cornish, but masculine in breton. it is assumed to originate from a neuter s-stem which is attested in oir mag (s, n.) ‘plain, cultivated field’ and in gaulish place names.118 it is likely that its plural form is preserved in Welsh place-names like Myddfai, Cilfai (gPc s.v. mai2). for the common noun, the analogical plural maou119 is used in oW and obr. during the medieval period, however, the word fell out of use in both languages. The expected developments are: sg. Pcl Pl. Pcl *magos > *maγos > *maγ > oW, obr ma *magesā̆ > *maγe’(h)a > *maγīa > *maγī > *maγi > oW, obr *maï 120 it remains unclear why the suffix is feminine in Welsh and cornish. The reanalysis of a neuter plural in -a as feminine singular, which is frequently found in the romance languages, is to be excluded here, as the plural is Pbr *maγi > oW, obr *maï. 116 The W suffix -aid (< MW -eit) forms denominal nouns which denote contents and measures and which normally possess the gender of the base, e.g. longaid f. ‘shipload, cargo’ : llong f. ‘ship’, llestraid m. ‘vesselful, bushel, hiveful’ : llestr m. ‘vessel’: cf. Thomas 1996: 665. Zimmer 2000: 291 suggests the reconstruction of a neuter suffix *-ăti̯on. after the loss of neuter gender, the derivatives would have adopted the gender of their bases. This does not seem likely, as such a development is not attested in other cases. also latin possesses suffixes which do not change the gender of the base: cf. leumann 1977: 305. in the romance languages, neuter nouns formed with these suffixes are assigned a different gender category. 117 cf. Zimmer 2000: 370f. for a discussion of the etymology of the variant -fan. 118 cf. goi 216, leia M-8, delamarre 2003: 214, fleuriot 1964a: 370. 119 cf. gPc s.v ma1, fleuriot 1964a: 249, falileyev 2000: 46. 120 cf. schrijver 1995: 390ff. with regard to MW tei ‘houses’ < *tegesa. 104 britta irslinger alternatively, the gender of ma could have been influenced by a noun with similar semantics. however, W lle, Mco le, which appears as a second member in compounds denoting places, is masculine. a probable candidate is then the feminine noun bro ‘region, country, land, vale, lowland’, especially as it frequently occurs in place-names. in addition, its form Pbr *broγ < *broγis < Pcl *mrogis was similar to *maγ. as for the breton noun, there are two possible explanations for its masculine gender: if it results from the standard development neuter > masculine, it means that Pbr *maγ had double gender. alternatively, obr ma, br -va was originally feminine and changed its gender only later, under the influence of lec’h m. ‘place’. 3.2 3.2.1 Suffixes Originally Forming Adjectives W -iant, Br -iant, -ent Pedersen (1909–1913, i: 48) postulates that abstract and action nouns in W -iant, br -iant, -ent were originally feminine (< Pcl *-antī), and that the gender of the Welsh suffix changed over time. he compares nouns belonging to the above group with latin derivatives of the type constantia. This comparison is, however, unlikely to provide any insights with regard to the gender of the brittonic suffixes. Morris Jones (1913: 232) and Zimmer (2000: 394ff). explain that the Welsh suffix -iant stems from the masculine form of the active present participle in Pcl *-ant < Pie *-n̥t. in breton, -iant is rare. both badeziant ‘baptism’ and skiant ‘science’ from latin scientia are feminine. in the case of the former, a possible change of gender is still open for debate. 3.2.2 W -ydd, -edd, Br -ez, MCo -edh These suffixes are based on the Pie thematic suffixes *-i̯o-, *-i̯ā-, which formed denominal adjectives as well as neuter and feminine abstract nouns. in Pbr, *-ii̯o- and *-ii̯ā- form agent nouns and deadjectival abstract nouns. as Pcl ā-stems affected the vowels early Pbr *-i- and *-u- of penultimate syllables to -e- and -o-, the latter had become to some extent a phonetic marker of feminine gender. This function is only partly preserved in Welsh, while in later sWbr the two vowels merged into -e- and the suffixes were reanalyzed as feminine. gender of abstract noun suffixes in brittonic languages 105 adjectives W gwynn m. : gwenn f. ‘white’ br gwenn m./f. ‘white’ Mco gwyn(n) m./f. ‘white’121 < Pcl *u̯indos : *u̯indā suffixes W -yn m. : -en f. br -enn f. Mco -yn m. : -en f.122 W -ydd m. : -edd f. br -ez f. Mco -edh m. < Pbr *-innos : *-innā < Pbr *-ii̯o- : *-ii̯ā123 The suffix *-ii̯ā is typically regarded as feminine singular. however, it may also be interpreted as a continuation of the plural form of *-ii̯om, in which case its gender could be classified as neuter. W llawenydd, llewenydd m. ‘joy’ is reconstructed as *lau̯enii̯o- e.g. by schrijver (1995: 283), whereas the breton cognate levenez ‘joy’ is traced back to *lau̯enii̯ā by fleuriot (1964a: 88), since the word is feminine in Modbr. in obr, the noun is attested as louuinid and in oW as leguenid. according to schrijver (2011: 18, 22), the oW spelling represents the osWbr vowel /ı/, oW /ɨ/ rather than the vowel resulting from ā-affection, which should be spelled <e>. based on this observation, the question arises whether the obr word might represent the masculine (< neuter) noun *lau̯enii̯o-, which became feminine after the lowering of /ı/ to /e/. alternatively, the Modbr word might be a direct continuation of the neuter plural.124 The same development could be suggested for the ModW suffix -edd, since it is also likely to stem either from feminine singular or from neuter plural *-ii̯ā. This assumption then leads to the question of whether neuter plurals were always reinterpreted as feminine singulars or whether they became masculine singulars as well. abstract nouns like W mawredd ‘bigness’, noethedd ‘nakedness’ are masculine despite the occurrence of -e- in their suffix. nonetheless, many of them occur also with feminine agreement. 121 for the devopment *e to *ı in cornish: cf. the following fn. 122 cf. schrijver 1995: 260, 324; 2011: 43f., fleuriot 1964a: 49f., 234. Middle cornish retains several spellings with -yn. according to schrijver 2011: 43, this may be due to the tendency of *e to become *ı before nasals in sWbr and thus it would not confirm the existence of the masculine suffix. 123 cf. schrijver 1995: 288, no. 2. 124 in the case of lat. gaudium n. ‘joy’, the plural gaudia is continued e.g. by french joie and italian gioia (both feminine). cf. schön 1971: 51f. on the frequency of singular and plural forms of the latin word. 106 britta irslinger While breton has generalized feminine gender, cornish seems to have a mixed distribution like Welsh which probably includes several cases of double gender. The distribution described by the handbooks looks artificial: while a generalization of masculine gender is assumed for Mco -edh, its younger variant -a is claimed to be feminine despite the general expansion of masculine gender in cornish. The foregoing discussion shows that the nouns included in the present group vary with regard to their origins and some of them may have changed their genders several times. due to the lack of documentation, it is impossible to track these changes, which is why the reconstruction of proto-forms for individual words is not attempted here. in this respect, the gender development of latin o/ā-stems in the romance languages constitutes a good parallel, since it is not uniform and, among other things, dependent on the frequency of certain (case) forms, semantic differentiation, reanalyses etc. (schön 1971). The widespread practice to infer nominal gender of reconstructed Pbr, Pcl or even Pie nouns from the gender of these nouns in the medieval or modern brittonic languages does not generate any reliable results. Welsh -edd productive examples possible developments mawredd m. ‘bigness’ gwirionedd m./f. - f. sg. > m. sg. - n. pl. (> f. sg.?) > m. sg. -ydd llewenydd m. ‘joy’ unproductive carennydd m./f. ‘kinship’125 breton -ez cornish -edh -edh > -a - m. sg. - n. sg. > m. sg. gwirionez f. ‘truth’ levenez f. ‘joy’, obr louuinid (m.?) karantez f. ‘love’ - f. sg. - n. pl. > f. sg. - n. sg. > m. sg. > f. sg. meuredh m. ‘bigness’ gwiryonedh m. ‘truth’126 lowena f. ‘joy’127 kerensa f. ‘love’ - f. sg. > m. sg. - n. pl. (> f. sg.?) > m. sg. 125 Pedersen 1909–1913, i: 373 relates this noun to the oir neuter cairde ‘pact, peace’, but parallel formations are possible: cf. leia c-37. 126 The article followed by unlenited noun is attested in Pc 1974: an gwyryoneth. 127 The noun followed by unlenited adjective is attested in Pc 1042: Lowenna tekca gothfy ‘the fairest joy thou knowest’. gender of abstract noun suffixes in brittonic languages 107 4 conclusions for the brittonic languages, gender plays a role with regard to the classification of the lexicon, notably the secondary lexicon. although all brittonic languages display remarkable similarities with regard to their nominal morphology, the restructuring which each language experienced varies considerably. The Welsh inventory of suffixes is the most conservative one, containing only celtic suffixes or early Pbr borrowings from latin.128 no borrowings from english are documented. Welsh suffixes are predominantly masculine, which might be due to the fact that neuter suffixes took on masculine gender when neuter gender was lost. in historical times, the change from feminine to masculine gender is documented in several instances, while the reverse process is rarely attested. differentiation according to gender is thus very limited in the secondary lexicon of Welsh. over the long term, this could reduce the complexity of classification in this part of the lexicon, so that gender might be restricted to primary nouns or even animates as it is in english. breton, on the other hand, developed a system similar to french by increasing the number of feminine suffixes through change of gender, the creation of new suffixes and borrowing. This might explain the higher proportion of feminine suffixes compared to Welsh and cornish. Thus, breton constitutes a case where, partly due to language contact, the general expansion of the masculine gender came to a halt and the feminine gender prospered. The analysis of the cornish system yielded largely inconclusive results due to the lack of textual attestations showing agreement. suffixes created on the basis of cornish morphemes are masculine. on the other hand, the loan suffixes -(y)ans and -yjyon have undergone a change from feminine to masculine, assuming that they have been borrowed from the romance language family. although a few minor instances are attested in which gender and nominal aspect are related, the association of feminine gender with abstractness may not be confirmed for brittonic. it seems that this was already the case in Pbr and it might be linked to the loss of endings which frequently served as overt gender markers. before such a conclusion is drawn, however, more research factoring in the irish language will be necessary. 128 This is also true for suffixes not attested or lost in breton like -dwr or -eb. 108 britta irslinger according to Matasović (2004: 119ff., 124), abstract nouns (with suffixes other than *-(a)h2-) were either feminine or neuter in late Pie. The evidence he provides consists mainly of feminine ti- and neuter s-stems. none of the two suffixes was productive in celtic, so that a possible inheritance of the correlation between gender and types of abstract nouns can be ruled out. The development of initial mutations, which indicate gender also in the modern languages, has been influenced considerably by o- and ā-stems. These, however, have not been attributed any functional correlations.129 breton or at least some of its regional or situational (oral) varieties have been claimed to employ a functional gender-system. With regard to the nature of this system, several hypotheses have been put forth. however, there is no consensus concerning its central characteristics, such as the nouns involved or the semantic difference produced by the gender change. according to Kersulec’s interpretation of the data, the central feature is [±particularizing]. breton would thus be an ideal example to sustain Weber’s hypothesis. as it has been stated previously (irslinger 2010a), functional gendersystems employing gender independently of nominal suffixes are rather rare. They have been found to work within smaller parts of the nominal lexicon, but do not cover the complete inventory of nouns of any given language.130 although the classification of nouns according to a specific feature [±particularizing] is clearly a function of gender in ie languages, it is never gender alone that brings it about. instead it involves a 129 note also that the suffixes *-ā and *-ī, which became productive for the derivation of nouns denoting females in many ie languages, are not employed in this function in insular celtic (de bernardo stempel 1995: 432). 130 Weber 2001: 39ff. claims that this function is found with the genders of Wintu (a Penuti-language in northern califonia), but there are reasonable doubts whether Wintu is a gender language at all. Wintu distinguishes nouns according to the notions [generic] ~ [particular] by forming noun stems with the help of two suffixes (including a range of allomorphs). Pitkin 1984: 202 assumes a classification of nouns according to nominal aspect. The resulting aspectual differences are rarely translated accurately, but they often involve the distinctions [animate] ~ [inanimate] and [singular] ~ [plural] (ibid. 211). however, according to unterbeck’s definition (above 1), a gender language should include the feature [+sex], but Wintu does not even include gendered pronouns. in older descriptions of Wintu, these distinctions between gender and other systems of nominal classification are not yet considered. demetracopoulou-lee 1942, who first described the aspectual difference, makes random use of the terms “noun classes”, “categories” and “genders”. The latter term is employed exclusively in Johannes lohmann’s introduction to demetracopoulou-lee’s article. gender of abstract noun suffixes in brittonic languages 109 considerable number of lexical, syntactical and morphological strategies. This type of classification is therefore not the “proper” function of gender which Weber (2000: 496) has been looking for. abbreviations Grammatical terms aor. def.art. f., fem. hab. imp. ind.art. ipf. m., masc. mut. n. ppf. pron. prs. rel. pt. sj. vn. aorist definite article feminine habitual imperative indefinite article imperfect masculine initial mutation neuter pluperfect pronoun present relative particle subjunctive verbal noun Languages br co e fr icl ie ir lat. M Mod o Pbr Pcl Pie sWbr W V breton cornish english french insular celtic indo-european irish latin Middle (e.g. Mco = Middle cornish) Modern (e.g. Modbr = Modern breton) old (e.g. oW = old Welsh) Proto-brittonic Proto-celtic Proto-indo-european south-West-brittonic Welsh Vannetais breton 110 britta irslinger references benzing, Johannes. 1955. Lamutische Grammatik. Mit Bibliographie, Sprachproben und Glossar. (Veröffentlichungen der orientalischen Kommission, akademie der Wissenschaften und der literatur 6) Wiesbaden: steiner. ——. 1955 [1956]. Die tungusischen Sprachen. Versuch einer vergleichenden Grammatik. (abhandlungen der geistes- und sozialwissenschaftlichen Klasse, akademie der Wissenschaften und der literatur in Mainz 1955, 11) Mainz: Verlag der akademie der Wissenschaften und der literatur. bK = graham Thomas and nicholas Williams (eds.). 2007. Bewnans Ke. The life of St Kea. A critical edition with translation. exeter: univ. of exeter Press, The national library of Wales. bM = Whitley stokes (ed.). 1996. Beunans Meriasek/Bywnans Meryasek. converted to Kernewek kemmyn by Keith syed. sutton coldfield: Kernewek dre lyther. brown, Wella. 1993. A Grammar of Modern Cornish. [s.l.], 2nd ed. callington: Kesva an Taves Kernewek. corbett, greville g. 2000. Number. cambridge: cambridge university Press. cW = The Creacion of the World. A Critical Edition and Translation. ed. and transl. by Paula neuss. (garland Medieval Texts 3) new York and london: garland 1983. dalton-Puffer, christiane. 1996. The French influence on Middle English morphology. A corpus-based study of derivation. (Topics in english linguistics 20) berlin and new York: Mouton de gruyter. de bernardo stempel, Patrizia. 1995. Zum genus femininum als ableitbarer Kategorie im Keltischen. in Wojciech smoczyński (ed.), Analecta Indoeuropaea Cracoviensia II. Kuryłowicz Memorial Volume. Part One, 427–46. cracow: universitas. ——. 1999. Nominale Wortbildung des älteren Irischen. Stammbildung und Derivation. (buchreihe der Zeitschrift für celtische Philologie 15) Tübingen: niemeyer. de rostrenen, grégoire. 1732. Dictionnaire françois-celtique ou françois-breton. rennes: Julien Vatar. delamarre, xavier. 2003. Dictionnaire de la langue gauloise. Une approche linguistique du vieux-celtique continental. 2e éd. rev. et augm. (collection des hesperides) Paris: Éditions errance. demetracopoulou-lee, d. 1942. noun categories in Wintu. The generic and the particular. Zeitschrift für vergleichende Sprachforschung 67.197–210. deshayes, albert. 2003. Dictionnaire étymologique du breton. douarnenez: chasse-Marrée. di Meola, claudio. 2007. genuszuweisung im deutschen als globaler und lokaler strukturierungsfaktor des nominalen lexikons. Deutsche Sprache 35/2.138–58. doyle, aidan. 1992. Noun derivation in Modern Irish. Selected categories, rules and suffixes. lublin: red. Wyd. Katolickiego uniw. lubelskiego. dressler, Wolfgang, and lavinia Merlini barbaresi. 1994. Morphopragmatics: Diminutives and Intensifiers in Italian, German, and Other Languages. (Trends in linguistics / studies and monographs 76) berlin and new York: Mouton de gruyter. evans, simon d. 1964. A Grammar of Middle Welsh. (Medieval and Modern Welsh series, suppl. vol.) dublin: inst. for advanced studies (repr. 1989). faiss, Klaus. 1992. English historical morphology and word-formation. Loss versus enrichment. (fokus 8) Trier: WVT. falileyev, alexander. 2000. Etymological Glossary of Old Welsh (buchreihe der Zeitschrift für celtische Philologie 18). Tübingen: niemeyer. favereau, francis. 1997. Grammaire du breton contemporain / Yezhadur ar brezhoneg a-vremañ. Morlaix: skol Vreizh. fisiak, Jacek. 1968. A short grammar of Middle English. I: graphemics, phonemics and morphemics. Warszawa and london: PWn and oxford university Press. fleischer, Wolfgang, and irmhild barz. 1995. Wortbildung der deutschen Gegenwartssprache. unter Mitarbeit von Marianne schröder. 2nd ed. Tübingen: niemeyer. gender of abstract noun suffixes in brittonic languages 111 fleischman, suzanne. 1977. Cultural and linguistic factors in word formation. An integrated approach to the development of the suffix -age. (university of california publications in linguistics 86) berkeley: university press. fleuriot, léon. 1964a. Le vieux breton. Éléments d’une grammaire. Paris: Klincksieck. ——. 1964b. Dictionnaire des gloses en vieux breton. Paris: Klincksieck. george, Ken. 1991. The noun suffixes -ter/-der, -(y)ans and -neth in cornish. Études Celtiques 29.203–13. ——. 1993. Gerlyver Kernewek kemmyn. An gerlyver meur. Kernewek—Sowsnek. Cornish— English Dictionary. seaton, Torpoint: cornish language board. gib = roparz hemon. 1979–1998. Geriadur istorel ar brezhoneg / Dictionnaire historique du breton. 2. éd. Plomelin: Preder. gM = h. Meurig evans and W.o. Thomas. 1989. Y geiriadur mawr. The complete WelshEnglish English-Welsh dictionary. 15th ed. llandybïe: christopher davies and gwasg gomer. goi = rudolf Thurneysen. 1946. A grammar of Old Irish. rev. and enlarged ed. Transl. from the german by d.a. binchy and osborn bergin. dublin: institute for advanced studies (repr. 1993). gPc = r.J. Thomas et al. (eds.). 1950–. Geriadur Prifysgol Cymru. A dictionary of the Welsh language. caerdydd: gwasg Prifysgol cymru. grandi, nicola, and sergio scalise. 1999. des règles d’altération nominale en italien. Silexcales 2.83–93. hemon, roparz. 1975. A historical morphology and syntax of Breton. (Medieval and Modern breton series 3) dublin: institute for advanced studies (repr. 1984). irslinger, britta. 2002. Abstrakta mit Dentalsuffixen im Altirischen. heidelberg: Winter. ——. 2009. singulativ und Kollektiv in den britannischen sprachen. in uwe hinrichs, norbert reiter and siegfried Tornow (eds.), Eurolinguistik. Entwicklung und Perspektiven, 233–53. (eurolinguisitische arbeiten 5) Wiesbaden: harrassowitz. ——. 2010a. genus und nominalaspekt. Historische Sprachforschung (2009 [2010]), 122.1–30. ——. 2010b. les dérivés avec gallois, cornique -yn/-en, breton -enn et irlandais -ne: fonction et sémantique. La Bretagne Linguistique 15.43–81. Jackendoff, ray. 1991. Parts and boundaries. Cognition 41.9–45. Jackson, Kenneth h. 1953. Language and history in Early Britain. edinburgh: university Press (dublin: four courts Press 61994). Kersulec, Pierre-Yves. 2010a. Étude morphologique des noms en -erezh en breton moderne. Morphologie et grammaire. Manuscrit de thèse université rennes ii. (pdf may be downloaded at http://morphobretonne.weebly.com/). ——. 2010b. les noms en -erezh à référence collective: esquisse de classement. La Bretagne Linguistique 15. ——. (forthcoming). remarques sur les noms en -enn construits sur un nom singulier comptable en breton contemporain. in conference proceedings edited by denis costaouec and Tanguy solliec. emgleo breiz (the complete bibliographic information will be given on http://morphobretonne.weebly.com/). Kervella, franzes. 1947. Yezhadur bras ar brezhoneg. brest, al liamm (31995). ——. 1973. Kudennoù reizh ha kudennoù niver e brezhoneg. Hor Yezh 86–87.71–84. Kilarski, Marcin, and Peter Trudgill. 2000. The function of linguistic gender: a discussion. Poznan Studies in Contemporary Linguistics 36.191–201. Köpcke, Klaus-Michael, and david a. Zubin. 2009. genus. in elke hentschel and Petra M. Vogel (eds.), Deutsche Morphologie, 132–154. (de gruyter lexikon) berlin et al.: de gruyter. Krahe, hans, and Wolfgang Meid. 1969. Germanische Sprachwissenschaft. III: Wortbildungslehre. (sammlung göschen 2234) berlin: de gruyter. l’armeyrie [de Kerampoul, cillart]. 1744. Dictionnaire françois-breton ou françois-celtique du dialecte de Vannes, par Monsieur l’A***. leide, compagnie. 112 britta irslinger leia = J. Vendryes, e. bachellery and P.-Y. lambert. Lexique étymologique de l’irlandais ancien, a 1959; MnoP 1960; rs 1974; Tu 1978; b 1980; c 1987; d 1996. dublin and Paris: dublin institute for advanced studies and cnrs. le Menn, gwennole (ed.): Les dictionnaires français-breton et breton-français du R.P. Julien Maunoir 1659. réédité et annoté par gwennole le Menn. 2 vols. [saint-brieuc] (bibliothèque bretonne). lewis, henry. 1990. Handbuch des Mittelkornischen. deutsche bearbeitung von stefan Zimmer. Mit einem bibliographischen anhang von andrew hawke. (innsbrucker beiträge zur sprachwissenschaft 61) innsbruck: institut für sprachwissenschaft der universität inssbruck. lewis, henry, and holger Pedersen. 1961. A concise comparative Celtic grammar. repr. with corr. and a suppl. göttingen: Vandenhoeck & ruprecht. leumann, Manu. 1977. Lateinische Laut- und Formenlehre. (lateinische grammatik von leumann/hofmann/szantyr bd. 1) München: beck. Matasović, ranko. 2004. Gender in Indo-European. heidelberg, Winter. ——. 2007. linguistic convergence in the british isles in the roman and early medieval periods. in hildegard l.c. Tristram (ed.), The Celtic languages in contact. Papers from the workshop within the framework of the XIII International Conference of Celtic Studies, Bonn, 26–27 July 2007, 93–112. Potsdam: Potsdam university Press (urn: urn:nbn:de: kobv: 517–opus-15682, url: http://opus.kobv.de/ubp/volltexte/ 2007/1568.) Moal, J. 1890: Supplément lexico-grammatical au dictionnaire français-breton du colonel Troude en dialecte de Léon. landerneau. Morris Jones, John. 1913. A Welsh grammar. Historical and comparative. oxford: clarendon (repr. 1955). nance, Morton r. 1955. A Cornish-English dictionary. The federation of old cornwall societies. repr. [s.l.], cornish language board 1971. norris, edwin. 1859. The ancient Cornish drama. 2 vols. ed. and transl. oxford (repr. new York and london: benjamin blom 1968). nyrop, Kristoffer. 1979. Grammaire historique de la langue française. 6 vols. 4. éd. rev., réimpr. des éd. de Paris-copenhague, 1914–1960. genève: slatkine reprints. oM = Origo mundi, first play of the Cornish mystery cycle, the Ordinalia. a new edition by Phyllis Pier harris. university of Washington dissertation 1964 (see also norris 1859 i.) Pa = ray edwards (ed.). 1996. Pascon agan Arluth. passhyon agan Arloedh (Kan an passhyon/The passion poem/Mount Calvary). ed. 5th print. sutton coldfied: Kernewek dre lyther. Padel, oliver J. 1985. Cornish place-name elements. nottingham: english Place-name society. Pc = Passio domini nostri Jhesu Christi = norris 1859 i. Pedersen, holger. 1909–1913. Vergleichende Grammatik der keltischen Sprachen. 2 bde. göttingen: Vandenhoeck & ruprecht. Pinsker, hans e. 1963. Historische englische Grammatik. Elemente der Laut-, Formen- und Wortbildungslehre. 2., verb. u. erw. aufl. München. Pitkin, harvey. 1984. Wintu grammar. (university of california publications in linguistics 94) berkeley: university of california Press. Plourin, Jean-Yves. 2000. Initiation au Breton familier et argotique. crozon: Éditions armeline. ——. 2009. continu et discontinu dans le groupe nominal. La Bretagne linguistique 14.143–74. Quiquer, guillaume. 1633. Le nomenclator latin-français-breton (1633) en 2 tomes, avec index général de Guillaume Quiquer de Roscoff. Réédité et annoté par Gwennole Le Menn. (bibliothèque bretonne 10) saint-brieuc: skol (2000). rd = Ressurectio domini = norris 1859 ii. rijkhoff, Jan. 1991. nominal aspect. Journal of Semantics 8.291–309. gender of abstract noun suffixes in brittonic languages 113 russell, Paul. 1990. Celtic word-formation. The velar suffixes. dublin: dublin institute for advanced studies. ——. 2002. The noun phrase. oxford: oxford university Press. schön, ilse. 1971. Neutrum und Kollektivum. Das Morphem -a im Lateinischen und Romanischen. (innsbrucker beiträge zur sprachwissenschaft 6) innsbruck: institut für sprachwissenschaft der universität innsbruck. schrijver, Peter. 1995. Studies in British Celtic historical phonology. (leiden studies in indoeuropean 5) amsterdam: rodopi. schumacher, stefan. 2000. The historical morphology of the Welsh verbal noun. (Maynooth studies in celtic linguistics iV) Maynooth: department of old irish, national university of ireland. ——. 2011. old british. in Ternes (2011), 1–84. surridge, Marie e. 1989. factors for the assignment of grammatical gender in Welsh. Études Celtiques 26.187–209. Ternes, elmar (ed.). 2011. Brythonic Celtic—Britannisches Keltisch. From Medieval British to Modern Breton. (Münchner forschungen zur historischen sprachwissenschaft 11) bremen, hempen. Th = The Tregear homilies [Johannes Tregear . . . Publ. by christopher bice] [repr. of the ed. of 1969/new ed. by ray edwards]. 1994. [liskeard]: Kesva an Tavas Kernewek. Thomas, graham, and nicholas Williams (eds.). 2007. Bewnans Ke. The life of St Kea. A critical edition with translation. exeter: univ. of exeter Press, The national library of Wales. Thomas, Peter Wynn. 1996. Gramadeg y Gymraeg. caerdydd: gwasg Prifysgol cymru. Trépos, Pierre. 1982. Le pluriel breton. 2. éd. [s.l.], brud nevez. Trudgill, Peter. 1999. language contact and the function of linguistic gender. Poznan Studies in Contemporary Linguistics 35.133–52. unterbeck, barbara. 2000. gender. new light on an old category. an introduction. in unterbeck/rissanen (2000), xv–xlvi. unterbeck, barbara, and Matti rissanen (eds.). 2000. Gender in grammar and cognition. I Approaches to gender. II Manifestations of gender. berlin and new York: Mouton de gruyter. Vc = eugene Van Tassel graves (ed.). 1962. Vocabularium Cornicum. The Old Cornish vocabulary. columbia university dissertation. Vogel, Petra Maria. 1996. Wortarten und Wortartenwechsel. Zu Konversion und verwandten Erscheinungen im Deutschen und in anderen Sprachen. berlin and new York: de gruyter. ——. 2000. nominal abstracts and gender in Modern german: a “quantitative” approach towards the function of gender. in unterbeck/rissanen (2000), 461–93. Watkins, T. arwyn. 1961. Ieithyddiaeth. Agweddau ar astudio iaith. caerdydd: gwasg Prifysgol cymru. Weber, doris. 2000. on the function of gender. in unterbeck/rissanen (2000), 495–509. ——. 2001. Genus. Zur Funktion einer Nominalkategorie exemplarisch dargestellt am Deutschen. frankfurt a. M.: Peter lang. Williams, nicholas J.a. 2000. English-Cornish dictionary. Gerlyver Sawsnek-Kernowek. dublin and redruth: everson gunn Teoranta and agan Tavas. ——. 2011. Middle and late cornish. in Ternes (2011), 237–355. Williams, robert. 1865. Lexicon Cornu-Britannicum. llandovery and london: roderic and Trubner. Williams, stephen. 1980. A Welsh grammar. caerdydd: university of Wales Press. Wmffre, iwan. 1998. Late Cornish. München: lincom (repr. 1999). Zimmer, stefan. 2000. Studies in Welsh word formation. dublin: school of celtic studies and institute for advanced studies.