[Nummer 2], Tijdschrift voor Nederlandse Taal- en Letterkunde. Jaargang 129 - DBNL (2023)

[pagina 87]

[p. 87]

[Nummer 2]

Tinne van Rompaey
A diachronic account of Dutch -nis, -heid, -dom and -schap
Rivalry within the paradigm of abstract suffixesGa naar voetnoot*

Abstract - In this case study I will investigate the history of the Dutch suffixes -nis (‘-ness’), - heid (‘-hood’), -dom (‘-dom’) and -schap (‘-ship’) in abstract nouns such as hechtenis (‘custody’), wijsheid (‘wisdom’), rijkdom (‘richness’) and moederschap (‘motherhood’). Even though they appear to have interchangeable schematic meanings at first sight, I will argue that over time they have established distinct semantic frames, different morphological patterns and varying degrees of productivity. By means of a corpus-based investigation I will examine the semantic and morphophonological changes in their development from 12^th century to present-day Dutch. My reconstruction shows that, despite their different sources, all the suffixes under investigation undergo parallel processes of semantic generalisation and reanalysis. This study differs from others such as Trips (2009) on English -hood, -dom and -ship in that it includes the hitherto underresearched suffix -nis, which, unlike -heid, -dom and -schap, did not originate in autonomous nouns. Yet, when all suffixes meet in the paradigm of abstract suffixation, they display similar characteristics and engage in interaction and rivalry, leading to different diachronic paths and distinct present-day Dutch functions.

1 Introduction

Abstract nouns such as beauty, growth or love are generally defined as describing ‘typically non-observable and non-measurable’ entities (Quirk et al. 1985: 247). They do not refer to a concrete perceptible or accessible entity in the external world, but to generic concepts conceived in the mind. Most abstract nouns are nominalizations containing at least one free morpheme which refers to a certain quality (expressed by an adjective, e.g. truth from true), a process (expressed by a verb, e.g. growth from grow) or an instantiation of a quality or process (expressed by a noun, e.g. wonderness from wonder). The element encoding the abstract meaning can be a zero-morpheme (e.g. love), but is very often a phonetically represented sign, i.e. a nominalising affix that is, at least in Germanic languages, mostly added at the end of the word (following the ‘right hand head rule’, Trommelen & Zonneveld 1986).

From synchronic studies on word formation we know that the Dutch language

[pagina 88]

[p. 88]

has many suffixes at its disposal to form abstract nouns. De Vooys (1976: 214) mentions -ing, -er, -de/-te, pays special attention to -(e)nis, -dom, -schap, -heid and also discusses -wezen and -rijk. De Vries (s.d.: 87) adds -igheid, -(e)ment, -(i) teit, -(er)ij, -atie, -laag and obsolete -i/-e to this list, De Haas & Trommelen (1993) -ie, -asme, -ade, -uur and -se, and Algemene Nederlandse Spraakkunst (1997) includes the suffixes -isme, -age, -atie and -st. Vercoullie (1922: 112-115) makes a fundamental distinction between -m, -sem, -sel, -nis, -e, -de, -te, -st, -ing and -lijk and what he calls the ‘autonomous suffixes’ -dom, -heid, -schap.

As is the case with free morphemes, the present variety of abstract suffixes and their current distribution is the result of diachronic language processes. In this case study I will trace the history of the in origin Germanic abstract suffixes -heid, -dom, -schap and -nis, which all exhibit strong similarities in their semantic and compositional make-up, as exemplified by the co-existence of English wisdom, Dutch wijsheid and German Weisheit (‘wis-hood’), En. darkness and goodness corresponding to Dut. duisternis (‘dusk-ness’) and goedheid (‘good-hood’), En. bishophood versus Dut. bisschopdom (‘bishop-dom’), En. priesthood versus Ger. Priesterschaft (‘priest-ship’), En. drunkenness versus Dut. dronkenschap (‘drunken-ship’) and Ger. Mutterschaft (‘mother-ship’) versus En. motherhood. This article will investigate how the four main abstract suffixes in Dutch, three originating in nouns and the fourth one in an earlier affix, arrived at the present stage of apparent semantic and compositional similarity. I will reconstruct how the suffixes, once they had entered the paradigm of abstract suffixation, engaged in rivalry with each other. I will zoom in on crucial moments in their development when they took over semantic functions and morphological patterns from each other via processes of analogy and reinterpretation, resulting in semantic and morphological overlap in their functional domains. This fuelled the competition between the suffixes and lead to shifts in productivity, with one suffix becoming more productive than the other, or even in suffix substitution in already existing derivations.

This diachronic study will be structured as follows. In section 2 I will briefly introduce the suffixes -nis, -heid, -schap and -dom and their origins and show that their development involved processes of semantic bleaching, generalization and reanalysis. In section 3 I will look at the specific trajectories of the suffixes throughout the stages of the Dutch language with again focus on reanalysis or reinterpretation and subsequent extension of patterns. In the concluding section 4 I will summarize the main lines of this development, which I will interpret as a case of rivalry within the paradigm of abstract nominalization suffixes, yielding different diachronic paths and distinct present-day functions.

2 Origins and developments of the abstract suffixes

Accounts of the evolution of -dom, -heid en -schap have revealed that these suffixes originated in autonomous nouns (e.g. Schönfeld 1970 and Trips 2009). A particularly clear example is the Indo-European noun *(s)kai (‘image, sign, shape’), which evolved into the suffix -heid. As a first step in the process of morphologisation, the noun assumed a more abstract meaning, such as ‘rank’, ‘position’, ‘dignity’, ‘standing’ or ‘honour, as in Old Saxon hēd, Old High German heit, Old

[pagina 89]

[p. 89]

Norwegian heiðr and Old English hād(or). Examples of the autonomous noun use conveying the meaning ‘manner’ are found in Gothic haidus and in Old High German heidim. Similarly, the suffix -schap stems from the autonomous noun *skap (‘creation, creature’, see Schönfeld 1970: 202 and Trips 2009: 121), which in its turn evolved via zero-derivation or implicit transposition from the Indo-European verb *skapi-z. As observed by Schönfeld (1970: 202) and De Vries & De Tollenaere (2004), *skap led to nouns ga-skafts in Gothic, scaf in Old High German, Old Norwegian skap (‘shape, kind’), Old English gesceap (‘appearance’) and Old Saxon giskaft (‘fate’). The source noun from which the suffix -dom was de-rived, dďm, was as such already a derivation with an abstract suffix, namely Indo-European -m /-moz (oed 1989) or infix -m-, -mo(n)-, -men- of -mi- (Schönfeld 1970: 203). Through addition of this /m/-suffix to the verbal root dō- (‘to place or to put’, Wilmanns 1930: 392), a new noun dō-m (‘situation’) came about, which finally resulted in what Schönfeld characterizes as the exclusively WestgermanicGa naar voetnoot1 suffix -dom (1970: 201).

So far, the etymology of the suffixes conforms to the widespread hypothesis (Booij & van Santen 1998: 273) that almost all bound morphemes arise out of free words, i.e. ‘[...] the main diachronic source of affixes is in grammaticalized and agglutinated previous full words’ (Haspelmath 1994: 2). A couple of adjectival morphemes also originated in this way, for instance Dutch -lijk and English -ly from the Gothic noun leik (‘body’) or Dutch -baar from the verb baren (cf. Got. bairan, ‘carry’). Several linguists have considered their emergence as a case of either grammaticalization (Lehmann 2002: 14) or lexicalization (Hopper & Traugott 2003: 58), depending on which status, lexical or grammatical, they assign to the category of derivational affixes. Leaving aside the question of whether grammaticalization or lexicalization is at stake, it can be noted that these two diachronic developments have at least one process in common, namely desemantization or semantic attrition (Lehmann 1985: 306) of one of the components. The purely lexical, concrete noun, for instance ie *skap (‘shape’) or ie *(s)kai (‘image’), first acquired a more abstract, metaphorically related (Traugott & König 1991: 190) meaning, namely ‘kind/sort’ or ‘situation/manner’ as in on skap or Got haidus. As a consequence, *haiduz, *skap and dōm could occur in contexts in which they were not found before, i.e. they generalized (Bybee 2003: 152). Pivotal in the further development from noun to affix was the use of these semantically bleached nouns in compounds. The presumed development from noun to suffix can thus be represented as in Figure 1.

Figure 1 Presumed cline from noun to suffix (taken from Lightfoot 2005:596)

However, as observed by Lightfoot (2005: 594), information about the intermediate stage between isolated noun and abstract compound, the ‘pre-compounded

[pagina 90]

[p. 90]

syntagm’, is often lacking: ‘evidence for the presumed pre-compounded phases is rare’, regarding, for instance, ‘the time when magad and heit [from Old High German magadheit, ‘virginity’] would have appeared in written form as two independent nouns syntactically poised to fuse’. Moreover, such rare data often illustrate incidental cases of ‘layering’ or ‘renewal’ (Hopper & Traugott 2003: 9 and Fischer 1997: 156) rather than the transition from simple word to compound. In example (1), for instance, it is clear that schap acquired a more abstract meaning in ridderscape, but the co-occurrence of the autonomous word and the suffix is probably due to divergent language processes in which one path led to the emergence of the suffix and another to the Middle Dutch noun schape (‘creature’, ‘kind’). The same holds for the suffix -dom, which existed alongside etymologically related Old Dut. duom, Middle Dut. do(e)m (‘judgement’).

Together with a reduction in semantic weight, the suffixes lost their syntactic freedom, first by being integrated in a compound and later by turning into a derivational suffix. A sequence of words, e.g. uuizent and heit (cf. ‘wis’ and ‘dom’), fused when the syntagmatic combination became entrenched in the mental lexicon as one ‘re-packaged [...] single processing unit’ (Bybee 2003: 153). The original semantics of the second part of the compound were then likely to disappear (e.g. isolated *skap, ‘creation’) and the abstract meaning (e.g. -scap/-scip, ‘state’) became permanently coded in the emerging suffix. The morphological boundaries were disregarded and the combination of two words was reinterpreted as one word, which is, in a sense, a case of rebracketing, as ‘fusion involves changes in the assignment of boundaries’ (Hopper & Traugott 2003: 41).

The abstract suffix -nis(se) seems to be an exception to the noun-to-affix cline, since a possible nominal root for this morpheme has not been found so far. Instead it originated in the Indo-European infix *-tu- or Germanic *-þu- or *-du-, which formed abstract nouns from verbs ending in a vowel and a dental plosive (Kluge 1926: §137). When attached to Gothic verbs in -at or -atjan (with causative suffix -jan), reanalysis and following phonological assimilation (Schönfeld 1970: 211) of /t/ and /þ/ into -at and -þu- yielded the suffix *-assuGa naar voetnoot2, as in the Gothic example (2). This formal reanalysis, I believe, led to a comparable development as the reanalysis involved in the development from autonomous nouns into -heid, -dom and -schap. Because of the highly frequent application and the semantic bleaching of the suffix, the output of the word formation rule and the morphological boundaries became less transparent with all four suffixes. Rebracketing

[pagina 91]

[p. 91]

took place as -at and -þu- merged and at the same time, the suffix *-assu expanded its derivational domain even more: it became relatively productive with verbs that did not end in -at(jan), as in (3), and with non-verbal bases such as prepositions. Due to ambiguous formations, in which the Gothic verb in -inōn stemmed from a noun without this -in-, reinterpretation (Kluge 1926: §138) may have taken place. The sequence -in-, originally belonging to the verbal -inōn, was absorbed by the Gothic suffix -assus which generated -(i)nassus (4). Formal reanalysis and assimilation thus typified the history of the suffix and were still at work in the early Dutch period. The final consonant cluster of the base adjusted to the first consonant of the suffix in Middle Dutch gevancknisse (with /k/ instead of /n/ from the verb gevangen, ‘to capture’) or the elision of <de> in vonnis (‘judgement’, originally vondenis from the past simple of the verb vinden, ‘to find’).

Finally, all the described suffixes were subject to considerable phonological changes, of which several are undoubtedly connected with the semantic abstractness of the suffix, which motivated the primary stress on the base. Old Germanic *-nassu for instance lost its full /a/ sound and appeared in Old Saxon -nussi/-nissi/nessi with less prominent vowels (De Vries & De Tollenaere 2004). The final vowel /i/ was levelled to /ǝ/ (Van Loon 1987: 72) and was eventually omitted in German monosyllabic -nis and Old English -nis/-nys/-nes. Although the two variants -nisse and -nesse still occurred in Middle Dutch, 16^th century Dutch selected the eroded -nis as the standard form. Except for the loss of inflectional final /ǝ/, -heit, -heide or -hede did not erode to the same extent as -nis/-ness. The /ai/ sound in *haidu was contracted to /ei/ in central Westgermanic dialects, the ancestors of Dutch and German, and to /e/ in others (e.g. in Old Saxon). In Old Dutch, Westgermanic /ō/ diphtongized to /uo/ in -duom (Szulc 1987: 80) or developed into /ū/ in Northern dialects (compare noun doem or En. doom, Van Loon 1987: 62). In unstressed suffix position, this vowel has been shortened to /o/ in 15^th century Dutch. The suffix -scap has been preserved in almost its original form. Although -scap(e) followed a rather rigid inflectional system in Old and early Middle DutchGa naar voetnoot3, it lost its ending vowel. Due to palatalization and unrounding, Indo-European /a/ developed into Ingvaeonic /e/ and even /i/, leading to Old Frisian -skep/skip(i). In western and more coastal dialects of Dutch, -scip (Schönfeld 1970: 201) remained in use, but it was replaced in the 17^th century by the /a/ variant from the central dialects (Van Loey 1948: 13, Marynissen 1996: 251-254).

[pagina 92]

[p. 92]

3 The development of Dutch abstract suffixes

In this section, I will first discuss the empirical basis and the methodological assumptions of this study (3.1). I will then lay out my reconstruction of the diachronic paths of -nis (3.2), -heid (3.3), -dom (3.4) and -schap (3.5).

3.1 Data and methodology

My investigation into the development of Dutch abstract suffixes is largely based on datasets containing written language from the 12^th to the 20^th century, taken from the dbnl corpus, a freely accessible collection of Dutch literary texts on the website Digitale Bibliotheek voor de Nederlandse Letteren (sponsored by Stichting dbnl, Maatschappij der Nederlandse Letterkunde, nwo and Nederlandse Taalunie).

I divided the data into several subsets, each representing a time span of approximately 50 years. I classified 15,053 attestations (N) for -nis, 88,773 for -heid, 3,457 for -dom and 13,384 for -schip, which yielded a total of 10,403 different abstract derivations (V, e.g. ridderschap, maagschap, etc.). I analyzed these derivations in terms of their morphological pattern, that is, the category of their base, i.e. nominal (noun - adjective) or verbal (infinitive - participle). For each morphological pattern I identified the prototypical semantic frames. These semantic labels rely partly on contextual and semantic analysis of a random subset of 100 tokens of the pattern and partly on descriptions from the dictionaries Oudnederlands woordenboek (onw, covering Old Dutch), Middelnederlandsch Woordenboek (mnw, Middle Dutch), Woordenboek der Nederlandsche Taal (wnt New Dutch) and Van Dale Groot Woordenboek der Nederlandse Taal (present-day Dutch).

In this way, I combined the traditional qualitative approach of diachronic morphology with more extensive quantitative information about the occurrence of the basic morphological patterns in each period. This has allowed me, I believe, to increase the delicacy with which patterns and transitions could be observed. The extended empirical datasets revealed a number of hitherto overlooked forms, which required recognizing more ambiguity than had happened so far in the morphological analyses. Ambiguity is in fact a crucial starting point for reanalysis and extension of the suffixes over new schemes or patterns. Once such often incidentally formed new patterns show an increasing type frequency, there is ‘semantic coherence’ (Aronoff 1973 in Hüning 1999: 69) and the rule becomes fully productive (Trips 2009: 29).

For the quantification of the data and the interpretation of the quantified data, I adopted the following procedures. In all the figures and tables, the quantitative term type refers to a derivation of which several instances may exist, e.g. bescermenesse and bescermenisse are two tokens or attestations (N) of one type (V) (symbols used by Baayen 1990 and Al & Booij 1989). I considered the relative proportion of types for each morphological pattern as an indication of the extent to which a productive rule dominated a stage in the history of the Dutch suffix. High token frequency (N) of a certain type I assumed to point at a high degree of lexicalization or fossilization, whereas a relatively large amount of hapax legomena, i.e. derivations just once attested, I took to reflect the productivity of a suffix.

[pagina 93]

[p. 93]

3.2 The suffix -(t)(e)nis: from deverbal act nouns to non-productivity

In line with the development of the Gothic suffix -(i)nassus, Old Dutch -(e)nussi was employed to derive deverbal nouns, a derivational class which dominated the samples extracted from the dbnl corpus. In most cases, the suffix was attached to verb stems, as in 10^th century giruornussi and irbarnussi from giruor-on (‘to move’ with infinitive suffix -on) and irbar-on (‘to reveal’), or farhugnissi and biriu(w)nissi from *farhug-on (‘to scorn’), and biriuw-en (‘to regret’). Apart from these apparently regular derivations from verb stems, Old Dutch had some forms with an inserted binding phoneme /ǝ/ as in gihugenussi (from gihug-en, ‘to remember, to think of something’) or irsuokenussi from irsuok-en (‘to investigate’). De Haas & Trommelen (1993: 245) hypothesize that -enis was the primary morpheme which attached to the verb stem. Whenever the ending consonant was a nasal (/n/) or liquid (/l/ or /r/), there was phonological harmony with the first /n/ of the suffix and the /ǝ/ was dropped, as in verwennisse (‘spoiling’), vangnesse (‘imprisonment’), verlancnisse (‘longing’), or verkulnisse (‘cooling’). At first sight, this hypothesis might appear to be confirmed by the diachronic data: /ǝ/ could be a levelled variant of the /i/ from the reinterpreted suffix -inassu, which might have triggered palatalisation in 13^th century gevencenisse (from gevangen, ‘imprisoned’), opverstentenisse (from opverstanden, ‘to rise’) and gedinkenisse (from gedenken, ‘to commemorate’) or 14^th century versmedenesse (from versmaden, ‘to scorn’). Most linguists agree, however, that the -i- did not survive in Old Dutch, as seen in the examples giruornussi and irbarnussi mentioned above. They argue that the palatalization of the stem vowels was caused by the /i/ sound in the second syllable of -enisse, the vowel in present-day -nis (Van Loon 1987:39). Even so, whether the underlying morpheme was -enis or -nis, in Old and Early Middle Dutch both forms seem to be roughly interchangeable (as in for instance gevangnisse and gevangenisse): they were completely synonymous and tended to be selected in a quasi random fashion. It is only in late Middle Dutch that the allomorph -enis secured its position in a stabilised system (De Haas & Trommelen 1993: 245) and that the phonological conditions became more rigid: stems ending in another sound than /n/, /l/ or /s/ almost always took -enis.

Importantly, due to the increase of the intervening phoneme /ǝ/, ambiguous uses came about. Since the Dutch infinitive morpheme -en has the same phonological realisation as /ǝ/ and nasal /n/ in the suffix -nis, it was not always clear whether the derivation contained the infinitive or the verbal stem, as in Early Middle Dutch nakenisse from naken (‘to approach’), scamenesse (from scamen, ‘to be ashamed’) and verstannesse (from the old verbal stem ik staen, ‘I stand’ or infinitive staan ‘to stand’, Schönfeld 1970: 178). Speakers may have assigned incorrect morphological structures to the derivations, based on formal analogy and semantic connections between verbal stems and infinitives (i.e. ‘analogical reinterpretation’, Van Bree 1996: 114 and ‘restructuring’, Booij & van Santen 1998: 280). This reinterpretation is often referred to as reanalysis, the ‘change by which a complex word comes to be regarded as matching a different wordschema from the one it was originally created by’ (Haspelmath 2002: 56). It is the first step in a process of analogical extension (Van Bree 1996: 104) or affix generalisation (Booij & van Santen 1998: 75). The derivational domain of -nis, originally attached to verb stems, was extended

[pagina 94]

[p. 94]

to other word classes. In this way, semantic and formal ambiguity between infinitive and stem in derivations with -nussi led quite early to the emergence of a new productive pattern, as found in Old Dutch ratonnussi and irfanknussi from raton (‘to prickle’) and irfān (‘to accuse’), both with the Old Dutch infinitive suffix -on or -an. Likewise, some verbs may have been reinterpreted as participles. Although most verb stems were monosyllabic (De Haas & Trommelen 1993: 245), a large amount of them (24.03% of the observations in my 12^th and 13^th century datasets) had been prefixed with Old Dutch far-, ir- and especially gi-, which correspond to ver- (Baayen 1990: 226), er- and ge- in present-day Dutch. These reinforced the perfective meaning of a verb, as in gebruken (‘to use’, cf. bruken, Got. brukjan) or getugen (‘to testimony’, cf. tugen, Got. tiuhan). This perfective prefix was optional until Late Middle Dutch, when its use became limited to and analogically spread over the grammatical class of participlesGa naar voetnoot4. This is why we encounter a lot of ambiguous derivations in Middle Dutch, such as 12^th century ghevanghenisse (from the verb or participle gevangen, ‘to catch, caught’, Van Loey 1948: 77) and gesceppenesse or ghescepenesse (from infinitive sceppen, ‘to create’ or participle gescepen or gesceppen, Van Loey 1948: 76). Figure 2 visualizes the ambiguity in the morphological composition of gevangenis. With the ge- prefix becoming obligatory in participles and becoming more or less exclusively used for this category, most of these derivations could be interpreted as containing past participles instead of infinitives. I propose that at this point a new deverbal morphological rule originated. Eventually, suffix -nis came to be attached to past participles in which this ge- was lost or absent, for instance in Early Middle Dutch von(de)nisse (‘verdict’) (Schönfeld 1970: 212), which is based on the strong verb vinden with vowel change to /o/ in past participle vond. Similar examples are gheboernesse (‘being born’) with /ō/, derived from participle gheboren (‘born’) and verb beren (‘to bear’), Late Middle Dutch ontbondenisse from ontbinden (Van Loey 1948: 69), as in (5), and bedrogenisse, which existed alongside bedriegenisse.

die grote dissolucie ofte ontbonde-nisse
the great dissolucie or dissolved-participle-ness-suffix
of [van] smeltinghen des lichamen
or [of] melting-genitive of body-genitive
‘the great dissolucie or dissolution or [of] the melting of the body’
(c. 1462, anonymous, Reis van Jan van Mandeville)

The deverbal suffix -nis, with verbal stem, infinitive or participle, proved one of the most productive suffixes up until 16th century Dutch and, as shown in (5), formed act nouns. Bedervenisse, behoudenesse, lavenesse or regierenissen referred to the act of bederven (‘to spoil’), behouden (‘to maintain’), laven (‘to slake’) and regeren (‘to rule’) respectively, or at least to one delineated instantiation of this act (‘instantial’ act nouns, Hüning 1999: 176). Nowadays, Dutch speakers would translate such act nouns by words ending in -ing, as in bescherm-ing (Eng. protection, Middle Dutch bescherm-enisse), (be)dreig-ing (Eng. a threat, Middle Dutch

[pagina 95]

[p. 95]

Figure 2 Overview of proposed morphological analyses with deverbal derivations in -nis

dreig-enesse) or vordering (Eng. improve-ment, Middle Dutch vorder-nisse).

The productivity of the deverbal suffix with past participle base may, according to my observations, have favoured the emergence of a deadjectival pattern. The first deadjectival derivations, thiusternussi (from thiuster ‘dark’) and īdilnussi (from īdil, ‘vain’, cf. idle), are attested in my data in the 10^th century. In 11^th and 12^th century Old Dutch, we find finstarnussi, thimsternussi and wuostnussi, derived from the adjectives finstar and thimster (‘dark’) and wuost (‘ferocious’). Nevertheless, deverbal derivations still dominate the data and except for Middle Dutch duysternisse, deadjectival derivations remained infrequent and marginal. In the 14^th and 15^th century, however, the deadjectival suffix -nisse produced several neologisms, such as donckernisseGa naar voetnoot5 (‘darkness’, cf. deemsternisse), swaernisse (from zwaar, ‘heavy’) and even soetenisse (from soet, ‘sweet’). These abstract nouns describe a characteristic and are quality nouns: they refer to the quality of ‘being dark/heavy/sweet’ or ‘being filthy’ in the case of vuilnisse (from vuil, ‘filthy’), and ‘being cool ’ in coelnisse (from koel, ‘cool’). I would like to point out the close link here between the adjectival pattern and the one with verbal origins, more specifically the participle. The characteristic expressed by the base is often a temporary state caused by an act, as in for instance moetnesse / vermoeienis (‘the quality of being tired’, ‘tiredness’), bedroefenesse (‘the state of being dejected’, ‘sadness’ from adjective bedroefd or verb bedroeven), schendenesse (‘the state of being disgraced’) or hechtenis (‘being imprisoned’). Probably, the patterns with participle functioned as an intermediate stage for the development of deadjectival derivations, as both participles and adjectives can be used as predicates. This formal reanalysis also triggered semantic reinterpretation (as defined by Van Bree 1987: 167) in a couple of cases: in (6), verradenisse with verbal stem verraden (literally ‘the act of betraying’) is used to refer to the prototypical quality of people who betray, namely ‘the quality of being false or disloyal’, or ‘treacherousness’.

[pagina 96]

[p. 96]

Nevertheless, many of the deadjectival forms disappeared quite early (from 9.47% in the late 16^th century to no attestations in the 17^th century) and the ones surviving acquired a more concrete locative meaning, as in wildernisse, heidenesse (‘land of the heathen’, from heiden, ‘heathen’), woestenis (‘a desert’), or duisternis, (‘darkness’, ‘dark place’). It may be under the impulse of such locative meanings that gevangenis, originally expressing the quality of ‘being captured’, developed its concrete semantics of ‘prison’.

As can be seen in Table 1, the deverbal pattern still dominated 17^th century -nis: it accounts for around 50% of all the observed types. However, a steady increase in denominal formations can be observed as well (from no attestations of exclusively denominal derivations in the 13^th century to 2.99%, 7.59%, 10.53% in the 15^th and 16^th century up to 16.67% in the early 18^th century). De Haas & Trommelen (1993: 245) state that -nis basically attached to Germanic verbs and adjectives. Some Middle Dutch derivations however could be analysed as derived from either a verb or from a noun, as in the dual-labeled column in Table 1. The two analyses are, for instance, possible with scamenisse, ghebrukenesse, lettenisse, ghedenckenisse, verraetnesse and verderfenisse with recognisable verbs scamen, ghebruken, letten, ghedencken, verraden and verderven or derived abstract nouns scame (‘shame’), gebruuc (‘use’), lette (‘obstruction’), gedenken (‘remembrance’), verraet (‘treason’) and verderfe (‘destruction’). By the beginning of the 16^th century we encounter some derivations that are no longer ambiguous and necessarily require an analysis on the basis of a noun, such as spysenesse (from spyse, ‘food’) and vergiffenis (‘forgiveness’, from obsolete vergif) as opposed to

[pagina 97]

[p. 97]

clearly deverbal vergevenis, which was common in Middle Dutch.

The rise of nominal bases with incorporated abstract suffix -te in the 16^th and 17^th century is essential in this development. A /t/ sound was already present in the early stages of the language, as the ending consonant of a verbal stem was easily devoicedGa naar voetnoot6 in Middle Dutch, yielding for instance 12^th century verbintenis from verbinden (‘to connect’), beeltenis (from beelden, ‘to depict’) or ontsteltenis (from participle ontsteld, ‘dismayed’). From then on, -nis could be attached to nominalizations on the abstract suffix -te: 13^th century gelofnisse, ghedenkenis or gheboernisse were replaced by late 16^th century geloftenisse, gedachtenis (‘thought’) and geboortenisse, derived from gelofte (verb geloven and -te), gedachte (‘thought’, derived from participle of denken ‘to think’) and geboorte (from participle geboren). Reanalysis took the form of affix telescoping (Haspelmath 2002: 56 and 1994: 3) here, that is, the process in which two morphemes, namely nominalising -te and -nis, fuse into one, yielding a third allomorph -tenis (De Haas & Trommelen 1993:245). In the case of 16^th century bekentenesse (‘confession’) and erkentenesse (‘acknowledgement’), both derived from the verb (be/er)kennen (‘know’) - a nominal base with -te (*bekente) never existed. This shows that /t/ was inextricably tied to the suffix, not to the base. Table 2 visualises how bekennisse/herkennis were replaced by bekentenis/erkentenis in the period 1200-1600.

Not all derivations were affected by this new productive rule: the mechanism was for instance blocked (in the sense of Rainer 1988, Haspelmath 2002: 249 and Booij & van Santen 1998: 69) in the highly frequent lexicalized form kennis (‘knowledge’, cf. German Kenntnis). The question arises then whether -nis still contributed to the meaning of the derivation, since suffix -te also formed abstract

[pagina 98]

[p. 98]

nouns. Beloftenis and geboortenis, for instance, have disappeared in present-day Dutch in favour of synonymous geboorte (‘birth’) and belofte (‘promise’).

In my view, the allomorph -tenis does not function as a phonologically conditioned variant of -nis, but its systematic emergence suggests a semantic development. It may have been a reaction to the overall lexicalizationGa naar voetnoot7 of derivations in -nis in the late 16^th and the beginning of the 17^th century and the concrete meanings that many of these highly frequent lexicalized forms came to convey in a rather unsystematic manner. Rather paradoxically, it is the abstractness of the suffix and its increasing availability that caused the word formation rules to become obscure (or ‘opaque’ and therefore difficult to apply, see Van Bree 1996: 165) and, hence, the outcome of suffixation to become more susceptible to semantic specialisation into concrete meaningsGa naar voetnoot8. My data showed that many act nouns came to refer to a person or thing that causes the action (‘agentive nouns’, Kronenberger 2002: 202), such as een hindernis (‘an obstacle’) or een stoornis (‘a disturbance’). Others referred to the result of the act (Hüning 1999: 184), for instance gesceppenesse/gestaltenisse (example 7, ‘a shape’ that has been created, from scheppen, ‘to create’). The distinction between cause and result is not always very clear: the nouns verbintenisse (‘an agreement’), getuigenis (‘a testimony’), vonnis (‘a conviction, judgement’) may refer to a legal document that causes the action of verbinden (‘to commit to’) or getuigen (‘to testimony’) as well as to the result of this verbal act, written down on paper. The rise of concrete secondary meanings eventually led to a new productive pattern with -nis that I observed in the 18^th and 19^th century, yielding derivations such as medichinesse (‘a medicine’, present-day Dutch medicijn) and gebeurtenis (‘an event’, from the verb gebeuren, ‘to happen’) which never had an abstract meaning. In these nouns, as well as in beloftenesse and ghebortenesse, the suffix -nis did not add much meaning to the derivation anymore: semantic bleaching had come to a final zero-point and, as shown by present-day belofte and geboorte with abstract -te and omitted -nis, even resulted in a null mor-

[pagina 99]

[p. 99]

pheme or zero-form (Givón 1979: 209). By this stage, -nis had taken up so many morphological patterns and meanings that -te may have been added to strengthen one subtype, namely the deverbal nominalising pattern with abstract meaning.

Similarly, some derivations developed a passive concrete meaning. Erfenesse (from erven, ‘to inherit’), for example, came to name the ‘affected object’ (Hüning 1999: 186) of the action, the thing that you can inherit, or ‘the inheritance’. The affected object meaning came, again, to be encoded by the suffix. Muizenis (8), for instance, was an 18^th century derivation that did not refer to the act of contemplating (from the verb muizen (‘muse’)), but only to specific worries that people have on their mind. Even deadjectival derivations were subject to this emergence of concrete semantics: vuilnis lost its original meaning of ‘the state of being filthy/rotten’ in favour of the present-day concrete semantics of ‘rubbish’.

[pagina 100]

[p. 100]

3.3 The suffix -(ig)heid: deadjectival quality nouns

Unlike deverbal -nis, the Dutch suffixes -heid, -dom and -schap originated in noun-noun compounds in Gothic and were initially attached to nominal bases in the broad traditional sense of nominal or adjectival (with the distinction between both being often hard to make). Old Dutch -heit seems to occur mainly with adjectival bases such as argheit, wankilheit, bitterheit, heiderheit, skōnheit, wīsheit or slahtheit, derived from adjectives arg (‘bad’), wankil (‘wobbly’), bitter (‘bitter’), heider (‘bright’), skōn (‘beautiful’), wīs (‘wise’) or slaht (‘slow’). Old Dutch had some denominal formations such as gotheit (from got, ‘god’), manaheit (from mana, ‘man’) and kristīnheit (from kristīn, ‘christian’). However, they shared many semantic similarities with the deadjectival quality nouns which they were - increasingly - outnumbered by in the earliest stages of the Dutch language (from 5.00% and 3.00% of exclusively denominal derivations in the 13^th-14^th century to around 0.50% and 0.40% in the 17^th and 18^th century). Most denominal derivations which originally denoted a specific rank could also refer to a quality that prototypically comes with occupying a position. The definition for mensheit in the Middle Dutch dictionary (mnw), for instance, reads ‘to be a human’ as well as ‘humaneness’. Because of the close link between the rank ‘childhood’ and the quality ‘childishness’, /s/ was inserted in the derivation kindsheid, probably in

[pagina 101]

[p. 101]

analogy with the adjective kinds instead of the noun kind (cf. High German kindheid, Middle High Ger. kintheit and En. childhood). This pattern with adjectival base and quality noun semantics would become highly productive throughout the history of -heid.

The main difference between deadjectival quality nouns in -heid and ones in -nis lies in the number of polymorphemic or compositional bases (Wilmanns 1930: 385). Adjectives with suffixes -lijk (‘-ly’), -zaam or -ig frequently occurred with -heit from Old Dutch onwards, as in for instance guotlīkheit, stādigheit, brōthigheit, wirthigheit from guotlīk (‘godly’), stādig (‘steady’), brōthig (vgl. brōthi, ‘weak’) and wirthig (‘worthy’). Wilmanns (1930: 385) suggests that these formations may have counterbalanced the loss in productivity of Middle Dutch abstract suffixes -(e)de and -te (derived from ī, cf. Got. -iþa/iðaGa naar voetnoot9). There may be no evidence of a causal connection between the disappearance of nominalising -(d)e (Schönfeld 1970: 201) and the emergence of the polymorphemic bases with synonymous -heid but the high commutability and the almost unrestricted flexibility of -heid is certainly one of the reasons why the suffix gained ground so easily within the paradigm of abstract suffixesGa naar voetnoot10. At the end of the 16^th century, 72.44% of the deadjectival derivations in -heid in my data involved compositional forms with -zaam, -baar, -ig or -achtig, compared to 41.18% in the 12^th century, as shown in Figure 3. The morphological constraints on the base were so few that a

Figure 3 Compositionality of deadjectival bases with -heid (type frequency)

[pagina 102]

[p. 102]

variety of synonymous derivations emerged, for instance reynheit, reynelijckhede and reynicheit (‘cleanness’). The deadjectival forms sometimes competed with the originally denominal derivations, such as goddelijkheid, menselijkheid and jeugdelijkheid which gradually pushed out quality nouns godheid, mensheid and jonkheid (resp. ‘godness’, ‘humanness’ and ‘youngness’).

Within this diversity of complex adjectival bases, the forms with -ig became very frequent and finally fused with the abstract suffix into the allomorph -igheid. In Late Middle Dutch, most lexicalized old adjectives in -ig were already morphologically opaque: the nouns or adjectives on which heil-ech (‘holy’), ghier-ech (‘miserly’) or guls-ich (‘gluttonous’) were based, disappeared or occurred both with and without -ig, as in soetecheit (‘sweetness’), lichtecheiden (‘lightness’) vs. soetheit, lichtheit (see De Haas & Trommelen on -ig 1993: 248). The transparency of the morphological structure was lost and in the 14^th and 15^th century, a few derivations emerged in which -ig did not belong to the adjectival base, but merely functioned as an intermediate, transitional syllable, as in 14th century geborenicheide (‘birth’) or meineedecheit (‘perjury’) (but * geborenich or *meineedec), 15^th century gesondicheit, and snelligheydt from gesond (‘healthy’) and snel (‘fast’) and 16^th century rijpicheyt (‘ripeness’) and strengicheyt (‘strictness’). Both morphemes -ig and -heid were so frequently combined that they eventually were reinterpreted as one fused or ‘telescoped’ suffix (compare -te and -nis in -tenis). Examples proving the claimed morpheme status of -igheid are stacked forms vruchtbaricheit (‘fertility’), eerlozicheyd (‘honourlessness’) or ghehoorsamichede (‘disobedience’). As -zaam or -baar changed the part of speech of the base into an adjective and -ig did not semantically contribute to the derivation anymore, the only reason why it was still inserted, was because it had become formally part of the suffix -(ig)heid. By the 15^th century, the allomorph -igheid even occurred in derivations which contained nouns (rebel in rebellicheit, ‘rebellion’), verbs (falen in falicheyt, ‘failure’) or noun phrases (duistere nacht, ‘dark night’, in duusternachticheyt). In German, this telescoped affix consisting of -ec and -heit further fused into the assimilated forms -igkeit and even -keit (cf. Wilmanns 1930: 385).Ga naar voetnoot11

Whether speakers opted for the use of -heid or -igheid increasingly depended on semantic grounds. In Middle Dutch, most derivations with -heid and -igheid co-existed and were more or less semantically equivalent: the compilers of the Middle Dutch Dictionary (mnw) cross-reference vetticheit, magericheit and sachticheit and their variants without -ig, namely vetheit, magerheit, sachtheit, i.e. ‘fatness’, ‘slimness’, ‘softness’. With the increase of the -igheid allomorph in the 16^th and 17^th century, I observed some semantic differentiation between the two allomorphs: derivations with Dutch -igheid acquired concrete meanings more easily (see Wilmanns 1930: 388 on the German equivalent -keit), whereas words on -heit remained their abstract meaning. The 16^th century derivation groenicheit, for instance, could still refer to ‘the quality of being green’ (cf. groenheit) but also

[pagina 103]

[p. 103]

Figure 4 Overview of morphological analyses of deadjectival adjectives in -(ig)heid

to ‘something green’. By contrast, 20^th century vuiligheid (9) or gauwigheid were only used as concrete nouns (‘something filthy or vuil’ and ‘something you do quickly or gauw’). A similar semantic split is manifested in the difference between present-day Dutch fraaiheid (‘prettiness’) and fraaiigheid (‘something pretty’), kleinheid (‘smallness’) and kleinigheid (‘something small’), zoetheid (‘sweetness’) and zoetigheid (‘a sweet’). In other words, the allomorph -igheid evolved into a fully operational morpheme producing concrete nouns.

The participle pattern with -heid was less common than the purely deadjectival structure, but its number rose steadily from early Middle Dutch onwards (from 5.58% past participles in the 13^th century to around 11% in the 18^th and 19^th century in my data). The past participle referred to a temporary state initiated by an action, such as bedectheit and ghemintheit from the weak verbs bedecken (‘to cover’) and minnen (‘to love’, notice the participle marking ghe-V-t), verborgenheit and ghenomenheit from the strong verbs verbergen (‘to hide’) and nemen (‘to take’) of which the participles were formed by vowel change in the stem (Ablaut).

However, the development of this participle derivation was different from that with suffix -nis in that -heid could also be attached to present participles, e.g. mogentheit (‘being almighty’ from mogend, ‘being able to’) or leventheit (‘being alive’ from levend, ‘living’).

The growth of the participle patterns with -heid was a prerequisite for the extension of the derivational domain to purely verbal bases, as in early Middle Dutch veranderheyt (‘change’, cf. verandernisse), komenheit (‘coming’, cf. comenisse in the mnw) or kunheit (‘skill’). As a first step in this process, derivations with a /t/ from the weak past participle form or the present participle (in -end/-int) became

[pagina 104]

[p. 104]

opaque. This /t/ was sometimes reinterpreted as a mere binding phoneme and could, perhaps as a result of hypercorrection, be inserted in 16^th and 17^th century derivations with past participle bases which did not contain this morpheme (such as gheschapentheyt from scheppen/geschapen, ‘to create/created’, verborghentheyt from verbergen/verborgen, ‘to hide/hidden’), verdorventheyt from verderven/verdorven (‘to debase/debased’), as in (10).

By the end of the 18^th century, this binding phoneme was systematically dropped again, leading to an increase of possible infinitive bases (in -en), such as 19^th century kwaadsprekenheid (‘slander’), oplettenheid (‘advertency’) or haatdragenheid (‘vengefulness’). Table 5 shows the rise in 1500-1700, and subsequent drop, in frequency of the forms with binding phoneme /t/.

[pagina 105]

[p. 105]

The high productivity of -heid around the 17^th century was apparent in the range of minor word classes it attached to, represented under ‘other’ in Table 4. Already in Middle Dutch, -heid could be combined with atypical bases such as ander (‘other’), meerder (‘more’), preposition over (‘over, above’) in abstract nouns anderheit (‘the quality of being different’), meerderheid (‘the quality of being superior’), or overheit (‘the quality of being mighty’, now lexicalized into ‘the government’). These nouns expressed a non-gradable relation between two entities, which is rare in abstract noun formation (e.g. van Santen 1992: 184). The quality noun semantics remained dominant and were imposed on roots which at first sight could not be associated with a quality meaning.

(11) Is anderheit ghesturven doot,
is other-noun/adverb-ness-suffix died dead,
‘when otherness has died, then [...]’
(c. 13^th century, anonymous, Een subtile ghedicht van hogher godliker mynnen, r. 15)

From the 17^th century onwards, many of the derivations lexicalized and acquired contextually determined concrete meanings. Christenheid/kerstijnheit or edelheit (originally ‘the quality of being a Christian or being noble’) were sometimes

[pagina 106]

[p. 106]

used to refer to a group of Christians or noble men. Meerderheid and minderheid (originally abstract ‘to be superior/inferior’) nowadays exclusively denote ‘a majority/minority’. Some derivations assumed a locative meaning, such as gelegen(t) heit (‘a well-situated place’, Kronenberger 2002: 204) and 16^th century uses of openheid (originally ‘the quality of being open’) referring to an open space or an opening.

Apart from these incidental collective or locative meanings, most lexicalized derivations came to denote a specific person or object which typically displays the quality expressed by the adjectival base. Outheit, which is currently only used to refer to a period, once had the sense of ‘an object from an old period’ or ‘relic’ (compare antiquiteit, ‘an antique’). Other examples of semantic specialisation are the use of een schoonheid (‘a beauty’) to refer to someone beautiful (cf. French une beauté) or heiligheid for a holy person. The diachronic development of the semantic derivation patterns is represented in Table 6.

3.4 The suffix -dom: from quality nouns to collective nouns with allomorph -endom

As a continuation of its original noun-noun compound structure, the suffix -dom was at first mainly attached to nominal bases in the broad sense of either noun or adjective. Original -dom as in Old Dutch biskopduom, 12^th century hertochdom or 13^th century maghedom was used to refer to the abstract notion of ‘status’ (Trips 2009: 82), ‘the position or rank of a bishop (biskop), a duke (hertoch) or a virgin (maghet)’. The denominal pattern, of which 14^th and 15^th century scependoem (from schepen, ‘alderman’), keefdoem (from keef/kevese, ‘concubine’) and keyserdom (from keizer, ‘emperor’) are some more examples, still persists nowadays as the default pattern of the suffix. Note that the two semantically re-

[pagina 107]

[p. 107]

lated meanings of the derivations in -dom, viz. ‘social status’ (e.g. maghedom) and ‘profession’ (e.g. scependoem), correspond with two different semantic types of nominal bases, viz. nouns referring to a person (e.g. maghet) and nouns referring to a rank (e.g. biskop). Table 7 shows that in its earliest use, -dom combines with both nominal bases.

Adjectival derivations were also already attested in Old Dutch, for instance in siekduom (‘sickness’ from siek, ‘sick’) and wÐsduom (‘wisdom’ from wijs, ‘wise’). Middle Dutch words such as rijc in rijckdom (‘richness’) and wise, edel, have both adjective and noun status (‘the wise, the noble’, etc.), as noted by Koelmans (1979: 43) and Hüning (1999: 112): they all refer to a quality as well as a state or rank. This might also explain why currently obsolete heildom (from noun heil, ‘blessing’) existed alongside the formally deadjectival deriviation heilichdom (from heilig, ‘holy’) in Middle Dutch.

Except for the 13^th century derivation wasdoem (from wassen, ‘to grow’), the suffix -dom has never really taken verbal stems. Koelmans et al. (1979: 43) therefore suggest that wasdoem (‘growth’) may have been derived from the noun was (‘growth’). The development of act noun meanings in some formations with nonverbal roots and -dom shows that the suffix is already stretching its semantic domain. In (13), for instance, hoerdom (from the noun hoer, ‘whore’) does not refer to the state or position the girl is in, but rather to the more abstract and often frequentative behaviour of promiscuity or adultery. However, this deverbal act noun pattern emerges only occasionally.

(13) hi seide hare tonrechte hoerdom toe
hi said her wrongly whore-noun-dom-suffix towards
‘he wrongly accused her of adultery’
(c. 1393-1402, Philip Utenbroecke, Spiegel historiael, ii.I, r. 12)

By the 16^th and 17^th century, the suffix -dom was often innovatively and creatively used. This resulted in the temporary breakthrough of adjectival derivations, such as iongedom, leechdom, vrydom, which competed with joncheit, leechheit, vryheit (‘youngness’, ‘emptiness’, ‘freedom’). However, it was mostly nominal derivations which profited fully from this increased productivity and consequently extended their use (from around 40 to 50% in the 15^th century to around 67% in the late 18^th century). In the early 17^th century pausdom (‘popedom’), priesterdom (‘priesthood’), prinsdom (‘princedom’) and ketterdom (‘hereticness’) emerged. As also noted by Koelmans et al. (1979:37), a fair amount of these neologisms were created by Joost Van den Vondel (so-called ‘poetic licence’, Haspelmath 2002: 101), such as zeedom (‘seamen’), besnedendom (‘Judaism’, from besneden, ‘circumcised’) and ridderdom (‘knighthood’).

One of the most remarkable tendencies, however, was the sudden use of plural forms in nominal stems (see Table 8), as in late 16^th century vorstendom (‘the kings’) or 17^th century jodendom (‘the Jews’) and godendom (‘the gods’). Apart from jufferdom (from juffer, ‘nurse’), zusterdom (from zuster, ‘sister’), koningdom (from koning, ‘king’), meesterdom (from meester, ‘master’) or rederijkerdom (from rederijker, ‘rhetorician’), most neologisms in the 18^th and 19^th century data consist of plural base forms: examples are protestantendom (~ ‘protestants’), comediantendom (~ ‘comedians’), germanedom (~ ‘Teutons’), republikeinendom (~

[pagina 108]

[p. 108]

‘republicans’), or vrouwedom (~ ‘women’), all with -en, the plural morpheme in Dutch. Hence, a new productive rule has to be posited. Even the existing Middle Dutch derivations engeldom (~ ‘angels’) or dichterdom (~ ‘poets’) as well as hoerdom were subject to this rule, yielding 18^th century englendom, dichtrendom and hoerendom. The insertion of e(n) may have been caused by reanalysis of deadjectival derivations ending in -en, such as heidendom or eygendom or derivations based on participles, such as besnedendom or geschapendom. The development can also be linked to the compound origins of the derivational pattern. Some Old-Germanic compounds required a genitive form for the first incorporated element, as can be seen in Middle High German vürstentuom (‘princedom’ but Old High German furisttuom), herizogentuom (‘dukedom’) and witewentuom (‘widowhood’), examples taken from Wilmanns (1930: 393). Given the intense contact between German and Dutch in the 17^th century, these genitive forms may have been reinterpreted as plural morphemes.

As observed by Koelmans (1979: 38), this rise of plural forms is problaby only the formal reflex of a semantic evolution. Table 7 shows an increase in nominal bases referring to persons. However, derivations such as griekendom, vrouwedom, patriottendom or mannendom did no longer refer to the state or quality of the Greeks, women, patriots or men, but designated the group as such. In the 16^th and 17^th century, many abstract quality nouns already took on a collective inter-

[pagina 109]

[p. 109]

pretation in certain contexts, such as rijkdom (‘the rich’), edeldom (‘the noble’) and wijsdom (‘the wise’). This collective meaning became fully productive in the 18^th and 19^th century. Most collective nouns in -dom were derived from a stem referring to people, to animals, as in apendom (‘apes’), beestendom (‘beasts’), wolvendom (‘wolves’) and bavianendom (‘baboons’), or, as Koelmans (1979: 39) puts it, to creatures with human features, such as schepsel (‘creature’) in schepslendom and geest (‘ghost’) in geestendom. The collective meaning then came to be coded by the suffix itself, as in the 18^th century formation mensdom (cf. mensheid in present-day Dutch), which refers to ‘all humans’ and no longer to ‘humanness’. In some cases, a secondary meaning developed from the collective semantics, namely the view, mentality or theory adhered to by a group of people, as in Christendom or jodendom. This appears to have become a productive pattern in the 20^th century, as illustrated by Hitlerdom and wertherdom (‘mentality of selfdestruction’, referring to the character of Die Leiden des jungen Werther by Goethe).

Derivations with -dom survived in 20^th century Standard Dutch only in lexicalized formations such as rijkdom (‘wealth’), ouderdom (‘age’) or jodendom. Hertogdom, bis(schop)dom and vorstendom are primarily used in their locative senseGa naar voetnoot12 nowadays, i.e. an area which belongs to a duke (hertog), bishop (bisschop) or lord (vorst). Some derivations acquired other concrete meanings, referring to objects such as eigendom (‘property’, which is personal or eigen to someone) or heiligdom (‘sanctuary, temple’). The evolution of the semantic derivation patterns is represented in Table 9.

[pagina 110]

[p. 110]

Neologisms that still emerge in present-day Dutch, despite the low productivity of the suffix -dom, are mainly used in ironic or negative contexts. The infrequency and archaic nature of the suffix (Koelmans et al. 1979: 44), confer a humorous effect on its use in informal contexts: the combination of the formal suffix and a depreciative stem, such as idioot (‘idiot’) or paap (‘popish person’) yields parodic formations, such as ploertendom (from ploert, ‘cad’), keezedom (from kaas/kees, ‘cheese’, referring to the Dutch), papedom or even Afrikaans idiotedom. The negative connotation of the base has an impact on the suffix too: derivations such as regentendom (14) or protestantendom inherit a disapproving value even though the stem is neutral.

(14) Toen zijn de studenten in opstand gekomen tegen
then are the students in resistance came against
het regent-en-dom.
the teacher-noun-s-PLURAL-dom-suffix
‘And then the students rose against the teachers’
(1969, F. Auwera, Schrijven of schieten interviews, Harry Mulisch, p.98)

3.5 The suffix -schap: denominal quality nouns referring to ranks

Although do(e)m and haidus were also found in compounds before the suffixes had emerged, it is schap which shows this compound origin most explicitly in the Old Dutch data of the dbnl corpus. The earliest attestation watarskap (‘water source’, or literally ‘place where water is being created’) is sometimes analysed as a compound (see onw) and sometimes as a derivation (see wnt). Similarly, bodaskap refers to the ‘messenger creation’ or the message that the messenger (boda) ‘creates’/transfers. The dominance of the compound use in Old Dutch may even entail that schap did not occur as a suffix yet in this period. The first clear suffix

[pagina 111]

[p. 111]

function of - schap is attested in 12th century deadjectival heithinskap, with a semantically abstract second element: heithinskap refers to the quality of ‘being a heathen’ (compare heathenism).

In my Early Middle Dutch data, -schap is mainly attested with nouns referring to people. It tends to differ from -dom in that nominal bases with -schap more often refer to a specific occupation or profession instead of a social status or family relationship (e.g. maghe or hoer in maghedom, hoerdom). This pattern was extremely productive throughout the history of the suffix and has proven to be fairly stable, as 12^th and 13^th century derivations meysterscap (from meyster, ‘scholar’), ridderscap (from ridder, ‘knight’) or coepmanscap (from coepman ‘merchant’) and 14^th century neologisms such as capiteinscap (kapitein, ‘captain’) or governoerscap (‘governor’) are still fully transparent today. The suffix seems to compete with -dom in derivations such as broederscap (‘brotherhood’, compare zusterdom, ‘sisterhood’), juedscap (from jood, ‘Jew’, ~ jodendom) and maechscap (from mage, ‘a relative’, ~ magedoem). The distributional distinction between profession and social status is, however, not rigid: cnaepscap (15), which is derived from cnape, may refer to the state of being a knight or servant, as well as to the social status of a young unmarried boy. The close link between family and social status is also present in broederschap (‘brotherhood’) which may relate to the family relationship as well as to the ‘(spiritual) solidarity’ or in vaderschap (cf. fatherhood), which may also refer metaphorically to the typical qualities of a person who ‘functionally and relationally acts as a father’ (Geeraerts & Moerdijk 1983: 93). These abstract nouns often name an interhuman relationship, from family relationships (zusterschap, ‘sisterhood’ or recent ouderschap, ‘parenthood’) to social relationships (vriendschap, ‘friendship’ or Middle Dutch maatschap, ‘companionship’).

Similar to the development of -dom, the 12^th and 13^th century nominal bases with an equivalent adjectival use were again the ones making further extension of -schap to the adjectival derivations possible, viz. vroetscap (16), nutscap and vrientscap which could be based on the adjectives vroet (‘wise’), nut(te) (‘useful’) and vrient (‘friendly’) or on the homonymous nouns yielded by zeroderivation. This led to 13^th century deadjectival quality nouns ghemeenscap (‘fellowship’, from adjective ghemeen) and bliscap (‘gladness’, from adjective bli, without the nominal ending in -d, cf. blidscap), which referred to a specific human characteristic, or sometimes to a temporary state with past participle bases (Middle Dutch dronkenschap, ‘being drunk’ or 19^th century gevangenschap, ‘being captivated’). The increasing productivity of the adjectival pattern in Middle Dutch is illustrated by neologisms familyaerschap, ijpocrijtscap or jaloirscap, based on familyaer (‘presumptuous’), ijpocrijt (‘hypocriet’) and jaloers (‘jealous’, derived from French ja-

[pagina 112]

[p. 112]

lous). Nevertheless, this expansion was restricted to Middle Dutch: no new deadjectival derivations were generated in 16^th century Dutch.

I propose that deadjectival -schap influenced the semantics of the denominal pattern, as even some denominal derivations came to refer to a quality instead of a rank or status. In these cases, the prototypical characteristic or state of a person became the focus: Middle Dutch gezelschap is probably most accurately translated by ‘fidelity’ or ‘kindness’, which is to be expected from a gezel (‘companion’). Cnechtschap acquires the meaning ‘helpfulness’ and meesterscap (lit. ‘mastery’) refers to the quality ‘competence’. Other examples are vriendschap (‘friendliness’) or vijandschap (‘hostility’). Besides this semantic side path, the denominal pattern with profession semantics remained highly productive. In the 16th and 17th century neologisms apostelschap (lit. ‘apostle-ship’), roffiaenschap (from roffiaan, ‘matchmaker’ or ‘brothel keeper’), burgemeesterschap (‘mayoralty’), herderschap (‘pastorship’) or colonelschap (‘colonelship’) emerged.

The high productivity of the suffix in the late 16^th century manifested itself in the emergence of deverbal derivations (from no exclusively deverbal attestations in the 12^th and 13^th century to 10.81% and 13.21% in the early and late 16^th century). Some 14^th and 15^th century deadjectival as well as denominal constructions had indirect verbal origins, such as dronkenscap (from participle drunken), ghereetscap (from gereiden, ‘to prepare’), conscap (cf. conde, ‘skill’) or ghequelscap (noun derived from kwellen, ‘to torment’). This resulted in a deverbal productive pattern of which 16^th century abstract nouns vereenschap (from vereenen,

[pagina 113]

[p. 113]

‘to unite’), verraedschap (from ‘to betray’) or versnelschap (from ‘to accelerate’) are examples. Two 17^th century derivations, wetenschap and weddenschap were based on the infinitive form (weten, ‘to know’ and wedden, ‘to bet’) instead of the verbal stem. Deinfinitival (on)wetenschap eventually replaced Middle Dutch conste (‘skill’) and weddeschap is a synonym of now obsolete wedding. Apparently, these deverbal and deinfinitival derivations often assumed an act noun meaning. Zeggenschap originally referred to ‘the act of saying something’ and wetenschap has been attributed the definition of ‘the act of knowing’ in the wnt. The formation of act nouns with -schap conflicted with deverbal -ing, as in 17^th century rekeninge, weddingeGa naar voetnoot13, versnelling or kwelling, cf. ghequelscap.

Perhaps exactly because of this overlap with the domain of -ing, act nouns in -schap have known a very short productive period, which meant that most deverbal derivations became soon lexicalized and underwent semantic specialisation. Zeggenschap nowadays refers to ‘the right to decide’, rekenschap no longer refers to ‘the act of calculation’ but to the justification that comes with a certain calculation and nalatenschap names the objects, goods or ideas that have been transmitted by ancestors (‘legacy’, expressed by erfgoet before the 17^th century). A relatively common present-day deverbal derivation is the lexicalized beterschap (‘improvement, progress, recovery’). Beterschap originated in the 17^th century, based on a comparative form (beter, from goed, ‘good’) or the derived verb beteren (‘to improve’), and finally replaced the deverbal suffix -nis of obsolete beternisse. Table 11 specifies how beternis(se) was replaced by beterschap in the 18^th century.

[pagina 114]

[p. 114]

In the 18th and 19th century data, most neologisms were denominal and had profession semantics such as professorschap (‘professorship’), acteurschap (lit. ‘actorship’), schrijverschap (lit. ‘writership’) or 19th century beulschap (lit. ‘executionership’), boerschap (lit. ‘farmership’), bruidegommeschap (lit. ‘bridegroomship’) and diktatorschap (‘dictatorship’). The only exception to this rule is 19^th century zwangerschap (‘pregnancy’), which replaced synonymous zwangerheid.

In certain contexts, the derivations acquired concrete meanings. Although the collective meaning was less common than it was with -dom, -schap could also refer to a group of people, as in jongelingschap (‘youngsters’), clerckschap (‘the clerks’) and burgerschap (‘the citizens’). Present-day gezelschap has this meaning of both ‘the group companions’ (cf. Middle Dutch ghezinne or compagnie) and ‘an association’ (cf. versaminghe or German gesellschaft ‘society’). Locative concretisations of derivations in -schap could refer to an area or an institute. One example is graafschap (‘county’, e.g. ‘t Graefschap van Vlaenderen, ‘the County of Flanders’) which denotes the area over which the graaf (‘count’) has power (compare hertogdom). Recent locative derivations denote an institute, as in vennootschap (‘company’) or agentschap (‘agency’). Some derivations came to designate a thing or person exhibiting the prototypical feature or state expressed by the base noun, such as gereetschap (collective ‘tools’, cf. Middle Dutch tuyg), or een maatschap and een heerschap referring to a mate (maat) or gentleman (heer), in which the -schap suffix does not seem to contribute to the denotational meaning of the derivation. These lexicalisations did not always survive: the use of vriendschap (lit. ‘friendship’) in the sense of ‘a pleasure’ (for instance in een vriendschap doen, ‘to do a favour’), is nowadays obsolete.

In the 20^th century, the original denominal pattern with -schap has proven to be the most stable and productive, as shown by neologisms presidentschap, dandyschap, pionierschap (from ‘pioneer’) or voyeurschap, sometimes with ethnic nouns in duitscherschap (lit. ‘Germanship’), europeërschap (lit. ‘Europeanship’) and mandarijnenschap (from mandarijnen, ‘Chinese mandarins’). Even in informal present-day Dutch, it shows more productivity than -nis or -dom: the formation of vrouwschap (‘womanhood’) or deadjectival bezopenschap (lit. ‘sloshedness’) have been attested. Although the suffix does not necessarily take stems with negative connotation, some of these recent derivations hint at a slightly coloured meaning. The clash between -schap naming prestigious ranks and some ‘hierarchically low’ (Moerdijk & Geeraerts 1983: 528) bases, such as vuilnisman (‘garbage collector’), have led to ironic derivations (e.g. vuilnismanschap). This irony may

[pagina 115]

[p. 115]

be present in some recent formations, but most neologisms in my data have a neutral connotation, such as papaschap (lit. ‘daddyship’). Except for some occasional creative derivations, the negative connotation or ironic aura has not become part of the suffix itself.

Van Bree (1996: 161) has noted that in certain Northern dialects, -schap was reinterpreted as a free morpheme and could be extracted. In neologisms productschap and bedrijfschap, schap no longer functions as a suffix, but is instead reinterpreted as the second part of a compound: productschap refers to an organisation of enterprises which process the same material or ‘product’ (cf. Van Dale 1999). Schap has here a specific lexical meaning of its own, namely ‘organisation of enterprises’. This reanalysis is probably based on the analogy with vennootschap, which originally referred to the abstract notion of ‘being a vennoot, a partner’, but came to refer to a co-operation in economical affairs. This concrete meaning of an ‘organisation’ or ‘institute’ has become associated with the lexical morpheme schap, which is even able to stand on its own as a noun (17, not to be found in Van Dale 1999). The emergence of autonomous schap is the result of analogy, contextual reinterpretation and finally extraction or ‘debonding’ (Norde 2006), but does not involve a revival of the original semantics of *skap (‘creature’). Only one semantic aspect of the derivation is being extrapolated to the autonomous morpheme: een schap does not refer to ‘a rank’ nor ‘an area’, but to an ‘organisation’ or ‘institute’.

[pagina 116]

[p. 116]

4 Concluding discussion: rivalry within the paradigm of abstract suffixation

The suffixes -heid, -dom and -schap have often been discussed together because of their similar evolution, but in this investigation I hope to have shown that very similar semantic and morphological processes also marked the history of -nis. Noun-noun compounds with the autonomous nominal elements haidus, do(e) m and scap(e) formed the source construction of -heid, -dom en -schap whereas the suffix -nis probably originated in a deverbal derivational pattern. However, through bleaching of their original lexical meaning and continuing processes of analogy and reinterpretation, all of the suffixes diverged from their original structure and came to exploit morphologically different bases or new meanings, often causing overlap with each other. Formal ambiguity between word classes, as well as metaphorical or metonymic relations, nourished these shifts in derivational domain, for example from denominal kindheid (‘social position of being a child’) to deadjectival kindsheid (‘behaviour of childishness’) or from hindernis as ‘the act/process of obstructing’ to ‘an obstruction, the result of the obstructing’.

The suffixes differ, however, in that they do not display one and the same productive pattern at the same moment. Figure 5 below, based on the relative spread of these suffixes or ‘productivity indexes’ (I, see Al & Booij 1989), shows the dominance of -heid in the adjectival pattern (5a) and its increase with participial bases (5b). The figure visualizes the decrease of participial and deverbal -nis (5b and 5c) and the predominance of -dom and especially -schap in the denominal pattern (5d). It also shows the upcoming differentiation between kinships and relationships in -dom (5d) and professions in -schap (5e) and the temporary extension of -(te)nis to bases referring to abstract nouns of the type geboorte, gelofte and gedachte (5f). As shown by the opposite developments in (5c) and (5f), the decline of -nis as a deverbal suffix coincided with the emergence and increase of denominal derivations.

Although the morphological and semantic changes described are thus in many ways similar, different chronologies and constant interaction between -nis, -heid, -dom and -schap have probably caused the suffixes to develop their own sense and prototypicalGa naar voetnoot14 morphological structure. Old Dutch -heid, denominal in origin and clearly distinguished from its concrete use by the 17^th century allomorph -igheid, produced deadjectival quality nouns from an early stage on. Together with the increase of productivity of quality nouns with -heid, which imposed no restrictions on the complexity of the adjectival bases, the productivity of deadjectival -dom and -schap decreased. Consequently, -schap and -dom remained generally close to

[pagina 117]

[p. 117]

Figure 5 Evolution of the productivity index of -nis, -heid, -schap and -dom (based on type frequency of unambiguous types)

[pagina 118]

[p. 118]

their noun-noun compound origins (‘the rank/state of X’) and especially -schap evolved into a fairly productive suffix to form denominal quality nouns referring to ranks or professions. It is not clear whether we should consider the fossilization of denominal abstract -dom and the following emergence of the collective meaning of -dom, formally expressed by the allomorph -endom, as a catalyzer or as a result of the growing overlap with quality nouns in -schap. The described changes also influenced the prototypical patterns of long-established, but in earlier studies neglected, -nis, a very productive Old Dutch suffix generating deverbal act nouns. Perhaps because of strong competition with deverbal -ing (e.g. beproevenesse and beproevinge), -nis first shifted to the domain of deadjectival quality nouns (e.g. duisternis or soetenisse), which was by then however already dominated by -heid. In the 17^th century, some deinfinitival act nouns ending in -heid and -schap were created (e.g. wetenheid or zeggenschap), which entered into competition with deverbal -nis. Finally, both deverbal and deadjectival -nis disappeared. By the 18^th and 19^th century the suffix mainly formed denominal concrete nouns. The emergence of -tenis with reinforcing abstract -te marked the inevitable non-productivity of -nis: -tenis was, unlike -nis/ -enis, not a phonologically determined allomorph, but rather a device to counterbalance the semantic erosion of the original suffix.

On the basis of all these observations I propose that one of the most decisive factors in the development of the distinct suffixes was rivalry. With the notion of rivalryGa naar voetnoot15 I want to generalize over the diachronic interactions that can be observed between the four nominalizing suffixes. Analogy and reinterpretation led to partial functional doublets, which may create pressure towards semantic differentiation but may also result in the gradual loss of a form/function pairing. Figure 6 visualizes the history of productive patterns with each of the suffixes. I would argue that the extinction of deadjectival -dom, -nis and -schap and their following divergent paths are at first caused by the high productivity of deadjectival -heid. In English, for example, denominal -hood did not extend its domain to the same extent (e.g. manhood, citizenhood or companionhood), as a result of which -dom and -ship and more specifically -ness successfully took on deadjectival quality nouns (e.g. freedom, obs. awaredom, kinship or obs. gladship and blindness, heaviness or even parabolicalness, see oed). Rivalry between -schap and -dom, in its turn, triggered differentiation between relationship nouns and finally collective nouns in -dom and profession quality nouns in -schap. At moments when nearsynonymous suffixes interfaced in non-lexicalized derivations, they were easily substituted by more productive ones (Wilmanns 1930: 392 and Van Bree 1996: 168). This is why gelegenisse, quelnesse and beternisse were replaced by derivations in -heid, -schap or -ing (cf. gelegenheit, gequelheit/kwelling or beterschap) or why collective -dom in present-day jodendom pushed out joodscap or joodsheit. Occasionally, ‘overcharacterization’ (Booij & van Santen 1998: 262) took place: the suffixal meaning was reinforced by the addition of another suffix, yielding concatenated forms such as 13^th century heerscapheit or maghedoemlijchheit.

[pagina 119]

[p. 119]

Figure 6 Schematic overview of the replacements of -nis, -heid, -dom and -schap

This first explorative diachronic outline of Dutch -nis, -heid, -dom and -schap has thus revealed that their semantic and morphological shifts sharply affected their present-day functions and that all abstract suffixes, regardless of their historic (dis)similarity, should be considered. Further research into abstract nominalising suffixes will give us more insight into the impact of interrelated changes within the paradigm of abstract suffixes, which hitherto have been studied too much in isolation. This case study might therefore be complemented in interesting ways with extensive comparative examinations into the English or German abstract affixes. Likewise, comprehensive study is warranted of underresearched Dutch suffixes -ing, -de/-te (‘-th’) or borrowed -(er)ij (‘-(er)y’), -isme (‘-ism’), -age, -atie (-‘ation’) and -iteit (‘-ity’), because they have all partaken in the rivalry within the abstract suffixal system of the Dutch language.

Bibliography

[pagina 120]

[p. 120]

[pagina 121]

[p. 121]

[Om privacyredenen is dit tekstgedeelte niet zichtbaar.]