Perfect participle agreement and restructuring in Norwegian

Perfect participle agreement and restructuring in Norwegian Teodor Ekblad Aagaard FTL March 30, 2016 1 The phenomenon The phenomenon under investigat...
Author: Preston Mason
6 downloads 1 Views 393KB Size
Perfect participle agreement and restructuring in Norwegian Teodor Ekblad Aagaard FTL March 30, 2016

1 The phenomenon The phenomenon under investigation is sentences like (1) and (2), where the verbal complement of a perfect participle (henceforth PPC) itself appears as a perfect participle where the standard language would have an infinitive. This apparent agreement is optional and has no semantic effect. (1) Jeg har glemt å pakket positiv innstilling og godt humør da… I have forgotten to packed positive attitude and good spirits though (NoWaC) (2) Eneste som overrasker meg er at de har klart å animert dårlig skuespill only that surprises me is that they have managed to animated bad acting “The only thing that surprises me is that they have managed to animate bad acting” (NoWaC) PPC agreement is fairly widespread in Faroese, Norwegian and Swedish. It is much more limited in Danish and does not occur in Icelandic (see Larsson 2014 and works cited there for an overview). In some Swedish dialects, verbal feature agreement is possible throughout the paradigm. Data from Wiklund (2007:3-4): (3) Han börjar o skriver dikter he begins to writes poems (4) Han började o skrev dikter he began to wrote poems (5) Börja o skriv dikter! begin.IMP to write.IMP poems! (6) Han hade börjat o skrivit dikter he had begun to written poems Present and past tense agreement is more restricted than imperative and PPC agreement in Swedish (Wiklund 2007:64-65; Teleman et al. 1999/3: 273-274) and the same is apparently true of Norwegian. This is different from verbal feature agreement in pseudocoordinations. With pseudocoordination verbs like være ‘be’, agreement is obligatory throughout the paradigm. (See Lødrup 2014a for further differences regarding voice agreement). (7) Jeg har vært og drukket/*drikke en øl I have been and drunk/*drink a beer Jeg var og drakk/*drikke en øl I was and drunk/*drink a beer Jeg har klart å drukket/drikke en øl I have managed to drunk/drink a beer Jeg klarte å ?drakk/drikke en øl I managed to ?drank/drink a beer

2 Verbal feature agreement as a restructuring effect Wiklund (2007) analyzes verbal feature agreement in Swedish as a restructuring (aka reanalysis, clause union) effect. Restructuring is the phenomenon that two or more predicates behave as one predicate with


respect to clause bounded syntactic operations. In typical minimalist approaches, restructuring verbs are either functional, or they select “small” complements, such as VP. In either case, the verbs involved share functional projections. (8) [CP…[FP…[VP Vrestr [VP V]]]] (Lexical) restructuring (9) [CP…[FP…[VP V [CP…[FP…[VP V]]]]]] Non-restructuring On Wiklund’s approach, restructuring infinitives may be as big as CP. However, they are underspecified for TMA features and inherit feature specifications from the matrix clause. (10) [CP C[val]i [TP T[val]j [AspP Asp[val]k [VP V [CP C[]i [TP T[]j [AspP Asp[]k [VP V]]]]]]]] (Adapted from Wiklund 2007:158) The long passive (aka long object movement) is a cross-linguistically common restructuring phenomenon, and exists also in Norwegian (Lødrup 2014b; see Wurmbrand 2014 for some typology). (11) at vaskemaskinen må huskes å slås på that washing-machine must remember.INF.PASS to turn.INF.PASS on «that you should remember to turn the washing machine on” (Lødrup 2014b:368) The Norwegian long passive offers support for the hypothesis that verbal feature agreement is connected to restructuring. It optionally – or even preferably – involves agreement in passive features, as in (11). The long passive independently calls for a monoclausal analysis. As shown by Niño (1997), Sells (2004) and Lødrup (2014b), the LFG analysis of restructuring (as in Alsina 1996) nicely accounts for verbal feature agreement in monoclausal structures. The relevant level of representation is functional structure (f-structure). F-structure contains information about grammatical relations and morphosyntactic features. The simplified f-structure for the long passive in (12) is given in (13): (12) Dette forsøkes å gjøres this try.INF.PASS to do.INF.PASS

(Lødrup 2014b:380)

' try - do  (SUBJ) ' PRED SUBJ  PRED ' this'   (13) PASS     TENSE PRES  The PRED values of the two verbs combine: The argument structure of do fills a slot in the argument structure of try, yielding a complex predicate taking one set of grammatical functions (Alsina 1996:201203). Whatever morphosyntactic features the verbs carry must unify. Under a monoclausal analysis, sentence (1) gets the simplified f-structure in (14). All the verbs involved correspond to one single f-structure with the feature [TENS PRESPERF] (or whatever features are associated with the present perfect).

' forget - pack  (SUBJ)(OBJ) ' PRED SUBJ  PRED ' I '   (14) OBJ PRED ' positive attitude...'    TENSE PRESPERF  2

3 The perfect participle as a marker of irrealis mood It has been observed that PPC agreement is more common in counterfactual contexts (Falk and Torp 1900:213-214; Sandøy 1991), i.e. when the higher PPC expresses irrealis mood. In these cases the higher PPC is licensed by a past tense modal verb or the past perfect auxiliary hadde. (15) Jeg ville også prøvd å gitt de tre unitene dine Rune of Stoicism I would also tried to given the three units yours Rune of Stoicism “I would also try to give your three units Rune of Stoicism (NoWaC) More interestingly, Sandøy (1991) claims that counterfactual PPC agreement is possible with a larger set of first verbs. Based on an acceptability survey, he gives the following judgements: (16) Dei har bestemt seg for å *gått/gå heim i kveld they have decided self for to *gone/go home in night Intended: «They have decided to go home tonight» (Sandøy 1991:256) (17) Dei hadde nok bestemt seg for å gått/gå heim att viss dei hadde kjent til denne regelen they had PART decided self for to gone/go home PART if they had known to this rule “They would probably have decided to go home if they had known about this rule” (Sandøy 1991:258) Julien (2003:148) observes the same pattern for planlegge ‘plan’. Sandøy attributes the unacceptability of (16) to an aspectual conflict: The event denoted by the complement of decide must be understood as being unrealized at the time of the matrix event (in this particular example the present perfect seems to force an interpretation in which it is unrealized even at speech time). According to him, (17) is fine because the aspectual distinction between the perfect participle and the infinitive is neutralized in irrealis contexts. So far we have only looked at PPC agreement in (what would under a biclausal analysis be) control sentences, but we do find it in other constructions. In (18) the extraposed infinitive agrees with the matrix predicate. This seems to be restricted to counterfactuals for a number of speakers. In the NoWaC corpus we find only a handful of examples like (19), which sounds bad to me. (18) Det hadde vært gøy å festet og sånn med henne, tror jeg It had been fun to partied and stuff with her think I «I think it would be fun to party and stuff with her» (NoWaC) (19) …det har vært koselig å vært på jobb It has been nice to been on work «It has been nice to be at work» (NoWaC) The above-said distribution is one of Julien’s (2003) motivations for rejecting an agreement analysis of double PPCs in counterfactual contexts. She proposes that the second PPC carries its own irrealis mood feature. This view is shared by Sæbø (2009) and Eide (2011). On this approach, the double PPC is treated similarly to counterfactual PPCs after past tense modal verbs, as in (20) (which belongs to the standard language): the PPC can express irrealis mood given an appropriate licensing context. Sæbø (2009) observes that for some speakers, this context is not restricted to a past modal or a higher PPC, cf. (21). (20) Du burde dratt i går you should left yesterday “You should have left yesterday” (21) Skulle ønske jeg hadde mot til å gjort noe nytt should wish I had courage to to done something new «I wish I had the courage to do something new” (Sæbø 2009:5)


4 General properties of restructuring verbs According to Wurmbrand (2001:7), it is commonly accepted that there is a crosslinguistic core class of restructuring predicates. These are modal verbs, motion verbs, aspectual verbs and causatives. Other common restructuring predicates can be classified as implicative (e.g. manage) and irrealis (e.g. try). Propositional verbs (e.g. claim) and factive verbs (e.g. regret) are never restructuring. Karttunen (1971:340) said the following about implicative verbs: “An asserted main sentence with one of these verbs as predicate commits the speaker to an implied proposition which consists of the complement sentence as augmented by the tense and other modifiers of the main sentence.” Thus: (22) Tom managed to overcook the chicken yesterday -> Tom overcooked the chicken yesterday Negative implicatives behave essentially the same, but “incorporate negation” (Karttunen 1971:352): (23) Tom forgot to fry the chicken yesterday -> Tom did not fry the chicken yesterday Complements of irrealis verbs are understood as unrealized at the matrix time. Several researches assume that restructuring infinitives lack an independent internal tense specification (see Wurmbrand 2001:7 and references therein). That is, the time reference of the restructuring infinitive is fully determined by the matrix time. (24) Hans hat versucht (*morgen) zu verreisen John has tried (*tomorrow) to go-on-a-trip (Wurmbrand 2001:73) (25) Hans hat beschlossen (morgen) zu verreisen John has decided (tomorrow) to go-on-a-trip (Wurmbrand 2001:73) Some German verbs that allow the long passive, such as try, consistently disallow their complement to have independent time reference. Others (such as erlauben ‘allow’) can take tensed infinitival complements, but only in the absence of restructuring effects like the long passive (Wurmbrand 2007:82). (26) Hans erlaubte dem Kind (?morgen) einen Kuchen zu essen John allowed the child-DAT (?tomorrow) a cake to eat ‘John allowed the child to eat a cake (tomorrow)’ (Wurmbrand 2001:80) (27) Dem Kind wurden nur Kekse (*morgen) zu essen erlaubt the child-DAT were only cookies (*tomorrow) to eat allowed ‘The child was only allowed to eat cookies (tomorrow)’ (Wurmbrand 2001:82) With some verbs, such as planen ‘plan’ and beschließen ‘decide’, the complement is interpreted as situated temporally after the matrix time, regardless of temporal adverbs. These disallow the long passive. (28) *weil der Wagen zu reparieren geplant wurde since the.NOM car to repair planned was ‘since they planned to repair the car’ (Wurmbrand 2014:280) The availability of long passive and the absence of internally specified tense are both due to the size of the restructuring infinitive. There is no vP, so the internal argument of the infinitive must receive case from the matrix clause. And there is no T-projection.


We do, however, find long passives with Norwegian planlegge ‘plan’. These sound fine to me: (29) … hvor giftig mudder nå planlegges å tas opp… where toxic sludge now plan.INF.PASS to take.INF.PASS up ‘where they now plan to take up toxic sludge’ (30) …når hele 10 lensmannskontor nå planlegges å legges ned when whole 10 sheriff-offices now plan.INF.PASS to shut.INF.PASS down ‘when they now plan to shut down 10 sheriff departments’



5 The NoWaC data Most of the verbs found with PPC agreement in NoWaC disallow independent embedded tense. Several of these are normally implicative: klare ‘manage’, begynne ‘begin’, rekke ‘make (in time)’, greie ‘manage’, gidde ‘bother’, oppleve ‘go through’, orke ‘manage/bother’, pleie ‘usually do’, glemme ‘forget’, unngå ‘avoid’, bruke ‘usually do’, finne på ‘do’, la være ‘refrain from’, sørge for ‘make sure’, nøle med ‘hesitate’, vise seg ‘appear’. The raising-to-object verbs se ‘see’ and høre ‘hear’ possibly belong in this group too. The verb prøve ‘try’ has an implicative and a non-implicative use, and we find plenty of PPC agreement with both. (31) Har prøvd å tatt han med, men han kommer bare i veien have tried to taken him with, but he comes only in the-way ‘I have tried to bring him along, but he only gets in the way’ (32) Har prøvd å gjort han tam, lykkes nesten 1 gang! have tried to done him tame, succeeded almost 1 time ‘I have tried to make him tame, I almost succeeded once!’



Some of the verbs found with PPC agreement in NoWaC have an implicative use, but still allow complements to be understood as taking place after the matrix time in some context. These are tørre ‘dare’, velge ‘choose’, slippe ‘not have to’, nekte ‘refuse’, våge ‘dare’, la ‘let, make’ and få til ‘make’. In most of the actual NoWaC sentences, the natural interpretation is more or less implicative, as in (33). Sentence (34) exemplifies a constructed future oriented use of choose. (33) Jeg ser på jenter som fremdeles lever i sin egen lille boble I look at girls who still live in their own little bubble på tross av det store steget de har valgt å tatt in spite of the big leap they have chosen to taken (34) Jeg har allerede valgt å stemme blankt til høsten I have already chosen to vote blank to the-autumn


The verbs vurdere ‘consider’, anbefale ‘recommend’, ønske ‘wish’ and tenke (seg) take complements that are understood as future to the matrix time. With these, we find PPC agreement almost exclusively in counterfactual contexts. This is in line with Sandøy’s (1991) discoveries. (35) Jeg ville vurdert å byttet platen i butikken I would considered to swapped the-record in the-shop With three verbs, we find a very large amount of PPC agreement, and almost exclusively in counterfactual contexts. These are like ‘like’, tenke seg ‘would like’ and komme til ‘will’. These do not have relevant noncounterfactual counterparts. In all but two examples of PPC agreement with like, the verb expresses an attitude to the specific event denoted by the complement, and not a general property.


(36) NON-STATIVE – abundant in NoWaC Jeg skulle likt å hørt samtalene mellom … britiske militære ledere I should liked to heard the-conversations between British military leaders (NoWaC) (37) STATIVE – very scarce in NoWaC … hadde hun likt å gått på tur, så ville vi dratt på galopp tur… had she liked to gone on trip then would we gone on gallop trip ‘If she had liked to go on trips, then we would go on gallop trips’ (NoWaC) To the extent that non-stative like has a non-counterfactual counterpart, it is factive. Sentence (39) is possibly an example of PPC agreement with factive like. (38) Jeg har likt å høre samtalene mellom britiske militære ledere I have liked to hear the-conversations between British military leaders (Constructed) (39) Jeg har alltid likt å stått i mål… I have always liked to stood in goal ‘I have always liked to be the goalkeeper’ (NoWaC) In counterfactuals, tenke seg means something like ‘would like to’, and has no non-counterfactual counterpart. If acceptable, sentence (41) expresses intention. (40) kunne virkelig tenkt meg å sett den filmen jeg could really thought myself to seen that movie I ‘I would really like to see that movie» (NoWaC) (41) ?har tenkt meg å se den filmen jeg have thought myself to see that movie I “I intend to see that movie” (Constructed) Komme til signals future tense, and is not available in the present perfect. In cases of non-counterfactual PPC agreement with komme til, the verb expresses some non-agentive resultativity. (42) Vi hadde ikke kommet til å gjort hverandre lykkelig sånn som det gikk we had not come.PPC to to done each-other happy such that it went ‘We wouldn’t have made each other happy the way that it went’ (43) Forskerne har ofte kommet til å oppfattet seg som talsmenn for bestemte klienter the-scientists have often come.PPC to to perceived themselves as spokespersons for certain clients ‘Scientists have often come to perceive of themselves as spokespersons for certain clients’ (NoWaC)

6 The counterfactual effect It seems that what needs to be accounted for is the apparent unavailability of non-counterfactual PPC agreement with future oriented verbs like vurdere ‘consider’, as observed also by Sandøy (1991). One possibility is that the different distribution of counterfactual and non-counterfactual PPC agreement is due to cases that do not involve agreement. Problems: -

If a higher PPC is required, the second PPC can carry its own irrealis mood feature only in contexts in which it is redundant Counterfactual PPC also seems to be subject to lexical conditions Non-counterfactual PPC agreement has to be something completely different (the solution in Julien 2003 and Eide 2011 is to make the wide class of pseudocoordination verbs even wider, which explains nothing)


Given three assumptions, we can account for a pattern like that in (44)-(45) within the restructuring analysis. The judgements are representative for the NoWaC tendencies, and consistent with Sandøy’s findings (cf. (16) and (17) above). (44) *John har vurdert å dratt John has considered to left (45) John ville vurdert å dratt John would considered to left -

The complement of a verb like consider is specified for tense The f-structure that the main verb corresponds to in a construction ‘has considered’, contains a tense feature The f-structure that the main verb corresponds to in a construction ‘would considered’, does not contain a tense feature

The two last assumptions are common in the LFG literature. After Butt, Niño and Segond (1996), the perfect auxiliary is often considered to be a functional category with no predicational content. Thus, the perfect auxiliary and the main verb correspond to the same f-structure. Exactly which tense feature this contains is not of great importance here. (46) F-structure for har vurdert ‘has considered’

PRED ' consider'  TENSE PRESPERF    (47) Alternative f-structure for har vurdert ‘has considered’

PRED ' consider' TENSE PRES    ASP PERF  Modal verbs, on the other hand, are treated as (raising or control) predicates (Butt et al. 1999; Falk 2008). The Norwegian ville is more modal-like than its English cognate, which is often treated as a purely temporal auxiliary, cf. Dyvik (1999); Falk (2008). I assume that the counterfactual PPC does not carry a tense feature (although the modal verb might). (48) F-structure for ville vurdert ‘would considered’


' would'  PRED ' consider'   MOOD IRR   

Now, if PPC agreement is due to restructuring, we expect a pattern like that in (44)-(45). In (44), consider and leave cannot correspond to the same f-structure because they have conflicting tense values. In (45), there is no such conflict. Wiklund (2007:64-66,84-85) gives a similar explanation for the unavailability of tense agreement with future oriented verbs in Swedish (although the data are somewhat different). There remains a formal issue concerning the syntactic representation of embedded tense in a monoclausal structure. Following Andrews and Manning (1999), we can represent the a-structural embedding of the second verb by making it the value of the ARG attribute in f-structure. This gives us the subordination needed to specify lexically induced embedded tense.


(49) vurdere ‘consider’ (↑ARG TENSE)=FUT Following Sells (2004), we implement the ARG attribute in predicate composition Alsina (1996) style: information in the outer f-structure and information in ARG is shared. PRED values are subject to composition; all other values are subject to unification. (50) F-structures for ‘John has considered to left’ and ‘John would considered to left’ (ignoring the modal predicate)

  LCS    TERM S     ARG    SUBJ     TENSE  

' consider  θi , leave  θi  SUBJ 

' leave  θ ' LCS TERM S  SUBJ     TENSE FUT  PRED

' John'


  '                    

  LCS    TERM S     ARG    SUBJ     M OOD  

' consider  θi , leave  θi  SUBJ 

' leave  θ ' LCS TERM S  SUBJ     TENSE FUT  PRED

' John'


  '                    

7 Problems We find what appears to be verbal feature agreement between verbs in non-local relations. This is not limited to counterfactuals, cf. (51), and not even to perfect participles, cf. the imperative agreement in (52). One could argue that the preposition utan in (51) introduces a counterfactual context, and that its PPC complement expresses irrealis mood. However, the higher PPC is required, and this does not express irrealis mood. (51) Har de levd ei heil veke utan å kokt kaffi, karar? have you lived a whole week without to boiled coffee, guys? ‘Have you gone an entire week without making coffee, guys?’ (Sandøy 1991:256) (52) Lev ei heil veke utan å kok kaffe! live.IMP a whole week without to boil.IMP coffee! Go an entire week without making coffee! (Julien 2003:155) Such cases are beyond the scope of a restructuring analysis.

References Alsina, A. (1997) The Role of Argument Structure in Grammar: Evidence from Romance. Stanford, CA: CSLI Publications.


Andrews, A. D. & Manning, C. D. (1999) Complex Predicates and Information Spreading in LFG. Stanford, CA: CSLI Publications. Butt, M., Niño, M. & Segond, F. (1996) Multilingual Processing of Auxiliaries within LFG. In Gibbon, D. ed. Natural Language Processing and Speech Technology. Berlin: Mouton the Gruyter, 111-122. Butt, M., King, T. H., Niño, M. & Segond, F. (1999) A Grammar Writer’s Cookbook. Stanford, CA: CSLI Publications. Dyvik, H. (1999) The universality of f-structure: discovery or stipulation? The case of modals. In Butt, M. and King, T. H. eds. Proceedings of the LFG 99 Conference, University of Manchester. Eide, K. M. (2011) The Ghost of the Old Norse Subjunctive: the Norwegian Subjunctive Participle. Groninger Arbeiten zur germanistischen Linguistik, 53 (2), 1–28. Falk, H. & Torp, A. (1900) Dansk-norskens syntax i historisk fremstilling. Kristiania: Aschehoug. Falk, Y. (2008) Functional relations in the English auxiliary system. Linguistics, 46 (5), 861-889. Julien, M. (2003) Dobbelsupinum og irrealis modus. Norsk Lingvistisk Tidsskrift, 21, 135-161. Karttunen, L. (1971) Implicative verbs. Language, 47 (2), 340-358. Larsson, I. (2014) Double supine. Nordic Atlas of Language Structures (NALS) Journal, 1. Lødrup, H. (2014a) How can a verb agree with a verb? Reanalysis and pseudocoordination in Norwegian. In Butt, M. and King, T. H. eds. Proceedings of the LFG14 Conference. CSLI Publications. Lødrup, H. (2014b) Long passives in Norwegian: Evidence for complex passives. Nordic Journal of Linguistics, 37 (3), 367-391. Niño, M. (1997) The Multiple Expression of Inflectional Information and Grammatical Architecture. In Corblin, F., Godard, D. and Marandin, J. eds. Empirical Issues in Formal Syntax and Semantics. Bern: Peter Lang, 127-147. Sandøy, H. (1991) Attraksjon av supinum i færøysk og norsk. Danske folkemål, 33, 251-262. Sells, P. (2004) Syntactic Information and its Morphological Expression. In Sadler, L. and Spencer, A. eds. Projecting Morphology. Stanford, CA: CSLI Publications, 187-225. Sæbø, K. J. (2009) A Scandinavian Mood Recovered: The Semantics of the Surreal Supine. NORMS Workshop: Counterfactuality and auxiliaries in Nordic Languages, Dec 7-8, 2009. Teleman, U., Hellberg, S. & Andersson, E. (1999) Svenska Akademiens grammatik. Stockholm: Norstedts. Wiklund, A. (2007) The Syntax of Tenselessness. Berlin: Mouton de Gruyter. Wurmbrand, S. (2001) Infinitives. Restructuring and Clause structure. Berlin: Mouton de Gruyter. Wurmbrand, S. (2014) Restructuring across the World. In Veselovská, L. and Janebová, M. eds. Complex Visibles Out There. Proceedings of the Olomouc Linguistics Colloquium 2014. Oloumoc: Palacký University, 275-294.


Suggest Documents