Hostname: page-component-6898cfcb55-w5hgh Total loading time: 0 Render date: 2026-11-11T04:32:27.740Z Has data issue: false hasContentIssue false
  • English
  • Français

The interaction of modality and negation in Finnish1

Published online by Cambridge University Press:  24 February 2012

ROSE THOMAS
ROSE THOMAS*
Affiliation:
University of Westminster
*
Author's address: University of Westminster, Department of English, Linguistics and Cultural Studies, 32–38 Wells Street, London W1T 3UW, UKluonnotar1065@yahoo.co.uk
Rights & Permissions [Opens in a new window]

Abstract

In Finnish, negation is expressed via an auxiliary, and no other verb may occur above this auxiliary in the structure. This gives rise to a problem with respect to the modals of obligation and necessity, which take scope over negation yet appear below it. It is tempting to account for this in terms of LF-movement, but evidence suggests that there are in fact two modal phrases in Finnish, one above negation and the other below it, the higher of which encodes necessity/obligation. Evidence for the higher phrase comes from the negative imperative. Although the PF part of a verb in a negative sentence cannot move to the head of this higher phrase, the head itself is in the right position to take scope over negation. Thus, rather than attributing the scope properties of the modals to LF-movement, it will instead be argued that the LF-interpretable part of a head is merged precisely where it takes scope, and that the relation between the LF- and PF-interpretable parts of the modal is one of checking at a distance. Head-movement will be regarded solely as a PF phenomenon. It will be seen that the scope relations of the modals and the imperative mood can be accounted for under this hypothesis. Thus, Finnish provides evidence for a view of syntax which identifies syntactic structure largely with the LF-interpretable part of a sentence, and sees head movement as fundamentally a PF phenomenon. There are two morphological moods in Finnish, which seem to provide counter-examples to this hypothesis, which will be left as a problem for future research.

Information

Type
Research Article
Information
Journal of Linguistics , Volume 48 , Issue 3 , November 2012, pp. 653 - 684
Copyright
Copyright © Cambridge University Press 2012

1. Introduction

It has been accepted since the early 1990s that in Finnish, a negative phrase (NegP) must be projected above a tense phrase (TP), but below a phrase which licenses subject agreement on the verb, in order to account for the order of tense and agreement markers in negative sentences (Mitchell Reference Mitchell1991, Reference Mitchell1993, Reference Mitchell2006; Holmberg et al. Reference Holmberg and Nikanne1993). In affirmative sentences, agreement markers occur after tense markers, as the paradigm in (2) shows (1st and 2nd person nominative subjects can be omitted in Finnish; see Appendix for a list of abbreviations).

  1. (1)

    1. (a) (Minä) men-i-n. I.nom go-past-1sg ‘I went.’

    2. (b) (Sinä) men-i-t. you.nom go-past-2sg ‘You went.’

    3. (c) Hän men-i. s/he.nom go-past.3sg ‘S/he went.’

    4. (d) (Me) men-i-mme.we.nom go-past-1pl ‘We went.’

    5. (e) (Te) men-i-tte. you.pl.nom go-past-2pl ‘You went.’

    6. (f) He men-i-vät. they.nom go-past-3pl ‘They went.’

Negation in Finnish is expressed using an auxiliary, sometimes called the verb of negation, which has the stem form e-. This auxiliary is marked for agreement, but not for tense, which appears on the main verb. The main verb appears in its stem form to indicate the non-past tense, and in the active past participle for the past tense. Thus, there is a separation of tense and agreement markers in negative sentences. The sentences in (2) illustrate the pattern with reference to 1st person singular; the pattern is similar for the rest of the person/number paradigm.

  1. (2)

    1. (a) Minä e-n mene. I.nomneg-1sg go-pres ‘I don't/won't go.’

    2. (b) Minä e-n men-nyt. I.nomneg-1sg go-past.prt ‘I didn't go.’

The fact that only agreement suffixes occur on the verb of negation in a negative sentence, while tense is shown on the verb which follows it, indicates that the verb of negation is a head (Holmberg et al. Reference Holmberg and Nikanne1993), and projects a NegP which is located below a phrase which licenses agreement and above TP.

Similar facts obtain for mood in Finnish. Apart from the indicative, there are three morphological moods in Finnish: the imperative, the conditional, and the rather formal potential, which expresses epistemic probability. When verbs are inflected for the conditional and potential moods (the imperative will be considered in Section 3.1 below), they cannot take agreement markers in a negative sentence – once again these appear on the verb of negation, while the mood markers appear on the main verb, as the following examples show:

  1. (3)

    1. (a) Men-isi-n. go-cond-1sg ‘I would go.’

    2. (b) E-n men-isi. neg-1sg go-cond ‘I would not go.’

    3. (c) Men-ne-n. go-pot-1sg ‘I'll probably go.’

    4. (d) E-n men-ne. neg-1sg go-pot ‘I probably won't go.’

Tense and mood never occur on the same stem in Finnish, and so it can be inferred that both tense and the conditional and potential moods occur at the same position in the structure, a position that can be realised as either TP or MP (Mood Phrase), and which is immediately dominated by NegP. Generalising from NegP to a more general polarity phrase PolP (Holmberg Reference Holmberg2001), the following structure for the higher projections of the Finnish IP seems likely:

  1. (4)

FP stands for Finiteness Phrase. The reason why this phrase, rather than AGRsP, is placed above PolP will be explained in Section 3.2.

The main verb can never occur before the verb of negation in Finnish. This applies to all verbs, including those which correspond to English modals, such as those in (5).

  1. (5)

    1. (a) Sinun ei pitäisi men-nä. you.genneg.3sg should go-inf1 ‘You should not go.’

    2. (a′) *Sinun pitäisi ei men-nä you.gen should neg.3sg go-inf1

    3. (b) Puhee-n ei tule ol-la pitkä. speech-genneg.3sg must be-inf1 long‘The speech must not be long.’

    4. (b′) *Puhee-n tule ei pitkä. speech-gen must neg.3sg long

    5. (c) Tämä ei voi ol-la totta! this.nomneg.3sg can be-inf1 true!‘This cannot be true!’

    6. (c′) *Tämä voi ei ol-la totta! this.nom can neg.3sg be-inf1 true

    7. (d) Sinä e-t saa men-nä. you.nomneg-2sg may go-inf1 ‘You may not go.’

    8. (d′) *Sinä saa e-t men-nä. you.nom may neg-2sg go-inf1

This raises the question of how modals take scope over negation in such sentences. Since the phrase below PolP can host both tense and mood, it is reasonable to assume that this is where the modals are located in examples (5a–d). In the case of (5c, d) there is no problem with this, as the modal verb is c-commanded by negation, and scopes below it (their meanings are, respectively ‘it is not possible that this is true’ and ‘you are not permitted to go’). However, in sentences (5a, b), known as obligative constructions, the modal scopes over negation (the meanings are ‘it is advisable that you do not go’ and ‘it is obligatory that the speech is not long’) and yet in the syntax, it occurs below it. How, then, can scope relationships in obligatives be accounted for?

One possibility is LF-raising, with the LF-interpretable part of the sign raised and adjoined to a higher position, perhaps somewhere in CP. An alternative possibility is suggested by Cormack & Smith (Reference Cormack and Smith1996, Reference Cormack and Smith1997) who, on the basis of work on serial and quasi-serial constructions, propose that functional heads can be merged as what they call ‘split signs’. That is, they are not necessarily merged as Saussurian signs, containing both LF- and PF-interpretable information (Cormack & Smith Reference Cormack and Smith1997: 224) but instead, the LF- and PF-interpretable parts of a head may be separated in the course of a derivation, and merged in separate positions. Cormack & Smith (Reference Cormack and Smith1996, Reference Cormack and Smith1997) further propose that the LF-interpretable part of a sign is merged in the position in which it is actually interpreted, and that there is no LF-movement. Likewise, the PF-interpretable part of the item is merged where it is pronounced. This claim is made only for functional heads. It is not clear how the proposals outlined below apply to those items which undergo A-movement. In this approach, functional heads such as Tense and Mood host only the LF-interpretable parts of these items, but have what Cormack & Smith call Infl heads associated with them, which is where the PF-interpretable parts of the signs are merged. These Infl heads are not necessarily adjacent to their associated functional heads, although this is the preferred option (Cormack & Smith Reference Cormack and Smith1998). Stripped of PF forms, it is claimed, a syntactic structure should be identical to a sentence's LF structure. As movement is eschewed, the relationship between the semantic and morphophonological parts of a sign is ensured by checking, which may be ‘at a distance’, i.e. long distance checking, and not necessarily in a Spec-Head or Head-adjoined configuration, although of course this is possible.

Long-distance checking is a concept which is already present in the notion of Agree (Chomsky Reference Chomsky, Martin, Michaels and Uriagereka2000, Reference Chomsky and Kenstowicz2001). Given long-distance checking, agreement need not involve movement, but can be accomplished merely by a matching of features between checker and checkee. A similar hypothesis has been proposed by Simpson (Reference Simpson2000) with respect to wh-in situ constructions. Rather than accounting for them by LF-movement, Simpson suggests that the wh-in situ expression checks its wh-feature simply by being within the scope of a +wh complementizer. The mechanics of checking at a distance are therefore identical to those of binding – the functional head merely needs to c-command the element with matching features, within a checking domain which, like binding domains, may vary cross-linguistically (Andrew Simpson, p.c.).

Applying this to modality, it is suggested (Cormack & Smith Reference Cormack and Smith1998, Reference Cormack, Smith, Barbiers, Beukema and van der Wurff2002) that the LF-interpretable parts of modals are merged exactly where they take scope. The English modals of obligation/necessity, for instance, take scope over negation, and indeed their PF forms appear in a pre-negation position. (It has often been noted, for example in McCloskey (Reference McCloskey and Haegeman1997), that cross-linguistically modals tend to take scope in their ‘surface’ positions.)

  1. (6) You should/must not go. OBL NOT

    (‘It is obligatory that you do not go.’)

The modals of possibility/permission, however, usually take scope below negation. May allows for ambiguity, while can appears in PF above negation:Footnote 2

  1. (7) Johnny may not go out. POSS NOT

    (‘It's possible Johnny won't go out/Johnny is not permitted to go out.’) NOT PER

  1. (8) Johnny can't eat nuts. NOT POSS

    (‘It's not possible for Johnny to eat nuts/Johnny is not permitted to eat nuts.’) NOT PER

Thus, Cormack & Smith (Reference Cormack and Smith1998, Reference Cormack, Smith, Barbiers, Beukema and van der Wurff2002) propose two modal phrases, one above NegP which hosts the necessity/obligation modals, and one below NegP which hosts the possibility/permission modals (the distinction is not between epistemic and deontic modality). It is of course the LF-interpretable parts of the modals which are merged in these positions (while their PF forms may appear elsewhere, as evinced by can). If this is correct, we seem to have a possible solution to the problem of scope relations in Finnish.

The existence of two such phrases is in fact supported by evidence from Finnish. This paper will argue that the correct way to approach the interaction of the Finnish modals with negation is via a view of syntactic structure in which it is identical to LF, which consequently allows us to dispense with LF-movement – a view which obviously has implications beyond Finnish. I will, however, dispense with Infl heads. There appears to be no reason why the PF part of a sign should not be merged in the same position as its associated LF part unless there is something to prevent this. This much is acknowledged in Cormack & Smith (Reference Cormack and Smith1998), where it is admitted that the preferred option is for signs to be unsplit. If this is so, then why should signs be split in some cases and not in others? This seems to be more readily explainable in terms of the idea that head-movement at least is a wholly phonological process (Chomsky Reference Chomsky, Martin, Michaels and Uriagereka2000, Reference Chomsky and Kenstowicz2001; Boeckx & Stjepanović Reference Boeckx and Stjepanović2001), in which the PF parts of signs move to the next highest functional head and are adjoined to it until something occurs to block this process – in Finnish, the merging of the PF part of the verb of negation, which prevents all further movement.

The paper is organized as follows. Section 2 gives an account of Finnish modal auxiliaries and their interaction with negation. Section 3 provides evidence for the existence of a higher modal phrase, and also discusses some other aspects of Finnish IP structure that have relevance to the problem under discussion. Section 4 considers how the hypothesis of two modal phrases can account for the scope relations which occur in Finnish by the hypothesis of long-distance checking. Section 5 considers a problem posed for the hypothesis by the conditional and potential moods, and Section 6 concludes and suggests future research.

Information on the frequency of occurrence of certain ways of expressing modality is taken from Kangasniemi (Reference Kangasniemi1992), which in turn largely uses data from the Oulu Corpus of spoken and written Finnish, collected by the Department of Finnish and Lappish at the University of Oulu (http://www.csc.fi/kielipankki/iulkaisut/opas/c623.phtml#AEN625), as well as examples from newspapers and magazines, and data elicited from informants.

2. Modality in finnish

This section will give an overview of the expression of modality by modal verbs in Finnish. Sections 2.1–2.5 will show how epistemic and deontic modality are expressed, and Section 2.6 will deal with their interactions with negation.

2.1 Finnish modal verbs

Fourteen verbs in Finnish have been identified as expressing modality (Kangasniemi Reference Kangasniemi1992: 291). For convenience, I will deal mostly with the following, which are given with their nearest English equivalents: voida ‘can, may, be able, be possible’, saada ‘may, must’, täytyä ‘must’, pitää ‘must, have to’, tulla ‘must’, tarvita ‘need, must’ (in negative sentences), and the obligative structure ole+V+(t)tAvA. Other modals will not be dealt with in the main body of the paper.Footnote 3 The structure ole+V+(t)tAvA consists of the verb olla ‘be’, which has the stem form ole-, and the passive present participle. The passive present participle is formed by suffixing the passive suffix -(t)tA- to a verb (its precise form depends on the form of the verb), which is then followed by the present participle suffix -vA. (Due to the presence of vowel harmony in Finnish, the vowel in these suffixes may be either the back vowel a or the front vowel ä, shown by A.) A literal translation of this structure into English might be: ‘is to be V+past participle’.

Also, for convenience, I will largely omit considerations of dynamic modality, and concentrate on deontic and epistemic modality as expressed by the verbs referred to above. These verbs have various properties which distinguish them from English modals. Saada and voida take nominative subjects and are inflected for subject agreement. Täytyä, pitää, tulla, and tarvita have genitive subjects and take a default 3rd person singular agreement. The verbs can also be inflected for tense and mood. They also have non-finite forms, though such forms are rarely used. Voida is the verb most likely to appear in non-finite forms (Kangasniemi Reference Kangasniemi1992: 294–306). The examples in (9) illustrate these properties, using the verbs saada, voida and täytyä.

  1. (9)

    1. (a) Saa-t laina-ta kirja-n. may-2sg borrow-inf1 book-acc ‘You may borrow the book.’

    2. (b) Te voi-tte maksa-a kerran kuu-ssa. you.pl.nom can-2pl pay-inf1 once month-ine ‘You can pay once a month.’

    3. (c) Peka-n täyty-i lähte-ä työ-hön. Pekka-gen must-past leave-inf1 work-ill ‘Pekka had to leave for work.’

    4. (d) Häne-n täyty-isi jo ol-la kotona. s/he-gen must-cond already be-inf1 at.home ‘S/he should already be at home.’

    5. (e) Voi-da-kse-ni teh-dä päätökse-n tarvitse-n aika-a. can-inf1-trans-1sg.poss make-inf1 decision-acc need-1sg time-partFootnote 4 ‘For me to be able to make the decision, I need time.’

The use of the conditional on täytyä, pitää and tulla gives them the meaning of ‘should’.

In addition, some of the above are used as main verbs: tulla ‘come’, pitää ‘hold’, tarvita ‘need’, and saada ‘obtain, get’. In this usage, they behave exactly like all other main verbs, being inflected for person and having nominative subjects. All the verbs illustrated above take a first infinitive complement. The first infinitive corresponds to the English infinitive to V. Other structures known as infinitives need not concern us.Footnote 5

2.2 Epistemic possibility

This section will discuss the expression of epistemic possibility in Finnish. The most common way of expressing epistemic possibility in Finnish is via the use of the verb voida. This verb can also be used to express dynamic modality, and thus is close in meaning to the English can. As in English, it can be difficult to distinguish between the two meanings when the subject is 1st or 2nd person, so I will use sentences with inanimate 3rd person subjects to illustrate its epistemic uses.

  1. (10)

    1. (a) Täma tilanne voi ol-la vaarallinen. this.nom situation.nom can.3sg be-inf1 dangerous ‘This situation may be dangerous.’

    2. (b) Lumi voi pudo-ta joskus syyskuu-ssa. snow.nom can.3sg fall-inf1 sometimes September-ine ‘Snow can fall sometimes in September.’

2.3 Deontic possibility

We will now consider the expression of deontic possibility, or permission. This can be expressed by voida, but more common is saada.Footnote 6

  1. (11)

    1. (a) Saa-t men-nä. may-2sg go-inf1 ‘You may go.’

    2. (b) Saa-mme-ko selittä-ä. may-1pl.q explain-inf1 ‘May we explain?’

2.4 Epistemic necessity

Epistemic necessity is most commonly expressed by täytyä.

  1. (12)

    1. (a) Teidän täyty-y ol – la järjiltänne! you.gen must-3sg be-inf1 out.of.your.mind ‘You must be out of your mind!’

    2. (b) Ongelma-n täyty-y ol-la tietokonee-ssa. problem-gen must-3sg be-inf1 computer-ine ‘The problem must be in the computer.’

Pitää can also be used to express epistemic necessity, usually in the conditional mood, when it expresses ‘should’.

  1. (13)

    1. (a) Tämä-n lääkkee-n pitä-isi lievittä-ä vaiva-t. this-gen medicine-gen should-cond soothe-inf1 pain-acc.pl ‘This medicine should ease the pains.’

    2. (b) Se-n pitä-isi ol-la ihana-a.it-gen should-cond be-inf1 splendid-part ‘It should be splendid.’

2.5 Deontic necessity

The most common modal of deontic necessity, or obligation, is pitää.

  1. (14)

    1. (a) Meidän pitä-ä lähte-ä.we.gen must-3sg leave-inf1‘We must leave.’

    2. (b) Sinu-n pitä-ä tul-la mukaan.you-gen must-3sg come-inf1 along‘You must come along.’

As with epistemic necessity, when the verb is marked with the conditional mood, it translates ‘ought to, should’:

  1. (15)

    1. (a) Minun pitä-isi häve-tä.I.gen should-cond be.ashamed-inf1‘I ought to be ashamed.’

    2. (b) Ihmis-ten pitä-isi syö-dä paljon hedelm-i-ä.person-gen.pl should-cond eat-inf1 much fruit-pl.part ‘People should eat a lot of fruit.’

Less common as a verb of obligation, but with stronger force, is täytyä:

  1. (16)

    1. (a) Sinun täyty-y soitta-a tänä yönä.you.gen must-3sg ring-inf1 tonight‘You must ring tonight.’

    2. (b) Heidän täyty-y lähte-ä heti.they.gen must-3sg leave-inf1 at.once‘They must leave at once.’

Tulla can also be used to express obligation. It is somewhat formal, and has less force than pitää:

  1. (17)

    1. (a) Meidä-n tule-e muista-a että se on vaarallinen.we-gen must-3sg remember-inf1 that it be.3sg dangerous‘We must remember that it is dangerous.’

    2. (b) Hän tietä-ä, mitä häne-n tul-isi teh-dä.s/he.nom know-3sg what s/he-gen should-cond do-inf1‘S/he knows what s/he should do.’

However, the most common way of expressing obligation, and the one with the strongest force (Kangasniemi Reference Kangasniemi1992: 103), is a construction formed with the 3rd person singular of the the verb olla ‘be’, and the passive present participle, giving us the structure ole+V+(t)tAvA. (It is conventional to refer to the participles as ‘past’ or ‘present’, though the tense distinction in Finnish is between past and non-past, and so the present participles should more correctly be referred to as non-past.) The verb olla is the only auxiliary in Finnish and is also used as a perfect auxiliary (see Section 3.3 below). When so used it is inflected for tense and person like any other verb. In the construction currently under discussion it is always 3rd person singular, but can occur in either the present tense form on or the past tense form oli. In a negative sentence it of course occurs in the stem form ole after the third person verb of negation ei (see next section). This construction is considered to express a very strong obligation.

  1. (18)

    1. (a) Ilmari-n ol-itotel-ta-va.Ilmari-gen be-past obey-pass.pres.prt ‘Ilmari had to obey.’

    2. (b) Häne-n onol-ta-va niin.s/he-gen be.3sg be-pass.pres.prt so‘S/he must be so.’

Note that the subject is again in the genitive case.

2.6 Interaction with negation

We must now look at how the modals interact with negation. When voida and saada interact with negation, the negation has scope over the modal. The scope relations are shown to the right of the example:

  1. (19)

    1. (a) Tämä ei voi autta-a nyt. NOT POSS this.nomneg.3sg can help-infi now ‘This cannot help now.’

    2. (b) Tämä ei voi ol-la totta! NOT POSS this.nomneg.3sg can be-inf1 true ‘This can't be true!’

    3. (c) Sinä e-t saa men-nä. NOT PER you.nomneg-2sg may go-inf1‘You may not go.’

    4. (d) He ei-vät saa tul-la mukaan. NOT PER they.nomneg-3pl may come-inf1 along ‘They may not come along.’

Thus, the surface word order corresponds to the actual scopal interpretation for these modals. When possibility/permission is required to scope over negation, it can be expressed by using tarvita ‘need’. This verb is also used when negation scopes over obligation/necessity, as in English. There is of course a logical equivalence between NOT NEC and POSS NOT (if it is not necessary that something is the case then it is possible that it is not the case).

  1. (20) Se-n ei tarvitse ol-la hauska ajatus. NOT NEC it-genneg.3sg need be-inf1 pleasant thought POSS NOT ‘It need/may not be a pleasant thought.’

Of the necessity/obligation modals, täytyä is virtually unknown in a negative context (Kangasniemi Reference Kangasniemi1992: 124–125). In fact, the most common way of expressing ‘must not’ in Finnish is to use the verb saada with the verb of negation, in other words, to express it by using the same form of words as to express ‘not permitted’.

  1. (21)

    1. (a) Sinä e-t saa sanoa sellas-i-a. OBL NOT you.nomneg-2sg must say-inf1 such-pl.part ‘You must not say such things.’

    2. (b) Ihmise-t ei-vät saa juo-da liian paljon alkoholi-a. OBL NOT person-plneg-3pl must drink-inf1 too much alcohol-part ‘People must not drink too much alcohol.’

Again, there is a logical equivalence between OBL NOT and NOT PER (if it is obligatory not to do something, it is not permitted to do it). The use of saada makes it possible to express the concept of obligation not to do something without a conflict between surface word order and underlying scope.

It is, however, possible for pitää, tulla, and the structure ole+V+(t)tAvA to occur with the verb of negation, with the scope of negation inside that of the modal (the force of ole+V+(t)tAvA is weakened when it is negated, giving it a meaning closer to ‘should’ (Kangasniemi Reference Kangasniemi1992: 116–121)).

  1. (22)

    1. (a) Puhee-n ei tule ol-la pitkä. NEC NOT speech-genneg.3sg must be-inf1 long‘The speech must not be long.’

    2. (b) Peka-n ei pidä ajatel-la sellais-ta. OBL NOT Pekka-genneg.3sg must think-inf1 such-part ‘Pekka must not think such a thing.’

    3. (c) Aamu-lla ei olejuo-ta-va alkoholi-a. OBL NOT morning-adeneg.3sg be drink-pass.pres.prt alcohol-part ‘One should not drink alcohol in the morning.’Footnote 7

    4. (d) Täma-n ei pitä-isi satutta-a. NEC NOT this-genneg.3sg must-cond hurt-inf1‘This should not hurt.’

    5. (e) Täma-n työ-n ei pitä-isi ol-la vaikea-a. NEC NOT this-gen work-genneg.3sg must-cond be-inf1 difficult-part ‘This work should not be difficult.’

When negation takes scope over necessity/obligation, the verb tarvita ‘need’ is used, as in English.

  1. (23) Teidä-n ei tarvitse lähte-ä vielä. NOT OBL you.pl.genneg.3sg need leave-inf1 yet ‘You needn't leave yet.’

Thus, while there is an avoidance of structures in which a modal takes scope over negation, there are circumstances in which such structures are used, and these require explanation.

3. Evidence for two modal phrases and the structure of the Finnish IP

In order to account for the scopal relations in the examples in (22) above, we have two possibilities. Either the LF-interpretable parts of the modal are moved to some position higher than PolP, or there is a phrase above PolP which encodes necessity/obligation. Therefore we must now consider what phrases are projected above vP in Finnish (for convenience, we will use IP to refer to the region of all the functional projections above vP) and what evidence exists for a second Modal Phrase.

3.1 The imperative mood

As was shown in Section 1, the potential and conditional moods provide evidence for the existence of a mood phrase below PolP. However, when we come to consider the imperative mood, a different picture emerges. First, consider the affirmative imperative:

  1. (24)

    1. (a) Men-kää-mme.go-imp-1pl ‘Let us go.’

    2. (b) Mene!go.imp ‘Go!’

    3. (c) Men-köö-n.go-imp-3sg ‘Let him/her go.’

    4. (d) Men-kää(-tte).go-imp(-2pl)‘Go! (plural)’

    5. (e) Men-köö-t.go-imp-3pl ‘Let them go.’

Here we have a marker indicating imperative mood in all forms except 2nd person singular, and agreement markers occuring after this. There is no 1st person imperative form, and the agreement marker in the 2nd person plural imperative is optional, and usually omitted. The 3rd person forms occur mostly in the written language. As can be seen, the imperative mood is here behaving like the conditional and potential moods, in that the mood marker is closest to the stem, and the agreeement markers follow it.

However, when we look at the negative imperative, we find the following pattern:

  1. (25)

    1. (a) Äl-kää-mme men-kö.neg.imp-1pl go-prt ‘Let us not go.’

    2. (b) Älä mene!neg.imp go‘Don't go!’

    3. (c) Äl-köö-n men-kö.neg.imp-3sg go-prt ‘Let him/her not go.’

    4. (d) Äl-kää(-tte) men-kö.neg.imp(-2pl) go-prt ‘Don't go! (plural)’

    5. (e) Äl-köö-t men-kö.neg.imp-3pl go-prt ‘Let them not go.’

Assuming that the root älä- is to be understood as the imperative form of the verb of negation, here the mood is marked on that verb, rather than on the main verb. (The suffix -kO on the main verb has a role which is not entirely clear, and is perhaps best interpreted as a kind of participle; see Mitchell Reference Mitchell1993. The capital letter indicates that it can be either a front or back vowel, o or ö, due to vowel harmony.) This is quite unlike the situation that occurs with the declarative verb of negation, and the negative conditional and potential moods. It is difficult to see how this can be accounted for other than by hypothesising that PolP (here realised as NegP) is below a phrase encoding imperative mood, which is itself below FP. This has been noted by Mitchell (Reference Mitchell1993), who suggests that a phrase encoding obligation, which she calls OblP, is projected above PolP while the phrase below it, which hosts the conditional and potential moods, should be called Hypothetical Phrase (HypP), as both of these moods express hypotheticalness.

Such a scenario would make sense given the scope relation that exists between the imperative mood and negation. The imperative mood expresses directive deontic modality (Kangasniemi Reference Kangasniemi1992: 97; Palmer Reference Palmer2001: 80), and thus should scope over negation in a negative sentence. The scope relations in ‘don't go’, for example are OBL NOT (go).

The hypothesis that two modal phrases, one encoding obligation/necessity and the other encoding possibility/permission, are projected on either side of a PolP has also been proposed by Cormack & Smith (Reference Cormack and Smith1998, Reference Cormack, Smith, Barbiers, Beukema and van der Wurff2002) on the basis of the scope relations of the English modals. Taking mood to be a morphological expression of modality, we have seen that the behaviour of the Finnish negative imperative provides support for such a higher phrase, encoding obligation at least, and the conditional and potential moods provide support for one below it, encoding at least possibility. Since the imperative expresses deontic modality, and the conditional and potential express epistemic modality, this might make it appear at first sight as though the distinction in Finnish is one of epistemic/deontic modality rather than one of necessity/obligation and possibility/permission. However, as we have seen in examples (19c, d) above, the ‘permission’ modal saada does not scope over negation, and appears in the same position as the ‘possibility’ modal voida, which indicates that the lower phrase does indeed encode possibility/permission rather than epistemic modality per se. Similarly, in (22d, e) above we see that epistemic necessity, like deontic necessity, scopes over negation. Thus, we will identify OblP and HypP with Cormack & Smith's two modal phrases and henceforth call them Mod1P (corresponding to OblP, the phrase which immediately dominates PolP), and Mod2P (corresponding to HypP, the phrase below it).Footnote 8

3.2 Finiteness Phrase and the Finiteness head

We should now consider the phrase which dominates Mod1P. This has considerable importance because if checking at a distance is indeed the mechanism by which the modals take scope, there is evidence that it is constrained within certain checking domains (Simpson Reference Simpson2000). In Finnish, the checking domain appears to be the finite clause, as will be shown in Section 4.1 below, and thus we must establish the position of finiteness within a clause and what its precise role is.

Given the separation of agreement and tense markers in negative sentences, it has been suggested that the phase which dominates PolP is a subject agreement phrase, AGRsP (Mitchell Reference Mitchell1991). However, it has also been argued that heads like AGRs, which appear to have no interpretable features but exist solely to satisfy the EPP, should be eliminated (Chomsky Reference Chomsky1995: 349–335). Instead of having an AGRs head, it seems simpler to account for agreement by the operation Agree (Chomsky Reference Chomsky and Kenstowicz2001: 3; Boeckx Reference Boeckx2008: 96), a relation which holds between two heads α and β, where α has interpretable agreement features and β has uninterpretable ones which delete under the operation. It is clearly preferable to eliminate heads with non-interpretable features, and yet if there is no head above PolP in Finnish, there is a problem. Subjects cannot be in (Spec, T) in Finnish, as they would appear below the verb of negation in negative sentences if this were so. The examples in (26) illustrate that this is not the case, regardless of whether the subject has nominative or genitive case:

  1. (26)

    1. (a) Minä e-n laula.I.nomneg-1sg sing ‘I won't sing.’

    2. (b) Sinu-n ei pitä-isi teh-dä si-tä.you-genneg.3sg must-cond do-inf1 that-part ‘You should not do that.’

Therefore there must be a position for subjects which is at least above PolP. To account for (26a) it must be the landing site for the nominative subject, and its head must host agreement markers, to which verbs are adjoined. As we have seen, agreement markers in the negative imperative occur after mood markers, hence this phrase must also dominate Mod1P (when this is present). It has been argued (Holmberg et al. Reference Holmberg and Nikanne1993; Holmberg & Nikanne Reference Holmberg, Nikanne and Svenonius2002) that the projection in question is a Finiteness Phrase (FP). This hypothesis is based on the fact that in all finite clauses in Finnish the verb contains a suffix in the same place as the agreement markers as its outermost marker, whether there is explicit agreement or not. Examples are the regular agreement suffixes, seen in (1a–f) above, the default 3rd person agreement seen in both affirmative and negative obligative constructions, and the outermost marker of an affirmative passive verb, formed by replicating the final vowel of a preceding suffix and adding -n, which occurs in the same position as the ordinary agreement markers. An example with the past tense marker -i- is shown in (27).

  1. (27) Minut vie-t-i-in.I.acc take-pass.past.f ‘I was taken away.’

Holmberg & Nikanne (Reference Holmberg, Nikanne and Svenonius2002) propose that the specifier of this phrase can host topics in general, including both nominative and non-nominative subjects – nominative when there is regular agreement on the verb, and non-nominative when either default third person agreement is present (as in the genitive subjects of obligative constructions) or the passive suffix is present (the passives shown above have accusative subjects, which will not be considered further).Footnote 9 As we have seen, the subjects of obligative constructions are genitive. With regard to such genitive subjects, it has been argued (Vainikka Reference Vainikka, Holmberg and Nikanne1993) that the genitive case is a structural default case, able to occur in the specifier of almost any projection in Finnish.

One might ask why finiteness should be able to host agreement markers, and why its specifier should be a position for subjects. This leads us to a consideration of what finiteness actually is. The distinction between finite and non-finite clauses is not solely a matter of the fact that a finite clause contains a tensed verb, if tense expresses the relation between the Reichenbachian speech time and event time (Reichenbach Reference Reichenbach1947). It is entirely possible for a non-finite clause to express an independent event time, as in example (28):

  1. (28) I remembered today meeting him last month at the office.

Here, the event of meeting must occur before the event of remembering. However, in the matrix clause, which contains a past tense verb, the event time (which in this case is simultaneous with the Reichenbachian reference time, as the verb is in the simple past tense) is independently related to the speech time, the speaker's ‘now’. In the subordinate clause the second event time is not, but is dependent on another point in time, namely the event time of the matrix clause – we understand the event of meeting to be in the past with respect to the event of remembering. Intuitively, it seems that a finite clause encodes an independent relationship between speech time and event/reference time while a non-finite clause does not (there is no independent speech time in the subordinate clause in (28) above). This has been expressed in terms of ‘absolute’ and ‘relative’ tense by Comrie (Reference Comrie1985), who points out that finite clauses in English express ‘absolute’ tense, that is, the event time of the clause is in a direct relation to the speech time, while non-finite clauses express ‘relative’ tense, that is, the event time is not related directly to speech time, but to some other temporal point.

Such an interpretation of finiteness has also been proposed by Holmberg & Platzack (Reference Holmberg and Platzack1995), who point out that the main role of the finite verb form appears to be to relate a predication to the time of utterance (speech time), and by Bianchi (Reference Bianchi, Gueron and Tasmovski2003), who goes further and suggests that finiteness encodes the logophoric centre of the clause, the actual speech event itself, which includes speech time, participants in the event, and attitudes of the participants to the event. In short, speech time, person, and epistemic modality can all be encoded in a finiteness head. A finite clause anchors participants, modality and event time to an independent speech event encoded by the finiteness head, while a non-finite clause has a logophoric centre which is anaphoric to some aspect of the matrix clause. For example, in (28) above, the logophoric centre of the subordinate clause is not an actual speech event, but a mental event occurring at the time of remembering. Thus, the temporal logophoric centre of the non-finite clause is the event time of the matrix clause.

On the basis of empirical evidence, finiteness is known to be a complex concept which, cross-linguistically, does not seem to be reducible to a single functional category (Cristofaro Reference Cristofaro and Nikolaeva2007). It is well known that both tense and agreement features can occur on complementizers which are located under finiteness (Rizzi Reference Rizzi and Haegeman1997); for example, tensed complementizers under finiteness are found in Irish (McCloskey Reference McCloskey1996; Adger Reference Adger and Nikolaeva2007), and complementizers are marked with agreement in West Flemish (Haegemann Reference Haegeman1992). Using evidence from agreement paradigms in Turkish and related languages, Aygen (Reference Aygen and Boekcx2006) suggests that epistemic modality too is located in the finiteness head. The characterization of finiteness suggested by Bianchi (Reference Bianchi, Gueron and Tasmovski2003) makes it possible to see why these things occur – they are all aspects of the speech event. We can see now why agreement markers would occur under finiteness – this is where the participants in the speech event are encoded.

Thus, I will assume that FP as proposed by Holmberg et al. (Reference Holmberg and Nikanne1993) is the phrase which is projected above Mod1P, and which hosts both nominative and genitive subjects.Footnote 10

3.3 The lower levels of IP

The lower levels of the Finnish IP are important in relation to the construction ole+V+(t)tAvA. A brief account of these levels will be given here.

The passive suffix is always closest to the verb stem, and thus the phrase which hosts it must be the phrase above VP (we will leave out vP for convenience). Below T/Mod2P there is an auxiliary phrase (AuxP), which hosts the perfect auxiliary olla ‘be’. This, with a verb in the active past participle, forms the perfect aspect in Finnish. The auxiliary is tensed, and thus must raise as high as TP. The example in (29a) with the structure in (29b) illustrates all these projections except for the passive.

  1. (29)

    1. (a) Pekka ei ol-lut osta-nut kirja-a.Pekka.nom neg.3sg be-past.prt buy-past.prt book-part ‘Pekka had not bought the/a book.’

    2. (b) [FP Pekka [F ei [PolP [Polei [TP [T ollut [AuxP [Auxollut [PrtP [Prt ostanut

      [VP [Vostanut [NP kirjaa]]]]]]]]]]]]]

The participle ostanut in (29a) cannot be in TP, as this position must be occupied by ollut (see Section 1 above), nor can it be in AuxP, where ollut is merged, and so it must be hosted in another phrase below AuxP, called in Holmberg et al. (Reference Holmberg and Nikanne1993) a lower tense phrase, but called here a participle phrase (PrtP), as it does in fact host the past and present participles. Like T, its head is marked either+or – PAST, depending on whether it is occupied by the past or present participle. Evidence for this comes from a type of subordinate clause known as a participial construction (to be considered in more detail in Section 4.1), in which the presence of a participle alone suffices to make the clause tensed. We have also seen that the past participle alone carries tense in negative sentences (see example (2b) above). It need not be surprising that there are two heads in a structure which can be marked for tense, as there are in fact three points in time to be considered in any tensed sentence, speech time, event time and reference time (Reichenbach Reference Reichenbach1947). In a simple tensed sentence, event and reference time are identical, therefore only one tense-carrying head is necessary, but in a sentence like (29) above, which is pluperfect, event time precedes reference time. Thus, two tense-carrying heads are necessary, with reference time shown by the past participle on the auxiliary, and event time by the participle on the main verb.

PrtP, then, is where we find the present passive participle in the structure ole+V+(t)tAvA, as in example (30).

  1. (30) Sinu-n on lähde-ttä-vä.you-gen be.3sg leave-pass.pres.prt ‘You must leave.’

Thus, for our purposes we shall assume the following structure for the Finnish IP (leaving out specifiers):

  1. (31)

This can be compared with the structure in Holmberg et al. (Reference Holmberg and Nikanne1993: 180), which differs from the above only in not including Mod1P and in having NegP instead of a more generalized PolP.

Not all of these phrases will be projected for every sentence of course, as many of them would be semantically anomalous in certain sentences (for example, Mod1P would be semantically anomalous in a conditional or potential sentence, and there would be no point in the projection of AuxP for a sentence which lacked perfect aspect). Holmberg et al. (Reference Holmberg and Nikanne1993) argue that FP and PrtP are the only necessary functional projections for a grammatical clause in Finnish.

4. Scope relations and checking at a distance

4.1 Theoretical assumptions

Having established the existence of two modal phrases, we might expect that those modal auxiliaries which express obligation or necessity take scope in Mod1P while those that express possibility or permission take scope in Mod2P. However, we know that the PF parts of the modals can never appear above the verb of negation. How then, do the obligative modals pitää, tulla and ole+V+(t)tAvA take scope?

One possibility is that their LF-interpretable part is raised to a scope position above negation. What might this be? Mod1P is the most likely candidate, but there seems to be no point in the LF-interpretable part of a sign raising to adjoin to a phrase whose head already expresses it. Indeed, all that is required for a scope relation to occur is that the LF-interpretable part of a sign should c-command the element it scopes over. Since Mod1 c-commands Pol in (31) above, the requisite scope relations are already present. There is no necessity for movement of the actual PF form of a sign to a scope position, in order for scope to be taken. As seen in Section 1 above, Cormack & Smith (Reference Cormack and Smith1998, Reference Cormack, Smith, Barbiers, Beukema and van der Wurff2002) have suggested that the LF-interpretable parts of modals are merged exactly where they take scope, removing the necessity for LF-movement entirely, with the relationship between the LF and PF parts of a sign ensured by checking at a distance. From a minimalist perspective, they argue, removing the necessity for LF-movement is a preferable state of affairs.

A similar hypothesis has been proposed by Simpson (Reference Simpson2000) with respect to wh-in situ constructions. Rather than accounting for them by LF-movement, Simpson suggests that the required relation is checking at a distance. In order for this to occur, the checked item merely needs to be within the scope of the checker, in other words, the checker needs only to c-command the checkee.

However, this hypothesis requires some restrictions to be placed on what are possible checking domains. Simpson (Reference Simpson2000: 70) proposes that cross-linguistically, there is evidence for three possible checking domains for wh-expressions: the sentence itself (Chinese), the tensed CP (Iraqi Arabic, Hindi) and a complementizer marked with the features +Q+wh, which forces movement of a wh-expression to its specifier (Romanian, Bulgarian). The last domain listed here is not really relevant to a consideration of modals, but either of the others could be the checking domain for Finnish modals. However, the tensed CP should more correctly be called the finite CP, as we have already seen that non-finite structures can be regarded as tensed, in that they contain an event time which is related to another point in time which just happens not to be an independent speech time. Henceforth, I will use the term finite CP.

There is evidence that the finite CP is the checking domain for Finnish modals. This comes from the non-finite structures known as participial constructions. Participial constructions express the same meaning as a finite CP headed by the complementizer että, ‘that’, and instead of a verb in the non-past or past tense, they use either a present or past active participle, while the nominative subject of the finite clause is put in the genitive case. The participle is also marked with the genitive case.Footnote 11 Thus, the että-clauses in (32a, b) are the equivalent of the participial constructions in (32c, d).

  1. (32)

    1. (a) He sano-vat että sinä lähde-t. they.nom say-3pl that you.nom leave-2sg ‘They say that you're leaving.’

    2. (b) Luule-n että he ovat men-neet.think-1sg that they. nom be.3pl go-past.prt ‘I think they have gone.’

    3. (c) He sano-vat sinu-n lähte-vä-n.they. nom say-3pl you-gen leave-pres.prt.gen ‘They say that you're leaving.’

    4. (d) Luule-n heidän lähte-nee-n.think-1sg they.gen leave-past.prt.gen ‘I think they have gone.’

These non-finite structures cannot be negative, nor can mood be present in them – there are no conditional or potential participles. However, it is possible for negation and mood, when they occur in a matrix clause, to scope over the proposition expressed by the participial construction, as in (33).

  1. (33) E-n toivo-isi häne-n lähte-vä-n.neg-1sg hope-cond he-gen leave-pres.prt.gen ‘I hoped he wouldn't leave.’

This example is taken from the story Vanhuus [Age] by the Finnish poet and writer Pentti Holappa, reprinted in Whitney (Reference Whitney1971: 29–43) and the context establishes its meaning. The writer is speaking about his son leaving home. In the example, although the verb of negation and conditional are in the matrix clause, they are interpreted as scoping over the subordinate clause. Thus, the non-finite clause must be transparent for scope. However, this is not the case in a finite clause. In an että clause, the above sentence would have to be as in (34).

  1. (34) Toivo-i-n että hän ei läht-isi.hope-past-1sg that he.nomneg.3sg leave-cond ‘I hoped he wouldn't leave.’

That is, the verb of negation must be inside the että clause to take scope over it. On the other hand, in (35), where the verb of negation is outside the että clause, it can only scope over the matrix clause, not the että clause.

  1. (35) E-n toivo-nut että hän läht-isi.neg-1sg hope-past.prt that he.nom leave-cond ‘I didn't hope he would leave.’

It appears, then, that in a finite CP, scope relations are confined to that CP. Since we are concerned with PF forms that enter into a checking relation with scope-taking heads, I will assume that the checking domain for long-distance checking in Finnish is the finite clause. Provided the checkee is in the same finite clause as the checker, long-distance checking can occur.

It is possible to go further. Burton-Roberts (Reference Burton-Roberts2009) suggests that the concept of syntactic objects as having a ‘double interface property’ is fundamentally flawed. That is, the idea that syntactic objects consist of a PF- and LF-interpretable part, which are then split apart at Spell-Out, leads to incoherence. He argues that if the two mutually uninterpretable parts of a syntactic object are split at the interface levels, then the object itself ceases to exist – it cannot, as the original syntactic object, be interpretable at either interface. If the interfaces are the only admissible levels of representation, then the very concepts of syntactic category and structure become inadmissable concepts. Thus, he rejects the idea of the double interface property and suggests that syntax is grounded solely in conceptual-intentional representations, and is, in fact, identical to the ‘language of thought’ (Fodor Reference Fodor2008). On this view, phonology does not figure in syntax at all. What this means in practice is that a syntactic structure, when it is generated, should be identical with what is usually seen as the LF representation of the sentence, as already suggested by Cormack & Smith (Reference Cormack and Smith1996, Reference Cormack and Smith1997). Where, then, does phonology fit in? Cormack & Smith (Reference Cormack and Smith1997: 232–234) hypothesise the existence of Infl heads at various places in the syntactic structure as positions where the PF-interpretable parts of signs are merged. But if we are not allowing phonology to play any role in syntax there is no reason why these should exist. There is, however, evidence that head-movement at least is a wholly phonological process as suggested by Chomsky (Reference Chomsky1995, Reference Chomsky, Martin, Michaels and Uriagereka2000, Reference Chomsky and Kenstowicz2001), who points out several problems with head-movement as a syntactic phenomenon, for example, the fact that it has no effect on interpretation and is never successive-cyclic (Chomsky Reference Chomsky and Kenstowicz2001: 37–38). Boeckx & Stjepanović (Reference Boeckx and Stjepanović2001) have provided further evidence for head-movement as a phonological process, based on a consideration of its interaction with object shift.

Let us assume, then, that both lexical items and functional heads are taken from the lexicon as {LF, PF} pairs, resulting in pairs such as {NOT, e-}, {PAST, -i-}, {OBL, pitä-}.Footnote 12 (The morph -i- is the past tense marker.) In the course of the derivation, the LF-interpretable parts are placed directly where they are interpreted. The head of NegP is the negative operator itself, the head of Mod1P is a necessity or obligation operator, and the head of TP is a tense operator. The phonological forms associated with these items are inserted under their appropriate heads where this is possible (in the case of pitää, for example, it cannot be directly inserted under Mod1 as it is a verb which can also carry tense and occur with the auxiliary olla ‘be’ and so must be inserted in a position where it can adjoin to tense, or to a participle). Movement, then, occurs solely as a phonological phenomenon. In the case of an agglutinative language like Finnish, it will be assumed that head-movement is a process of adjunction to suffixes (Baker Reference Baker1988). There are two circumstances in which this may not take place – either all necessary suffixes have been adjoined, or something occurs to prevent further adjunction. As an example of the first process, consider the third person singular imperative verb of negation, älköön. We will assume the structure for the Finnish IP given in (31) above. The imperative verb of negation itself, älä-, is inserted under Neg, then moves to adjoin to the imperative suffix -köö- under Mod1 (losing its final ä in the process) and finally adjoins to the agreement marker -n under F. As an example of the second, consider the indicative verb of negation e-. Once this is inserted under Neg, it blocks all further movement of any other head, and in consequence no verb can appear in PF above it. Only e- itself can move further, to adjoin to an agreement marker under F. Under such circumstances, the PF and LF parts of an item such as a modal verb of obligation must be separated, not because there are distinct Infl heads for the PF parts, but because the PF cannot move any further. In such a case, checking must be at a distance. We can now define long-distance checking:

  1. (36) Long-distance checking

    An LF-interpretable head X can check the features of its corresponding phonological form Y provided that

    1. (i) X c-commands Y, and

    2. (ii) Y is in the checking domain of X.

We can now apply these ideas to Finnish. In what follows, I will assume that only those projections that are necessary for a sentence's interpretation are projected, and also that the head below PolP (see diagram (31)) need not carry both tense and mood features simultaneously, but that these are projected as necessary. Thus, in an obligative construction which translates ‘must’, the phrase below PolP is TP – it has no features of hypothetical mood or possibility/permission, as these would be semantically anomalous. Since the syntactic structure is assumed to be identical with LF, the requirement that an LF structure should be well-formed suffices to prevent the projection of structures which are semantically meaningless. I also assume that the third person agreement which occurs on the verb of negation is default agreement, as in affirmative obligative constructions such as those shown in (14), (16) and (17) above, and does not involve movement of the verb to a higher head.

4.2 Modals of obligation

As we have seen, two modals of obligation, pitää and tulla, can occur in PF below the verb of negation while taking scope over it, as in examples (22a, b) above, repeated here in (37).

  1. (37)

    1. (a) Puhee-n ei tule ol-la pitkä. NEC NOTspeech-genneg.3sg must be-inf1 long‘The speech must not be long.’

    2. (b) Peka-n ei pidä ajatel-la sellais-ta. OBL NOTPekka-genneg.3sg must think-inf1 such-part ‘Pekka must not think such a thing.’

The structure hypothesised for such sentences is as shown in (38). The complements of the modals are placed under VP for convenience.

  1. (38)

Here negation is immediately dominated by Mod1P and so negation is in the scope of the obligation operator, as it should be. The phonological forms tule- and pidä-, which are both non-past, check their tense in TP and their obligative feature against Mod1 at a distance. The only intervening node is Pol which is not a possible checker, and hence the nearest checker for the modals is Mod1.

4.3 The expression of ‘should’

‘Should’ is expressed in Finnish by inflecting one of the obligative modals, usually pitää, with the conditional:

  1. (39) Sinu-n ei pitä-isi sano-a sellais-i-a. OBL NOT you-genneg.3sg must-cond say-inf1 such-pl.part ‘You should not say such things.’

This use of the conditional need not surprise us, since, as Palmer (Reference Palmer2001: 74) points out, ‘should’ is not simply a weaker version of ‘must’, but has a conditionality attached to it. ‘You should not do it’, means ‘you are obliged not to do it (if it is possible, given all other considerations)’. Thus, the structure for (39) is identical to that shown above.

Mod1P dominates PolP, giving us the scope relation OBL NOT. The only difference is that now the phrase below PolP has both mood and tense features, as is shown in (40).

  1. (40)

4.4 The structure of ole+V+(t)tAvA

Consider (22c), repeated in (41).

  1. (41) Aamu-lla ei ole juo-ta-va alkoholi-a.morning-adeneg.3sg be drink-pass.pres.prt alcohol-part ‘One should not drink alcohol in the morning.’

The syntactic structure of this sentence gives rise to no particular problems. It is as presented in (42), with OBL taking scope over negation, as required (the symbol e here indicates a deleted PF copy. It should not be confused with the verb of negation).

  1. (42)

The auxiliary ole is in the non-past tense and has been moved to T from Aux. The participle juotava is under Prt. The verb juo- has been moved from V and adjoined to the passive present participle. Although the participle is separated from Mod1 by three intervening heads (Aux, T and Pol; unlike the other modals we have seen so far, which are separated from it only by Pol as seen in (38) and (40) above), it appears that the participle is the phonological form which is checked, as it appears to carry the modal meaning, rather than the auxiliary. As illustrated in (43), the participle can be used as an adjective, and when it is so used, it generally conveys the idea of ‘something to be done’. This suggests that it carries obligative meaning by itself, without any need for an auxiliary.

  1. (43)

    1. (a) Mies on kutsu-tta-va.man.nom be.3sg invite-pass.pres.prt ‘The man is to be invited.’

    2. (b) On-ko vielä jotain teh-tä-vä?be.3sg.q still something do-pass.pres.prt ‘Is there still something to be done?’

The participle can also have the meaning ‘can be done’, so that (43a), for instance, can be interpreted as ‘the man can be invited’, depending on context (Karlsson Reference Karlsson1999: 197). Whichever interpretation we have, a modal meaning belongs to the participle, which indicates that it is the item which is checked. As it is in the same finite clause as Mod1P, and thus within its checking domain, I will assume that this is so.

4.5 Modals which scope below negation

No problems arise with the modals of possibility and permission, voida and saada. As the bracketed structures for the examples in (44) indicate, these scope below negation and so we would expect to find them in T/Mod2P, exactly where their PF forms appear.

  1. (44)

    1. (a) Tämä ei voi autta-a. NOT POSS this.nomneg.3sg can help-inf1 ‘This cannot help.’

    2. (a′) [FPTämä[Fei[PolP[Polei[Mod2P[Mod2voi[VP auttaa]]]]]]]

    3. (b) Sinä e-t saa tul-la mukaan. NOT PER you.nomneg-2sg may come-inf1 along‘You may not come along.’

    4. (b′) [FPSinä[Fet[PolP[Polet[Mod2P[Mod2saa[VP tulla mukaan]]]]]]]

Similarly tarvita, which is used when negation scopes over obligation/necessity, also appears below negation, in the position in which it takes scope. The following illustrate:

  1. (45)

    1. (a) Meidän ei tarvitse lähte-ä. NOT OBL we.genneg.3sg need leave-inf1‘We need not leave.’

    2. (b) [FP Meidän [PolP[Polei[Mod2P[Mod2tarvitse [VP lähteä]]]]]]

It is clear, then, that the behaviour of the modals can be accounted for under the hypothesis that a higher modal phrase exists, encoding their LF parts, and checking their PF parts at a distance.

5. A problem for the long-distance checking hypothesis

Although the hypothesis of long-distance checking can account for the behaviour of the modal verbs, modality is also expressed in Finnish by two morphological moods, the conditional and potential. These pose a problem for the above account. They scope over negation, but in this case there appears to be no head above negation which can account for this.

The conditional mood indicates that an action is hypothetical and conditional on some other action. It usually translates as the English ‘would’, although in Finnish it occurs on both verbs in a conditional sentence:

  1. (46) Tul-isi-t-ko, jos pyytä-isi-n?come-cond-2sg.q if ask-cond-1sg ‘Would you come if I asked?’

The potential mood indicates that an action is probable. It has no past tense form, though it can occur in the present perfect.

  1. (47)

    1. (a) Presidentti saapu-ne-e huomenna.president.nom arrive-pot-3sg tomorrow‘The President will probably arrive tomorrow.’

    2. (b) Hän liene-e ol-lut Ruotsi-ssa.s/he.nom be.pot-3sg be-past.prt Sweden-ine ‘S/he has probably been in Sweden.’

In their interaction with negation, the verb of negation as usual appears before the main verb, as is illustrated in (48).

  1. (48)

    1. (a) Hän ei tul-isi, jos pyytä-isi-mme. COND NOTs/he.nomneg.3sg come-cond if ask-cond-1pl ‘S/he would not come, if we asked.’

    2. (a′) [FP Hän [Fei [PolP [Polei [Mod2P [Mod2 tulisi …]]]]]]

    3. (b) Hän ei soitta-ne tänään. PROB NOTs/he.nomneg.3sg ring-pot today‘S/he probably won't ring today.’

    4. (b′) [FP Hän [F ei [PolP [Polei [Mod2P [Mod2 soittane tänään]]]]]]

However, there is a problem here for the hypothesis proposed in this paper. We are proposing that there is no LF-movement in the case of the modals, and that scope relations can be accounted for by the LF-interpretable parts of a modal being merged where they are actually interpreted. It seems that similar considerations should apply to the moods. However, it is very difficult to think of any circumstances in which the moods do not scope over negation in the above sentences, but the verb of negation clearly c-commands the moods, as can be seen in (48a′) and (48b′) above, and thus should scope over them. There is no higher Mod1P above negation to appeal to, as it would be quite semantically anomalous to have such a phrase projected for the conditional and potential moods, as both these moods express hypotheticals. It seems, then, that the only way to account for the scopal relations between the moods and negation is by appealing to LF-movement.

It is not immediately obvious how to account for the behaviour of the conditional and potential while maintaining the position that there is no LF-movement. One possibility lies in the head Finiteness, which asymmetrically c-commands all other heads in the Finnish IP. As was argued in Section 3.2 above, this head encodes the logophoric centre of the clause, including epistemic modality, as the logophoric centre also includes the speaker's attitudes to the situation expressed by the clause (Bianchi Reference Bianchi, Gueron and Tasmovski2003; Adger Reference Adger and Nikolaeva2007). The conditional and potential are both epistemic moods, and hence the presence of epistemic modality on Finiteness may account for the scope relations seen in the examples in (48) above. This would also accord with the fact that epistemic modality takes scope over an entire clause, including the subject, while deontic modality takes scope only over the predicate, excluding the subject (Butler Reference Butler2003). While this seems a promising hypothesis, the precise details need to be examined, in particular the relationship of finiteness to the two lower Modal heads Mod1 and Mod2, both of which can already encode epistemic modality. I will not pursue the matter of the conditional and potential moods further here.

6. Conclusion

The Finnish data shown above provide evidence for the view that there are two positions for modals in syntax, one for modals of obligation and necessity and another for modals of possibility and permission, as Cormack & Smith (Reference Cormack, Smith, Barbiers, Beukema and van der Wurff2002) have argued for English. The higher of these positions also accounts for the behaviour of the Finnish negative imperative, as shown by Mitchell (Reference Mitchell1993) . The data also support a view of syntax in which a syntactic structure is already an LF-interpretable object in its own right, and LF-movement need not occur at all. The LF-interpretable parts of the modals of obligation and necessity occur in a position which asymmetrically c-commands negation, and thus they can take scope over it, even though their PF parts are found beneath it. For the modals of permission and possibility, both the LF- and the PF-interpretable parts occur below negation, where these modals take scope. This analysis accounts for the behaviour of the Finnish modal verbs and the imperative mood. However, the other two morphological moods of Finnish are rather more problematic. The presence of epistemic modality on Finiteness, the highest head in the Finnish IP, may provide an explanation for their behaviour with respect to negation, but for the present this must be left as a problem for future research.

APPENDIX

Footnotes

[1]

I would like to thank Erika Mitchell for helpful discussions on the structure of the Finnish IP, Heikki Kangasniemi for help with participial constructions, Andrew Simpson for a discussion of long-distance checking, and two anonymous JL referees for many helpful comments.

[2] May is allowed to be merged in two positions in Cormack & Smith (Reference Cormack and Smith1998).

[3] Of the other modals, kyeta and pystyä are used solely to express dynamic possibility, and joutua to express dynamic necessity. Voida, pitää, täytyä and tarvita also have dynamic uses, but I will not consider them in the main body of the paper. Saattaa can be used interchangeably with voida. Taitaa and mahtaa are colloquial, and indicate probability. Sopia and päästä are both largely used to express dynamic modality, though sopia can also express deontic modality, in which case it does not behave differently to täytyä, pitää and tulla. Päästä always has implications of motion and expresses dynamic modality.

[4] The partitive case alternates with the accusative as an object case in Finnish and is used to indicate an unspecified quantity, or that an object is not wholly affected by the action denoted by a verb. It is also routinely found on the objects of negative sentences.

[5] Finnish has four infinitives, used in a variety of non-finite constructions. The first infinitive is the form found in dictionaries and generally corresponds to the English infinitive. The term ‘first’ does not imply that it is a first person form.

[6] Interestingly, this verb can also be used to mean ‘must’, though when it is used in this way, it has less deontic force than other verbs of obligation. Native speakers of Finnish judge the several means of expressing obligation which are available in the language on a scale according to how ‘strong’ the obligation is, or how much deontic ‘force’ the verb has. The imperative has a stronger deontic force than any of the modals (Kangasniemi Reference Kangasniemi1992: 110–112).

  1. (i) Sinä saa-t pala-ta.you.nom may/must-2sg return-inf1‘You may/must return.’

Obviously, such sentences will be ambiguous out of context. The interpretation depends on whether the referent of the subject of the sentence is understood by the listener to want to undertake the action of the main verb or not. If they are, then saada is interpreted as giving permission, and if not, it is interpreted as indicating obligation.

[7] An anonymous JL referee has stated that according to his/her intuitions, this sentence should be interpreted as ‘it is not obligatory to drink alcohol in the morning’, i.e. with negation scoping over obligation. However, Heikki Kangasniemi (p.c.) informs me that to most Finnish speakers, the interpretation is one in which obligation scopes over necessity, the other interpretation only being possible with exceptional intonation, for instance. This may be an area where different speakers have different intuitions. I will consider only the interpretation given in the text, i.e. ‘one should not drink alcohol in the morning’.

[8] It is not clear where declarative mood fits in here. It is an epistemic mood which expresses certainty on the part of the speaker, and thus it could not be found under a head which otherwise expresses hypotheticalness. Mod1 would seem a possible host for this mood, as epistemic necessity is also found under this head. For simplicity, I will assume here that in a declarative sentence Mod1 is present, and is in the same scope relationship to the verb (which, in the case of declarative mood, has a null mood suffix) as when an overt modal is present. A simple declarative sentence is always tensed, and hence the phrase below PolP will be TP (see Section 1 above). If the verb of negation is present, it will of course not be possible for the verb to move beyond it.

[9] The accusative in Finnish has two forms in the singular, one marked with the suffix -n and the other identical to the nominative, except for pronouns, where the accusative is indicated by the suffix -t. The nominative-like form is used in circumstances where no other nominative-marked phrase can be present. The ending -n is in fact identical to the genitive singular, although historically this is a coincidence, as the accusative ending was originally -m which became -n in modern Finnish. It can be argued that the -n form of the accusative is in fact the genitive case, while the nominative-like form is a nominative (Yip, Maling & Jackendoff 1987; Maling Reference Maling, Holmberg and Nikanne1993; Toivainen Reference Toivainen, Holmberg and Nikanne1993; Vainikka Reference Vainikka, Holmberg and Nikanne1993). However, Reime (Reference Reime, Holmberg and Nikanne1993) has given plausible arguments for regarding both the -n and nominative-like forms as together constituting a genuine accusative. The subjects of passives, when they are not pronouns, are in the nominative-like form.

[10] An alternative view comes from Mitchell (Reference Mitchell1993, Reference Mitchell1994), who suggests that agreement is not a functional category but a relational one, whose function is to link elements of the phrase together, rather than having a meaning in its own right. She suggests that it can be hosted by other projections, and in Finnish should be regarded as hosted by her OblP, which dominates PolP. If this is so, then the specifier of OblP must be able to host both nominative and non-nominative subjects, and its head must be the final stopping point for agreeing verbs. However, this is problematic as there is no reason why the phrase should be projected if it is not necessary – indeed in conditional and potential sentences, its presence would be semantically anomalous. However, such sentences still have nominative subjects, and in affirmative sentences, agreement on the verb.

[11] That is, it is marked with the suffix -n. This suffix is used for both accusative and genitive case in the singular.

[12] The LF and PF parts of an item should perhaps not be presented as an unordered pair. They are related by the relation ‘carries information’, that is to say, the PF part carries information about the LF part. Therefore, lexical items might be better represented as PF→LF, where → stands for ‘carries information’. This information-carrying relation may be what ensures that the PF parts of a sign are inserted under the correct heads, that is, the heads which encode their corresponding LFs.

References

REFERENCES

Adger, David. 2007. Three domains of finiteness: A Minimalist perspective. In Nikolaeva, (ed.), 23–58.Google Scholar
Aygen, Gülşat. 2006. Finiteness and the relation between agreement and nominative case. In Boekcx, Cedric (ed.), Agreement systems, 63–98. Amsterdam & Philadelphia, PA: John Benjamins.CrossRefGoogle Scholar
Baker, Mark. 1988. Incorporation: A theory of grammatical function changing. Chicago: University of Chicago Press.Google Scholar
Bianchi, Valentina. 2003. On finiteness as logophoric anchoring. In Gueron, Jacqueline & Tasmovski, Liliane (eds.), Temps et point de vue/Tense and point of view, 213–246. Nanterre: Universite Paris X.Google Scholar
Boeckx, Cedric. 2008. Aspects of the syntax of agreement. New York: Routledge.CrossRefGoogle Scholar
Boeckx, Cedric & Stjepanović, Sandra. 2001. Head-ing towards PF. Linguistic Inquiry 32, 345–355.CrossRefGoogle Scholar
Burton-Roberts, Noel. 2009. On the grounding of syntax – and more. Newcastle Working Papers in Linguistics 15, 22–39.Google Scholar
Butler, Johnny. 2003. A Minimalist treatment of modality. Lingua 113, 967–996.CrossRefGoogle Scholar
Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press.Google Scholar
Chomsky, Noam. 2000. Minimalist enquiries: The framework. In Martin, Roger, Michaels, David & Uriagereka, Juan (eds.), Step by step: Essays on Minimalist syntax in honor of Howard Lasnik, 89–155. Cambridge, MA: MIT Press.Google Scholar
Chomsky, Noam. 2001. Derivation by phase. In Kenstowicz, Michael (ed.), Ken Hale: A life in language, 1–52. Cambridge, MA: MIT Press.Google Scholar
Comrie, Bernard. 1985. Tense. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
Cormack, Annabel & Smith, Neil. 1996. Checking theory: Features, functional heads and checking-parameters. UCL Working Papers in Linguistics 8, 243–281.Google Scholar
Cormack, Annabel & Smith, Neil. 1997. Checking features and split signs. UCL Working Papers in Linguistics 9, 223–252.Google Scholar
Cormack, Annabel & Smith, Neil. 1998. Negation, polarity and V positions in English. UCL Working Papers in Linguistics 10, 285–322.Google Scholar
Cormack, Annabel & Smith, Neil. 2002. Modals and negation in English. In Barbiers, Sjef, Beukema, Frits & van der Wurff, Wim (eds.), Modality and its interaction with the verbal system, 133–164. Amsterdam & Philadelphia, PA: John Benjamins.CrossRefGoogle Scholar
Cristofaro, Sonia. 2007. Deconstructing categories: Finiteness in a functional-typological perspective. In Nikolaeva, (ed.), 91–114.Google Scholar
Fodor, Jerry. 2008. LOT2: The language of thought revisited. Oxford: Oxford University Press.CrossRefGoogle Scholar
Haegeman, Liliane. 1992. Theory and description in generative syntax: A case study in West Flemish. Cambridge: Cambridge University Press.Google Scholar
Haegeman, Liliane (ed.). 1997. Elements of grammar: Handbook in generative syntax. Dordrecht: Kluwer.CrossRefGoogle Scholar
Holmberg, Anders. 2001. The syntax of yes and no in Finnish. Studia Linguistica 55, 141–175.CrossRefGoogle Scholar
Holmberg, Anders & Nikanne, Urpo. 2002. Expletives, subjects and topics in Finnish. In Svenonius, Peter (ed.), Subjects, expletives and the EPP, 71–106. Oxford: Oxford University Press.CrossRefGoogle Scholar
Holmberg, Anders & Nikanne, Urpo (eds.). 1993. Case and other functional categories in Finnish syntax. Berlin & New York: Walter de Gruyter.CrossRefGoogle Scholar
Holmberg, Anders, Nikanne, Urpo, Oraviita, Irmeli, Reime, Hannu & Trosterud, Trond. 1993. The structure of INFL and the finite clause in Finnish. In Holmberg, & Nikanne, (eds.), 177–206.Google Scholar
Holmberg, Anders & Platzack, Christer. 1995. The role of inflection in Scandinavian syntax. Oxford: Oxford University Press.CrossRefGoogle Scholar
Kangasniemi, Heikki. 1992. Modal expressions in Finnish. Helsinki: Suomelaisen Kirjallisuuden Seura.Google Scholar
Karlsson, Fred. 1999. Finnish: An essential grammar. London & New York: Routledge.Google Scholar
Maling, Joan. 1993. Of nominative and accusative: The hierarchical assignment of grammatical cases in Finnish. In Holmberg, & Nikanne, (eds.), 49–74.Google Scholar
McCloskey, James. 1996. On the scope of verb movement in Irish. Natural Language & Linguistic Theory 14, 47–104.CrossRefGoogle Scholar
McCloskey, James. 1997. Subjecthood and subject positions. In Haegeman, (ed.), 197–235.Google Scholar
Mitchell, Erika. 1991. Evidence from Finnish for Pollock's theory of IP. Linguistic Inquiry 22, 373–379.Google Scholar
Mitchell, Erika. 1993. Morphological evidence for syntactic structure. Ph.D dissertation, Cornell University.Google Scholar
Mitchell, Erika. 1994. When AgrO is fused to AgrS: What morphology can tell us about the functional categories. MIT Working Papers in Linguistics 22, 113–130.Google Scholar
Mitchell, Erika. 2006. The morpho-syntax of negation and the positions of NegP in the Finno-Ugric languages. Lingua 116, 228–244.CrossRefGoogle Scholar
Nikolaeva, Irina (ed.). 2007. Finiteness: Theoretical and empirical foundations. Oxford: Oxford University Press.CrossRefGoogle Scholar
Palmer, Frank Robert. 2001. Mood and modality. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
Reichenbach, Hans. 1947. Elements of symbolic logic. New York: Free Press.Google Scholar
Reime, Hannu. 1993. Accusative marking in Finnish. In Holmberg, & Nikanne, (eds.), 89–109.Google Scholar
Rizzi, Luigi. 1997. The fine structure of the left periphery. In Haegeman, (ed.), 281–337.Google Scholar
Simpson, Andrew. 2000. Wh-movement and the theory of feature-checking. Amsterdam & Philadelphia, PA: John Benjamins.CrossRefGoogle Scholar
Toivainen, Jorma. 1993. The nature of the accusative in Finnish. In Holmberg, & Nikanne, (eds.), 111–128.Google Scholar
Vainikka, Anne. 1993. The three structural cases in Finnish. In Holmberg, & Nikanne, (eds.), 129–162.Google Scholar
Whitney, Arthur H. (ed.). 1971. Finnish reader: Extracts from modern Finnish literature. London: Hodder & Stoughton.Google Scholar
Yip, Moira, Maling, Joan & Jackendoff, Ray. 1987. Case in tiers. Language 63, 217–250.CrossRefGoogle Scholar