Adverbial Lexicalisation in Faroese

Norsk Lingvistisk Tidsskrift · Årgang 37 · 2019 269–287 Adverbial Lexicalisation in Faroese Hjalmar P. Petersen In Faroese there is univerbation of phrases like onkran stað ‘some place’ to onkrastað- with deletion of the accusative ending -n in onkran ‘some’. In this particular case an adverbial suffix -ni is added giving the adverb onkrastaðni ‘somewhere’. It is univerbation and derivation of this kind that will be addressed here, where the second part of the compounds is -beint ‘straight’, -leið ‘path’, -stað ‘place’, -steðs ‘place’, -tíð ‘time’ and -veg ‘road’. Keywords: adverbial lexicalisation, derivation, Faroese, exaptation, petrification. 1 Introduction1 In this paper univerbation and derivation into adverbs in Faroese (FA) will be described and discussed. The results of the univerbation process, lexicalisation in the cases under consideration, are analysable into subparts, although the absolute final stage of lexicalisation is unanalysable words. The question of productivity in connection with lexicalisation and grammaticalisation will be discussed, and it will be shown that lexicalisation is concerned with the combination of tokens, not types. I will furthermore show that there is no abstract meaning in the forms under consideration, contrary to what happens in grammaticalisation. For more on these points, see Section 4. The data stem from spoken Faroese from the FADAC corpus in Hamburg (see Section 2); additionally I use data from Faroese Dictionary (Føroysk Orðabók) and Faroese Thesaurus (Samheitaorðabók)2 and also Seðlasavnið (a collection of words). The main patterns under consideration are given in (1). Note that there is no phonetic reduction in herbeint ‘here’ from her beint ‘straight here’, only change in meaning. In other cases, there is reduction, as in nógvar stað-ni > 1. I would like to thank two anonymous reviewers for comments on the paper, and H. Reinert for proof reading. 2. Available online at www.sprotin.fo. 270 Hjalmar P. Petersen nógvastaðni and allar tíðirnar ‘all the times’ to allatíðina ‘always’ with deletion of [ɹ] -r. (1) a. her beint here straight.NOM/ACC.NEUT.SG > herbeint ‘here’ b. so leiði-s so path-GEN.SG > soleiðis ‘like this; thus’ c. nógvar stað-ni many.ACC.PL place-SUFF > nógvastaðni ‘everywhere’ d. nógva steðs many places > nógvasteðs ‘everywhere’ e. allar tíðir-nar all.ACC.PL times-ACC.PL.DEF > allatíðina(r) ‘always’ f. beinan vegin straight.ACC.SG road.ACC.SG > beinavegin(ar) ‘at once’ Univerbation is here understood as ““becoming a word” of phrases” (Booij 2007: 19), and all the examples in (1) show that the modifying words and the head nouns are lexicalised; furthermore, there is derivation in -stað+ni, with addition of the adverbial suffix -ni. More will be said on the origin of this suffix in 4.3. In (1a) the adverb her ‘here’ and the adverb beint ‘straight’ become herbeint ‘straight here’. It is, as we shall see, possible to add -nar to this base, where the endpoint is herbeint-nar ‘straight here’. The form beint is nominative/accusative, singular, neuter, and such neuter forms can be used as adverbs in FA, as well as in Icelandic and e.g. Old Norse (ON). The example in (1b), soleiðis ‘thus’, is reconstructed as *swālaiðiōs ‘such + way’ in Proto-Norse, cf. Danish (DA) således (Katlev 2000), Old Swedish svaledhis, and Norwegian (NO) dialects solei(de)s. The genitive ending -s is reanalysed as part of the stem in FA (as well as in DA, Swedish (SW) and NO): leiði-s > leiðis- ‘path’. In FA the root is leið- ‘path’, and the ON form -i (leiði) and -s are reanalysed as a suffix, -is. Adverbial Lexicalisation in Faroese 271 Univerbation of an indefinite pronoun like nógvur ‘many’ + -stað ‘place’, (1c), is also found in other Scandinavian languages, cf. DA nogensteds ‘anywhere’, SW någonstans ‘anywhere’, NO nokonstan ‘anywhere’.3 It is possible to add the suffix -ni to the second part of the compound in FA, -stað ‘place’ > -staðni, as in nakrastað-ni ‘anywhere’. The origin of -ni is explained in Section 4.3, but it can be mentioned that this adverbial suffix is widely used in FA, sometimes for emphasis as in hjartaliga ‘heartily’ => hjartaliga-ni ‘very heartily’ and spakuliga ‘slowly’ => spakuliga-ni ‘very slowly’, with addition of -ni to the suffix -liga. The result -ligani is a new suffix, as a derivation with a suffix -lig + yet another suffix -ni is not commonly found in FA. The example in (1d) is with the borrowed second part -steðs from DA -steds ‘place’; in FA -steðs is simply an adverbial suffix. In (1e) there is univerbation of allar ‘all’ and tíð ‘time’ to allatíð-, with addition of the suffix -ina(r) resulting in allatíðina(r) ‘always’. Univerbation is also found in (1f) of the adjective beinan ‘straight’ and the masculine noun veg ‘road’ to beinaveg, and to beinavegin ‘at once’ with deletion of parts the accusative ending, that is -n in beinan, and ultimately to beinaveginar ‘at once’ with addition of –(n)ar. The paper is organised as follows. First the data will be presented and discussed, after which I discuss lexicalisation, phonetic reduction and comment on nakrantíð ‘ever’ and onkuntíð ‘sometimes’. These forms are strange, because the first part is masculine, the second part feminine. Then I will explain the origin of the suffixes, and finally there is a conclusion. 2 The data The data are retrieved from the FADAC database (Faroese-Danish-Corpus). The corpus was originally compiled for research into Faroese-Danish language contact and bilingualism at SFB Sonderforschungsbereich Mehrsprachigkeit at the University of Hamburg in the period 2005–2011.4 For more on this project see the web page and Petersen (2010). The FADAC database contains recordings from three generations of female and male speakers: younger speakers (16–21 years old), middle-aged speakers (40–50 years old), and older speakers (70+). They come from different locations 3. Thanks to an anonymous reviewer for pointing out this dialect form in NO, and also for pointing out annaleis and annsleis ‘different’, which will be discussed in 3.2. 4. https://www.uni-hamburg.de/sfb538/forschungsprogramm/historische-aspekte/varianz.html 272 Hjalmar P. Petersen on the Faroe Islands, and they speak Faroese and Faro-Danish in the interviews, which are informal in nature. For this study I have used 7 older male and female speakers, 10 male and 11 female speakers from the mid generation, and 6 younger male speakers and 7 younger female speakers. I searched for the relevant sequences in EXMARaLDA, which is a program that researchers at SFB in Hamburg developed for working with spoken data. For more on this program, see exmaralda.org. In addition to spoken data, I have used Seðlasavnið at Føroyamálsdeildin. Seðlasavnið is a collection of words.5 In the tables, the data from Seðlasavnið have the abbreviation SSv. Other abbreviations are: W = woman, M = man; O = older, M = Mid, Y = Younger. I have also used Faroese Dictionary (Føroysk Orðabók) and Faroese Thesaurus (Samheitaorðabók). 3 Univerbation and derivation The examples in (1) are all cases of univerbation into adverbs, as in allatíðina(r) ‘always’ from allar tíðirnar ‘all the times’, where the indefinite pronoun in the accusative plural allar ‘all’ modifies the feminine noun tíð+irnar ‘time.ACC.PL.DEF’. Note that the singular form alla tíðina ‘all times’ cannot be the base of allatíðina ‘always’ as becomes evident from allatíðinar with word final -r originally from the definite ending in the accusative plural in the feminine. In (1) there are further examples of univerbation and derivation in onkran stað ‘some.ACC.SG.MASC place’, ‘some place’, which changes into onkrastað-. This word form is the base to which the adverbial suffix -ni is added, and the result is the adverb onkrastaðni ‘somewhere’, with deletion of part of the accusative ending, that is -n in onkran ‘some’. Another suffix is -nar in for example herbeint-nar ‘straight here’. I will address the origin of these suffixes in 4.3 after having gone through the actual forms under consideration in Section 3. A discussion of the numbers in the different figures will be postponed to Section 3.5. As the article builds on data from a limited corpus, I do not want to be too categorical and claim e.g. that certain forms are used mainly by a definite age group or gender. 5. https://setur.fo/gransking/soevn/talgildu-sovnini/ordasedlasavn/ Adverbial Lexicalisation in Faroese 273 3.1 Adverb + -beint ‘straight’ There is univerbation of the adverbs her ‘here’ and har ‘there’ and the adjective beinur ‘straight’ in the nominative/accusative neuter singular, beint ‘straight’. The results are the adverbs herbeint ‘straight here’ and harbeint ‘straight there’. Primary stress is on the first part of the compound and secondary stress on the last part, as expected in FA compounds. The tokens in FADAC are listed in Table 1, where the -nar form is mostly used by young female speakers. The older forms are obviously herbeint ‘here’ and harbeint ‘there’, and the younger forms are herbeintnar ‘straight here’ and harbeintnar ‘straight there’. Adverb herbeint harbeint herbeintnar harbeintnar Translation ‘here’ ‘there’ ‘here’ ‘there’ MO WO 1 MM WM 13 3 MY WY 1 1 2 7 Table 1. Tokens with -beint and -beintnar 3.2 Annarleiðis ‘different’, beinleiðis ‘straight’ and soleiðis ‘thus’ The second part of these compounds, -leiðis ‘path’, is the genitive singular form of ON ija-stems. Kvaran (1990/1991: 15) shows examples with -leiðis from the 14th century in ON, hafleiðis ‘seaways’ and réttleiðis ‘straight ahead’. Fritzner (1896: 91) in his large dictionary on Old Norse treats réttleiðis ‘straight ahead’ as an adverb. The genitive singular form of the ija-stems was used for time and place adverbials in Old Norse (Kvaran 1990/1991), and the forms that have survived into FA are also adverbs. Of the three adverbs in FADAC with -leiðis, annarleiðis ‘differently’ and soleiðis ‘like this, thus’ exist also in other Scandinavian languages, like DA anderledes ‘different’ from annarr ‘another.NOM.SG.MASC’ (Katlev 2000: 86), SW annorledes ‘different’, and NO has forms like annleis, annaleis and annsleis ‘different’. I thank an anonymous reviewer for pointing out these forms in NO dialects for me. Soleiðis ‘like this’ dates back to Proto-Norse *swālaiðiōs ‘such way’. Forms where the noun lund ‘way, character’ has been used to form an adverb in Scandinavian do not exist in FA, except as a loanword nøku(n)lunda ‘somehow’ from DA nogenlunde ‘somehow’ with preservation or deletion of the -n in FA. 274 Hjalmar P. Petersen The suffix -nar, which is used in herbeint-nar ‘straight here’ and harbeintnar ‘straight there’, shows up in a reduced form in soleiðis-na ‘thus’, while the nonreduced form is soleiðis-nar. Adverb soleiðis soleiðisnar soleiðisna soleiðisni annarleiðis beinleiðis Translation ‘thus’ ‘thus’ ‘thus’ ‘thus’ ‘differently’ ‘straight’ MO WO MM WM MY WY SSv. 17 36 93 131 43 52 11 19 3 16 42 11 7 3 1 17 7 1 1 6 5 Table 2. Tokens with -leiðis and related variants It is evident from Table 2 that the older form, soleiðis, is used by all generations, while -ni and -na are rare. The form soleiðisna ‘thus’ was used by only one male speaker, so it is quite possible that the use of it is not tied to generations or gender, but to individuals. In addition to the forms in Table 2, FA has somuleiðis ‘in the same way’. The form soleiðis-ni does also exist (https://seðlasavn.setur.fo). 3.3 Univerbation of quantifiers The examples in 3.3.1 to 3.3.3 have quantifiers as the first part of the compound. I have included the borrowed word forskelligur ‘different’ (from DA forskellig ‘different’) in the group of quantifiers, since it has the same distributive meaning as ymiskur ‘different’; cf. also ymsastaðni ‘in different places’ = forskelligastaðni ‘in different places’. The second parts of the compounds with quantifiers as their first part denote ‘time’ and ‘place’, and by having a quantifier as the first part, the results are adverbs like ‘everywhere’, ‘somewhere’ and so on. One very common reason behind word borrowing between languages is need. Need is also the driving force behind the univerbation of the quantifiers + NP under consideration. To the best of my knowledge, no simplex words existed in FA with the meaning ‘anywhere’, ‘elsewhere’, ‘everywhere’, ‘nowhere’ and ‘somewhere’ prior to the univerbation of the phrases under discussion. The same basically also holds for ‘never, sometimes, anytime’. Adverbial Lexicalisation in Faroese 275 Here are some examples: For aðrastaðni ‘elsewhere’, Faroese Dictionary gives á øðrum stað ‘at a different place’. For ‘sometimes’ the dictionary has á nøkrum sinni, lit.: ‘on some occasion’; hence the new adverb nakrantíð ‘sometimes’. Need must also have been the driving force behind what came to be ongantíð ‘never’. The only synonym to ongantíð ‘never’ is the borrowed DA word aldri ‘never’. Need is also the motivation behind onkuntíð ‘sometimes’. Faroese Dictionary lists eina ferð, lit.: ‘some time’, ‘sometimes’ as a synonym to onkuntíð and viðhvørt ‘sometimes’; that is, only viðhvørt ‘sometimes’ seems to have existed before the lexicalisation discussed. 3.3.1 Quantifier + -stað ‘place’ In Table 3, all the examples from the FADAC database with -stað ‘place’ are listed, and in all cases the first part of the compound is a quantifier. Univerbation of a quantifier + stað- ‘place’ is familiar in Scandinavian, compare SW någonstans ‘somewhere’ and so on. Note the suffixes that are used with -stað; they are -ni, -nis, -na and a combination of -na (in e.g. allastaðna ‘everywhere’) and -ni to -nani, which then is a new suffix altogether. Adverb aðrastaðni aðrastaðnis allastaðni allastaðnis allastaðna allastaðnar allastaðnani forskelligastaðni nakrastaðni nakrastaðnis nógvastaðna nógvastaðni Translation ‘elsewhere’ ‘elsewhere’ ‘everywhere’ ‘everywhere’ ‘everywhere’ ‘everywhere’ ‘everywhere’ ‘everywhere’ ‘anywhere’ ‘anywhere’ ‘everywhere’ ‘everywhere’ MO WO MM WM MY WY SSv 1 2 7 4 2 9 4 3 6 3 7 5 11 7 2 1 1 1 1 1 2 2 1 1 1 1 1 1 1 1 276 Hjalmar P. Petersen Adverb nógvastaðnani6 nógvastaðnis ongastaðna ongastaðni ongastaðnis onkrastaðnis onkustaðna onkustaðni onkustaðnis Translation ‘everywhere’ ‘everywhere’ ‘nowhere’ ‘nowhere’ ‘nowhere’ ‘somewhere’ ‘somewhere’ ‘somewhere’ ‘somewhere’ MO WO MM WM MY WY SSv 1 1 1 1 1 8 1 3 1 2 3 1 2 6 4 1 Table 3. Tokens with -staðni and related variants The different forms are discussed in detail in Sections 3.5 and 4.6 In addition to the forms in Table 3, Faroese Dictionary has ymsastaðni ‘everywhere’, and Faroese Thesaurus lists ymsastaðna ‘everywhere’. Both have summastaðni ‘somewhere’, and the latter also has summstaðni ‘somewhere’. In these cases, the first parts of the compounds are ymiskur ‘different’ and summir ‘some’. 3.3.2 Quantifier + -steðs ‘place’ The second part of the compounds in Table 4, -steðs, stems from DA -steds ‘place’, and it is an adverbial suffix in FA. Adverb allasteðnani allasteðs allasteðsinar aðrasteðs nakrasteðs Translation ‘everywhere’ ‘everywhere’ ‘everywhere’ ‘elsewhere’ ‘somewhere’ MO WO MM WM MY WY SSv 1 1 1 2 1 6. The form nógvastaðnani ‘everywhere’ is attributed to Janus í Dímun, who lived on the isolated island Dímun. L. Woxnes Niclasen from Suðuroy is familiar with the suffix -nani. I have not heard them before (my dialect is Western Faroese, Vágar). In FTB Føroyskur talumálsbanki (Corpus of Spoken Faroese, http://clarino.uib.no/korpuskel/corpus-list?session-id=246561192788539) I have found allastaðnani ‘everywhere’. The speaker is also from Suðuroy, a male born in 1942). Adverbial Lexicalisation in Faroese 277 Adverb ongasteðs onkrasteðs Translation ‘nowhere’ ‘somewhere’ MO WO MM WM MY WY SSv 1 3 Table 4. Tokens with -steðs and related variants 3.3.3 Quantifier + -tíð ‘time’ Univerbation with -tíð ‘time’ as the second part is well known in Germanic, cf. English (EN) sometimes, FA summar tíðir ‘some times’ and for example DA altid lit.: ‘all time’, ‘always’ and FA altíð lit.: ‘all time’, ‘always’. Adverb allatíð allatíðinar allatíðina ongantíð onkuntíð onkutíð onkutíðinar onkutíðina nakrantíð Translation ‘always’ ‘always’ ‘always’ ‘never’ ‘sometime’ ‘sometime’ ‘sometime’ ‘sometime’ ‘ever’ MO WO MM WM MY WY SsV 2 1 2 4 10 23 1 30 21 42 61 6 8 10 17 6 13 46 86 7 9 1 1 2 9 2 2 3 8 2 5 Table 5. Tokens with -tíð and related variants The numbers will be briefly discussed in Section 3.5. 3.4 Forms with -veg ‘way’ In EN, way occurs in adverbs like always, and likewise in FA we find lexicalisation in forms such as those in Table 6. Beinan vegin is literally ‘the straight road’ but here it means ‘at once’. The noun vegur ‘way’ also figures in eisini ‘also’, originally from eins veg ‘same road’ (Weyhe 2018). 278 Hjalmar P. Petersen Adverb allanvegin beinavegin beinaveg beinaveginar hinvegin Translation ‘the whole way’ ‘at once’ ‘at once’ ‘at once’ ‘on the other hand’ MO WO MM WM MY WY SsV 1 1 2 4 5 7 9 7 5 1 1 1 Table 6. Tokens with -veg and -vegin 3.5 Summary As the data are scant, it is best to be cautious and not too categorical with regard to any definite conclusion. It is perhaps not surprising that the base (shorter) forms are used by all speakers: herbeint ‘here’, soleiðis ‘thus’, onkustaðni ‘somewhere’, allasteðs ‘everywhere’, allatíð ‘always’, onkutíð ‘sometimes’ and ongantíð ‘never’. The second step is represented by herbeintnar ‘right here’ and forms like soleiðisnar ‘thus’ and e.g. onkutíðinar ‘sometimes’ and beinaveginar ‘at once’. A more reduced form like allatíðina ‘always’ is also used by all generations. The forms with -nani are few, and possibly dialectal (Suðuroy). 4. Lexicalisation The terms lexicalisation and grammaticalisation have been subject to lively debate among researchers working on the topics, and they are still much debated. It is not possible to go into any details about what exactly counts as grammaticalisation and lexicalisation in this short article, so for examples I refer to Traugott & Trousdale (2016), Brinton (2012: 135–167), Himmelmann (2004), Lehmann (2002), Anttila (1989: 149–151) and Bauer (1983: 42–61). A very short definition of lexicalisation is “whenever a linguistic form falls outside the productive rules of grammar it becomes lexicalised” (Anttila 1989: 151). This means that all the adverbs under consideration here are the result of lexicalisation. According to Lehmann (2002: 13), lexicalisation “involves a holistic access to a unit”, which means that a collocation is not divisible into parts (Brinton 2012: 144), and it is treated as a lexical item in the lexicon. Phrased differently, lexicalisation means decrease in schematicity and productivity (Traugott & Trousdale 2013). Weyhe (2018) shows how FA kortini ‘nevertheless’ develops from the subordinate conjunction hvárt (ið) er ‘what (that) is’ Adverbial Lexicalisation in Faroese 279 > *kvárti > korti (dialectal in Sandoy, now extinct). Then there is addition of the suffix -ni to kortini ‘nevertheless’. Likewise he shows how eisini ‘also’ has its origin in eins veg ‘same road’ > einsig- > eisi-ni > eisini. These forms, kortini ‘nevertheless’ and eisini ‘also’, are certainly accessed as a unit by speakers of FA and are not divisible into parts; they have developed into a unit. Speakers do not even recognise -ni as a suffix in kortini ‘nevertheless’ and eisini ‘also’. The same can be observed in alskyns /alʃɪns/ ‘all sort of’ from als kyns ‘all.GEN.SG. kyns.GEN.SG ‘kind’. In this word, the expected palatalization of kyin kyns to an affricate /ʧ/ has not happened. Instead –sky, alskyns, has palatalized into /ʃ/. This clearly shows how als kyns is lexicalized into one word alskyns, and that speakers access it as a unit, and not as a compound als+kyns *[alsʧɪns/. It is easy to agree with the claim that a lexicalised form involves holistic access to a unit, and that the result is a lexeme with a direct mapping between form and meaning like [allastaðni] ‘everywhere’ and [alskyns] ‘all sort of’. Additionally, there is no reason to dispute that the definite endpoint of lexicalisation is an opaque form, as shown with kortini ‘nevertheless’, eisini ‘also’ and alskyns ‘all sort of’. But a lexicalised form is definitely not always completely opaque. Bauer (1983: 49) points to EN warm-th, which is “analyzable but lexicalized”. He further suggests that items can be lexicalised in different ways, and he suggests a split into three different types of lexicalisation depending on the linguistic levels (phonological, morphological and syntactic). Phonological lexicalisation involves phonetic reduction. An example of morphological grammaticalisation is, as will be shown in detail in Section 4.3, when the plural ending -ir and the definite suffix -nar in the accusative plural feminine shows up as a suffix -na in for example allastaðna ‘everywhere’. Deandative is an example with syntactic grammaticalisation. This is seen when the verb at fara ‘to go’ is used as an auxiliary in FA to express the future: vit fara at renna ‘we will run’. Deandative is similarly found in EN be going to. In allatíðina ‘always’ and beinavegin ‘at once’ and all the other forms under consideration here, speakers are definitely able to analyse the compounds into subparts. This is more in accordance with Himmelmann’s (2004) understanding of lexicalisation, where he says that in lexicalisation “an originally compositional expression is treated more and more as a single processing unit, thereby becoming more and more opaque and finally an unanalyzable simple sign”. Lexicalization is thus gradual (L1 > L2 > L3) as also pointed out by Brinton & Traugott (2005), so that speakers are at first able to form a connection or a network between the word forms, before they develop into one single unit, e.g.: 280 Hjalmar P. Petersen L1 soleiði-s > soleið-is > L2 soleið-isna(r)/soleið-isni > L3 [soai:s] or [soui:s]. The forms that are reached at L3 exist in spoken FA, and if or when these become the only forms available to the speaker, we will have an unanalysable simple sign (and we could write them as soeiðs or soíðs). The loss of constituency in e.g [DP allar [NP tíðirnar]] to an adverb [AdvP allatíðina] is abrupt, but it is still possible to analyse the single parts in the compound, until finally it becomes completely opaque. One result of grammaticalisation7 is productivity, which means that forms become more widely applicable (Brinton 2012: 158). The formation of the adverbs under discussion is not productive; thus, it is not possible to add another pronoun like the demonstrative pronoun hesin ‘this’ and get a form *hesastaðnis ‘at this place; right here’. Lexicalisation is concerned with the combination of tokens, not types (Brinton 2012: 146) with reference to Himmelmann (2004)). This means that the tokens allar tíðirnar ‘all the times’ change into allatíðina ‘always’, which then ultimately becomes a new lexeme in the language, and a new type. Additionally, there is no abstract grammatical meaning in the forms under investigation; abstract grammatical meaning is one result of grammaticalisation, as when the verb at fara ‘to go’ is used as a future time marker in the construction fara ‘to go’ + infinitive (deandative). This is not the case when for example beinan ‘straight’ + vegin ‘road’ = ‘the straight road’ is lexicalised to beinavegin ‘at once’. 4.1 Increased autonomy and phonetic reduction without loss of analysability In this section, I will first explain the origin of the first parts of the compounds, and then address the question of analysability of the compounded adverbs that are mentioned in Sections 3.1 to 3.4. The first parts of the compounds stem for a large part from the accusative singular/plural, and it is not unusual in FA to have adverbial accusatives: (2) a. Vit koma mánadagin. we come Monday.ACC.SG.MASC.DEF ‘We will arrive on Monday.’ 7. Himmelmann (2004: 33) on grammaticalization: It is a process of context expansion on three levels: (a) host-class expansion (the class of elements a grammatical form is in construction with, the host-class, expands); (b) expansion of the syntactic context; and (c) expansion of the semantic-pragmatic context; see also Brinton (2012: 145). Adverbial Lexicalisation in Faroese 281 b. Vit arbeiða allan dagin. we work all day.ACC.SG.MASC.DEF ‘We will work the whole day.’ c. Vit arbeiða alla tíðina. we work all time.ACC.SG.FEM.DEF ‘We work all the time.’ The verbs koma ‘to come’ and arbeiða ‘to work’ in (2) are intransitive (unaccusative and unergative, respectively). This means that the nouns or noun phrases mánadagin ‘on Monday’, allan dagin ‘the whole day’ and alla tíðina ‘the whole time’ are not objects. Also, it is not the case that we ‘work on the whole day’ and that ‘we work on the whole time’. The nouns do not have a patient thematic role, but are adjoined adverbial phrases, which formally are nouns in the accusative. Univerbation of the accusative plural is seen in ongar stað-. The nominative plural of eingin ‘no one’ is eingir or ongir ‘no one’, and the accusative has the ending -ar: ong-ar. This turns up in ongastað-ni ‘nowhere’ from ongar staðplus addition of the suffix -ni. The final -r in ongar is deleted in the compounded form. A discussion of the origin of -ni is postponed to 4.3. The accusative plural is found in aðrar stað- ‘elsewhere’ (nom.m.: aðrir ‘other’); allar stað- ‘everywhere’ (nom.m.: allir ‘all’); nakrar stað‘somewhere’ (nom.m.: nakrir), nógvar stað- ‘everywhere’ (nom.m.: nógvir ‘some’). The masculine singular accusative is seen in annanstaðni ‘elsewhere’, in older Faroese in the ballads about Sjúrður that Schrøter wrote down in 1818 (Matras 1951–53: 16). Annanstaðni is not known in modern Faroese. When a form like aðrastaðni ‘elsewhere’ has emerged, the first part is replicated with the borrowed suffix -steðs in aðrasteðs ‘elsewhere’. The same replication happens with the other forms with the head -steðs. The base in onkustaðni ‘somewhere’ and other forms with onku(n) is the accusative singular form onkun ‘some’. Sandøy (1992) mentions this relatively rare form in his discussion of the use of the indefinite pronouns onkur ‘some’, nakrar ‘some’ and summur ‘some’ in FA. Hammershaimb (1854 [1969]: 298) has onkun and says, that it is truncated from einhvörn ‘some’, compare the ON indefinite pronoun einhverr ‘some’ and the accusative einhvern. In Faroese Dictionary (online: www.sprotin.fo) we find onkun stormdag fer træið um koll ‘some stormy day the tree will fall’. 282 Hjalmar P. Petersen The accusative plural is the base in allatíðinar ‘always’ from allar tíðirnar ‘all times’, lit. ‘all.ACC.PL.DEF.FEM time.ACC.PL.DEF.FEM’. In addition to -tíðinar we find -tíðina in Table 5, and this is a further development of -tíðinar > -tíðina. It is perhaps more difficult to explain the -n in nakra-n-tíð ‘ever’ and onkun-tíð ‘sometimes’, and this is the subject of Section 4.2. The accusative singular is the base behind beinavegin ‘at once’ from beinan ‘the straight’ + veg-in ‘the road’, with the definite article -in. It is unlikely that the weak form beina vegin is the base, as we find beinaveg ‘at once’ < beinan veg ‘straight road’ with an indefinite noun, veg ‘road’. In annarleiðis ‘different’ it is the nominative singular which is the base, cf. also DA anderledes (Katlev 2000: 86). Bybee (2010: 48) discusses cases where complex units maintain their internal structure and their relations with the other uses of their component parts. In the process of univerbation, the constituent structure is lost, so that [DP alla [NP tíðina]] (or [DP allar [NP tíðirnar]]) becomes [AdvP allatíðina]. But there is no reason to assume that speakers of Faroese are not able to recognise the subparts in the compound [AdvP allatíðina], and form a network between the adverbial compound and the corresponding lexemes ALLUR ‘all’ and TÍÐ ‘time’. In many of the cases under consideration, there is phonetic reduction of the first part, which has come about due to frequency of use, but this reduction has not, in most cases, resulted in loss of compositionality and analysability. 4.2 Nakrantíð ‘ever’ and onkuntíð ‘sometimes’ The form onkun- in onku(n)tíð ‘sometimes’ is suspect, and so is nakran- in nakrantíð ‘ever’, since the head noun, tíð ‘time’, is feminine, and onkun/nakran are acc.sg.masc. The bases behind the lexicalized adverbs are adverbial accusatives, which means that the expected forms should be *onkra+tíð lit.: some.SG.ACC.FEM + time.SG.ACC.FEM. and *nakra+tíð lit.: some.SG.ACC.FEM + time.SG.ACC.FEM. Neither *onkratíð nor *nakratíð exist in FA. The first part of onkuntíð/onkutíð ‘sometimes’ is the masculine singular accusative form onkun ‘some’ (nom.sg. onkur). Onkun is the first part in onkustaðnis ‘somewhere’ from onkun-ACC.SG.MASC stað-ACC.SG.MASC. Once [DP onkun [NPstað]]+ni has lexicalized into [onku(n)staðni] ‘somewhere’, onkunspreads analogically and is used in onku(n)tíð ‘sometimes’. Likewise the masculine sg. form nakran- from nakran-ACC.SG.MASC stað-ACC.SG.MASC spreads analogically to nakrantíð ‘sometimes’. This is possible after the loss of constituency, as the words now are not accessible for syntactic manipulation, but Adverbial Lexicalisation in Faroese 283 belong to the morphological component of the language. Speakers focus, in other words, on a relational match between Quantifier + stað/tíð, and the result is nakrantðíð ‘ever’ and onkuntíð ‘sometimes’. It is quite possible that lexicalisation of quantifier + place happened first, as place is so central to the speaker. After this, there was lexicalisation of quantifier + time. 4.3 The suffixes: origin and use Here I will address the second part of the compounds and explain the origin of the suffixes that are used in the univerbation process. For the sake of convenience, they are listed here with illustrative examples: (3) a. b. c. d. e. f. g. -is soleiðis ‘thus’ -(i)na(r) allatíðina(r) ‘always’ -nani allastað-nani ’everywhere’ -ni(s) allastaðni(s) ‘everywhere’ -t herbeint ‘right here’ -in beinavegin ‘at once’ -isna(r) soleiðisnar ‘thus’. As already mentioned, the origin of FA -is is the genitive-s of ON ija-stems. These stems had -i in the nominative, accusative, dative, singular, and -s in the genitive singular. The genitive singular forms were used as time and place adverbials (Kvaran 1990/1991), such as árdegis ‘early’. The -s of this genitive ending has spread analogically to the adverbial suffix -ni resulting in -nis as in allastað-nis ‘everywhere’. Note that the suffix -s in modern FA is synchronically different from the ON genitive -s, since the genitive is basically absent in FA (Petersen & Szczepaniak 2018). The -s in soleiði-s ‘thus’ and similar examples is a petrified form, and as it is reanalysed into the adverbial suffix -is, it spreads to -ni, giving -nis. It is also possible to have soleið-is+nar, with addition of -nar, which originally is the definite ending in the accusative plural. This gives us one more suffix: -isnar (3g). What then is the origin of the -ni suffix in adverbs like allastað-ni ‘everywhere’? Synchronically -ni is an adverbial suffix, but the origin of -ni is not the typical cline of lexicality, which is very often the base of derivational suffixes in Germanic, as in manns lík ‘a man’s body’ where lík ‘body’ is part of a phrase > FA man-s-lík (part of compound) to a derivational suffix mans-lig-ur ‘manly; masculine’. 284 Hjalmar P. Petersen It is very common to use the -ni suffix to derive adverbs in FA, or even to add it to existing adverbs like illa ‘badly’ > illa-ni ‘badly’, ofta ‘often’ > oftani ‘often’ and toluliga ‘tolerably’ > toluliga-ni ‘tolerably’. It is possible that it can be used for emphasis, as in hjartaliga tillukku ‘sincere congratulations’ > hjartaliga-ni tillukku ‘sincere congratulations (and I really mean it)’. With regard to age, there are examples from 1818 of annanstaðni from Schrøter’s transcription of Sjúrðarkvæði (Matras 1951–53: 16), and from 1819 from ballads Jhs. í Króki wrote down; he has houlija-ni = hóvliga-ni ‘gently’ (in verse 35, 36, 65, 66) (Weyhe 2011). Now to the origin of -ni. First the vowel -i was added to adverbs with word final -n like haðan ‘from there’ and vestan ‘from the west’ to haðan-i ‘from there’ and vestan-i ‘from the west’. This -i was originally used as a locative marker as in út ‘out’ > út-i ‘outside’, inn ‘in’ > inn-i ‘inside’, and it has further been added to forms like yvir ‘over’ to yvir-i ‘over’ and so on. Given a new base haðani ‘from there’ and eystani ‘from the east’, speakers have apparently reanalysed -n and -i as -ni, and have added it to different bases, of which one is -stað ‘place’, with the new form -stað-ni. Note that -s from -leiðis and other ON ija-stems can further be added to -ni giving -nis in for example -stað-nis (Petersen 2018; Weyhe 2018). Campbell (2013: 261) defines exaptation as “cases where phonological material takes on a new function, unrelated to its original or obsolete function in the language”. The suffixes -ni and -nis are, according to this, examples of exaptation. Brinton (2012: 135-136) discusses the familiar cline, content word > grammatical word > clitic > inflectional affix > Ø, cited in Hopper & Traugott (2003: 7). She notes that zero is not the only possible endpoint of grammaticalization. Exaptation is one possibility, and another is where inflectional endings may develop into derivational (lexical) affixes, and this is exactly what is the case for the suffix -(i)na)(r) in (3b). The suffix -(i)na(r) stems originally from the definite ending in the plural; that is allar tíðirnar lit.: ‘all-ACC.PL.FEM timeACC.PL.FEM’ ‘all the times’ > allatíð-ina-(r) ‘always’. Final -r is optional, and the unstressed -i can easily be deleted, so the endpoint can be -na as in soleiðis-na8 ‘thus’, allastað-na ‘everywhere’, nógvastað-na ‘somewhere’, ongastaðna ‘nowhere’. The definite ending has in this way developed into an adverbial suffix in Faroese and is part of Faroese derivational morphology. 8. As -is is an adverbial suffix in FA, -isna is a new adverbial suffix, as already mentioned. Adverbial Lexicalisation in Faroese 285 There exist further the interesting forms from the Seðlasavn, -nani, which really show that both -na and -ni have developed into independent suffixes, as they can be combined into one new suffix in allasteð-nani ‘everywhere’, nógvastað-nani ‘everywhere’. Finally, the form -in is found in beinaveg-in ‘at once’, and this is the definite ending in the accusative singular in the masculine. The forms are connected in a network, and it can be illustrated as follows, at least as a suggestion: (4) That is, there is influence from -(i)s on -ni to -nis (allastað-nis ‘everywhere’) and from -is to -na (soleiðisna); the -in in beinaveg-in has a relation to -(i)na(r) (beinaveginar). The -t in herbein-t can get -nar added as in herbeint-nar and finally we can get a combination of -na+ni to -nani. Conclusion The result of univerbation in the cases under consideration is lexicalisation, where there is an abrupt loss of constituency in addition to possible phonetic reduction. There is no reduction in her beint ‘right here’ > herbeint ‘here’, as opposed to the complex semi-idiomatic allatíðina ‘always’ from allar tíðirnar ‘all the times’. Both words are accessed as analysable units, contrary to the completely unanalysable simplex form kortini ‘nevertheless’ < hvárt (ið) er ‘what (that) is’. Although the main topic of the article is lexicalisation into adverbs, there are also examples of grammaticalisation, as shown with exaptation and cases where an inflectional ending has developed into a derivational suffix. One such suffix, -ni, is productive in Modern FA, and it is used for emphasis, as in hjartaliga-ni ‘very heartily’. More research is needed on the nature and 286 Hjalmar P. Petersen productivity of -ni. The examples that have been discussed in this paper show clearly that lexicalization is gradual, L1 > L2 > L3 as illustrated with L1 soleiði-s > soleið-is > L2 soleið-isna(r)/soleið-isni > L3 [soai:s], [ soui:s], and that the final stage is an opaque form as for example in eisini ‘also’ < eisi-ni < *einsig- *< eins veg ‘same road’. References Antilla, Raimo. 1989. Historical and Comparative Linguistics. Amsterdam/ Philadelphia: John Benjamins Publishing Company. Bauer, Laurie. 1983. English Word-formation. Cambridge: Cambridge University Press. Booij, Gert. 2007. The Grammar of Words. An Introduction to Morphology (2nd ed.). Oxford: Oxford University Press. Brinton, Laurel J. & Elisabeth Traugott Closs. 2005. Lexizalication and Language Change. Cambridge: Cambridge University Press. Brinton, Laurel J. 2012. The ghost of old morphology: Lexicalization or (de)grammaticalization? In Davidse, Kristin, Tine Breban, Liselotte Brems & Tanja Mortelmans (eds.), Grammaticalization and Language Change. New Reflections. Amsterdam/Philadelphia: John Benjamins, 135–166. Bybee, Joan. 2010. Language, Usage and Cognition. Cambridge: Cambridge University Press. Campbell, Lyle. 2013. Historical Linguistics. An Introduction (3rd ed.). Edinburgh: Edinburgh University Press. Fritzner, Johan. 1896. Ordbog over Det gamle norske Sprog. Oslo: Tryggve Juul Møller Forlag. Hammershaimb, Venceslaus U. 1854. Færøisk Sproglæra. D. Quist Bog- og Nodetrykkeri. Himmelmann, Nikolaus P. 2004. Lexicalization and grammaticalization: Opposite or orthogonal? In Bisang, Walter, Nikolaus P. Himmelmann & Björn Wiemer (eds.), What Makes Grammaticalization – A Look from its Components and its Fringes. Berlin: Mouton de Gruyter, 19–40. Hopper, Paul & Elisabeth Closs Traugott, Elisabeth. 2003. Grammaticalization (2nd ed). Cambridge: Cambridge University Press. Katlev, Jan. 2000. Politikens Etymologiske Ordbog. København: Politikens Forlag. Kvaran, Gurðrún. 1990/1991. Um -is endingu atviksorða. Íslenskt mál 12/13, 7–34. Adverbial Lexicalisation in Faroese 287 Lehmann, Christian. 2002. New reflections on grammaticalization and lexicalization. In Wischer, Ilse & Gabriele Diewald (eds.), New Reflections on Grammaticalization. Amsterdam/Philadelphia: John Benjamins, 1–18. Matras, Christian. 1951–53. J.H. Schrøters Optegnelser af Sjúrðar Kvæði. København: Munksgaard. Petersen, Hjalmar P. 2010. The Dynamics of Faroese-Danish Language Contact. Heidelberg: Universitätsverlag Winter. Petersen, Hjalmar P. 2018. Føroysk mállæra II. Orðmyndan, kyn og bending (manuscript). Hjalmar P. Petersen & Renata Szczepaniak. 2018. The Development of nonparadigmatic linking elements in Faroese and the decline of the genitive case. In Ackerman, Tanja, Simon J. Horst & Christian Zimmer (eds.), Germanic Genitives. Amsterdam/Philadelphia: John Benjamins, 115–145. Sandøy, Helge. 1992. Indefinite Pronouns in Faroese. In Louis-Jensen, Jonna & Jóhan H.W. Poulsen (eds.), The Nordic Languages and Modern Linguistics. Tórshavn: Føroya Fróðskaparfelag, 547–554. Traugott Closs, Elisabeth & Trousdale, Graeme. 2013. Constructionalization and Constructional Changes. Oxford: Oxford University Press. Weyhe, Eivind. 2011. Hentzasavn. Kvæðasavnið hjá Jóannesi í Króki frá 1819. Tórshavn: Fróðskapur. Weyhe, Eivind. 2018. Eisini, hissini og kortini (manuscript, submitted to Fróðskaparrit). Hjalmar P. Petersen Faculty of Faroese Language and Literature University of the Faroe Islands V.U. Hammershaimbsgøta 22 FO-100 Tórshavn [email protected]