Chapter 3 Towards a Criterial V2: an ideal mechanism
3.3. On Subject V2
3.3.3. V2 and Subject Initial Sentences: a typology of subjectV2
Even though Subjects can be Topics or Foci, it could also be the case that there exist instances of a Spec-head configuration in their dedicated Criterial position. Such a hypothesis follows the asymmetric analysis which has been proposed by Travis (1984) and Zwart (1997), as briefly sketched in chapter 2 and which I shall discuss in the next sub-sections.
3.3.3.1. Subject-Initial contexts: some theoretical background
As mentioned in chapter 2, Travis (1984) and Zwart (1997) limited cases in which movement of the verb to CP takes place. They proposed that the finite verb only raises to a high functional head of the inflectional domain (IP) in subject-initial main clauses. The literature refers to this account as V inside TP (VITP). On the other hand, Vikner and Schwartz (1996) assert that ''the verb always leaves IP52'' in V2 main clauses. This account
52 The title of their 1996 publication..
is referred to as V outside TP (VOTP). Van Craenenbroeck and Haegeman (2007) discussed evidence for such an account drawing from two Flemish dialects53 in (42).
Adopting from Van Craenenbroeck and Van Koppen (2002), “in embedded clauses by Van Craenenbroeck and Van Koppen (2002) […] in embedded clauses this object clitic obligatorily follows the complementizer and precedes the DP subject (42a). In non-subject- initial main clauses, on the other hand, the clitic follows the inflected verb and precedes the subject” (Van Craenenbroeck and Haegeman 2007: 169), as given in (42b).
(42) Wambeeck dialect of Flemish a. dan-t Marie al wetj.
that-it Marie already knows .... that Marie already knows it.'
b. Nou wenj-t Marie al.
now knows-it Marie already 'Now, Marie already knows it.'
(Van Craenenbroeck & Haegeman 2007: 169; 5)
In chapter 4, I shall develop an analysis for (42) in terms of a criterial approach to V2. In (43) the different syntactic structures according to the two accounts: (a) stands for subject-initial sentences and (b), (c) as non-subject subject-initial sentences (respectively another thematic role, e.g. object, and a Prepositional Phrase). English is employed as a metalanguage.
(43) VITP vs VOTP analyses of V2
VOTP (Schwartz & Vikner 1996, Van Craenenbroeck & Haegeman 2007) a. [SpecFinP/SpecTopP/SpecFocP John [FinP reads [IP the book [in the library]]]]
b. [SpecTopP/SpecFocP The book [SpecFinP - [FinP reads [IP John [in the library]]]]
c. [SpecTopP/SpecFocP In the library [FinP reads [IP John [the book]]]]
53 The object clitic –t in Wambeek dialect and the particle –tet in Lapscheure dialect are believed to occupy a very high position within the IP. These items follow the inflected verb in subject-initial clauses. Such evidence led the authors to claim that the verb moves to C in subject-initial sentences too.
VITP (Travis 1984, Zwart 1997)
a. [CP [SpecIP John [IP reads the book in the library]]
b. [SpecTopP/SpecFocP The book [SpecFinP - [FinP reads [IP John [in the library]]]]
c. [SpecTopP/SpecFocP In the library [FinP reads [IP John [the book]]]]
In line with the goals of a rigid Cartography of V2, it is desirable to agree with Travis (1984) and Zwart (1992, 1997). Adopting Poletto’s (2005: 216) words, “I would like to maintain Zwart’s basic intuistion that V2 effects can be achieved targeting different projections both cross-linguistically and language internally”.
Therefore, I shall propose two examples that may lead us to adopt a VITP analysis for subject-initial contexts, at the very least for Swiss Romansh varieties (section 3.3.3.2) and Icelandic (section 3.3.3.3) with different levels of analysis.
The question arises as to which position is the subject position in a cartography of subject positions. As already noted by Cardinaletti (2004: 116), "the preverbal subject field is more complex than usually thought, and more than one subject position should be assumed. The properties attributed to preverbal subjects (subject of predication, checking the EPP feature, checking phi features, checking nominative case) can be distributed across discrete functional projections, each of which realizes a feature or a set of features”.
Her work confirms the observation by McCloskey (1997: 197) that there is "a progressive deconstruction of the traditional category 'subject' so that the properties which are supposed to define it are distributed across a range of distinct (but derivationally linked) syntactic entities and positions". In other words, “there exists more than one preverbal subject position”. (Cardinaletti 2004: 120). Indeed, Cardinaletti (2004) was the first to propose a “Cartography of Subject positions”, given in (44).
(44) Cartography of subject positions [SpecSubjP [SpecEppP [specAgrSP ]]]
(Cardinaletti 2004: 154; 156)
Unlike Left Peripheral positions, the analysis of canonical subject V2 will follow a more general path. In other words, I propose that Standard German subjects may target different positions.
DP subjects in German target at least two positions (Cardinaletti 1990, Mohr 2002): one lower for indefinites and one higher usually labelled as ReferenceP (Koopman
& Szabolcsi 2000), as given in (45).
(45) German
a. Diesen Satz haben schon immer alle Studenten gehaßt.
this sentence have already always all students hated
‘This sentence, all students have always hated.’
b. *Diesen Satz haben schon immer die Studenten gehaßt.
this sentence have already always the students hated
c. ? Diesen Satz haben alle Studenten schon immer gehaßt.
this sentence have all students already always hated
d. Diesen Satz haben die Studenten schon immer gehaßt.
this sentence have the students already always hated
(Mohr 2002: 2; 5)
Naturally, DP and prononimal subjects also target Left Peripheral functional projections (Cf. Frascarelli & Hinterhölzl 2007), contrary to impersonal and expletive subjects.
Not all subject initial contexts may be the result of a Spec-Head configuration in SubjPs.
However, the lack of Spec-head configuration does not prevent a ‘superficial’ V2 in subject initial contexts, if nothing intervenes between the subject and the landing site of the inflected verb to a T position, as will be proposed in section 3.3.3.3.
3.3.3.2. Spec-Head configuration in SpecSubjP: Evidence from Swiss Romansh varieties and subject clitics
Following Anderson (2016: 169), the label Swiss Romansh (henceforth SR) indicates five varieties, all of which have written standards. Those indicated in (29) are spoken in the canton of Grisons (German Graubünden, Italian Grigioni, French Grisons) in Switzerland.
In the typology discussed in chapter 2, Wolfe (2016) considers SR varieties to be V2-to-Force languages, following Haiman & Benincà’s (1992: 150) claim that SR varieties
“avoid V3”.
(46) Swiss Romansh Varieties Variety
Surselvan Sutselvan Surmiran Puter Vallader
In all Swiss Romansh varieties54, the subject can be doubled with an enclitic on the inflected verb. The varieties only differ according to person and number. In (47), evidence from Surmiran is shown, in which a doubling subject enclitic is optional in non-subject initial contexts, whereas it is completely excluded in subject-initial sentences.
(47) Surmiran
a. Rumantsch discorra Ursus stupent Rumantsch speaks.3S Ursus excellently
‘Ursus speaks Rumantsch very well’
(Anderson 2005: 206; 7.44a)
54 Surselvan only allows doubling with an impersonal subject.
b. Rumantsch discorra=’l Ursus stupent Rumantsch speaks.3S.3SM Ursus excellently
‘Ursus speaks Rumantsch very well’
(Anderson 2005: 206; 7.45a)
c. Ursus discorra stupent Rumantsch Ursus speaks.3S excellently Rumantsch
‘Ursus speaks Rumantsch very well’
(Anderson 2005: 206; 7.43)
d. *Ursus discorra=’l stupent Rumantsch Ursus speaks.3S.3SM excellently Rumantsch
‘Ursus speaks Rumantsch very well’
(Anderson 2005: 207; 7.45b)
As (47) shows, subject enclitics doubling a lexical subject only occur in non-subject initial contexts. Among the varieties, doubling is possible in "all person/number combinations where a subject enclitic is available, that is, 1sg, 1pl and 3sg55, 3pl.’’ (Fuß 2005: 190).
Within the non-generative literature, clitic doubling was considered to be stylistic because it serves “to reinforce the subject’’. In particular, Sutselvan shows a different, and I would define more cartographic, form of clitic doubling: the clitic doubling which "appears to be more common’’ (Fuß 2005: 192), and in the words of Fuß (2005) is now "almost an obligatory phenomenon that has lost its function as a stylistically marked structural option". As a matter of fact, the DP does not receive stress (Linder 1987: 150). Relevant examples are given in (48):
55 In Putér, Vallader, and Surmiran, doubling seems exhibiting a definiteness effect: the full DP must be definite (or specific), that is, sentences such as ‘yesterday arrived he a man’ are not possible (Fuß 2005:
192).
(48) Sutselvan
a. Egn da quels lev-i ear jou One of those wanted-cl1s also I
‘I also wanted one of those’
(Linder 1987: 148) b. Ascheia vain-sa nus arviart igl mulegn…
so have-cl1p we unlocked the mill
‘So we have unlocked the mill’
(Linder 1987: 149) c. Igl fetschi preaschas, â-l el getg.
It is urgent, has-cl3sm he said.
‘He said it’s urgent’
(Linder 1987: 153) d. Cunquegl c’igl eara november, vev-la la scola antschiat
since it was November, had-cl3sf the school begun
‘Since it was November, the school had begun’
(Linder 1987: 153) e. Natiral vev-in las matàns radetg sei mailenders.
Of course had.cl3P the girls brought up Milans
‘Of course, the girl had brought up some Milans [pastries]’
(Linder 1987: 193)
SR varieties (Fuß 2005, Anderson 2005) may be considered VITP (Travis 1984, Zwart 1997) languages, namely that the verb (and the fronted XP) does not move to the Left Periphery in Subject Initial sentences, but in a lower Subject position, namely AgrS° in Cardinaletti’s (2004) system. The clitic possibly lies in the head of SubjP (as in Northern Italian dialects, Rizzi 2015: 333) and the clitic becomes part of the verbal inflection in the targeted head in the CP and spelled-out as -l. In (49) the derivation for (47).
(49) Surmiran V2s: subject vs. non-subject initial sentences a. Subject Initial
[CP [SpecSubjP Ursus [Subj° [Epp° [AgrS° discorra [IP stupent Rumantsch]]]]]]
b. Non-subject initial
[SpecFocP/TopP Rumantsch [Foc°/Top° discorral [SpecSubj Ursus[Subj° discorral[AgrS° discorra [IP ∆]]]]]]
(adapted from Anderson 2005 and Fuß 2005)
3.3.3.3. The verb remains in the IP: Icelandic V3 adverbs
In Icelandic, a class of adverbials is able to intervene between the subject and the inflected lexical verb, creating the order Subj – Adv – LexVerb. The adverbs triggering V3 structures (V3 adverbs, following Jónsson 2002 in Þráhinsson 2007: 39) are auðvitað 'obviously', líklega 'probably', sennilega 'probably', ennþa 'still', kannski 'maybe',
náttúrulega 'naturally', vonandi 'hopefully' and satisfy V2 requirements if they are preposed. Furthemore also focusing adverbs, like bara 'just' and einfaldlega 'simply' allow V3 structures in subject-initial contexts, but they do not satisfy V2 requirements (Cinque 1999: 169fn12; Nilsen 2003). On the other hand, other adverbs like aldrei 'never' and alltaf 'always' cannot intervene between the subject and the inflected verb. According to the description in Þráhinsson (2007: 39 – 40; 2.43) only auxiliaries can precede V3 adverbs.
In previous works (Samo 2014, 2016b), I have observed that these adverbs reside in the higher positions56 of the Inflectional Phrase, as in (50):
(50) [Modevaluative náttúrulega/vonandi [Modevidential auðvitað [Modepistemic likelega / sennilega [Modirrealis kannski [ASPcontinuative ennþa [Aspperfective alltaf / never ]]]]]]]
I thus propose that the subject moves to its criterial position SpecSubjP and the verb targets a T position within the IP higher than the ModEvaluativeP position (Cinque 1999) ‘always/never’. Auxiliaries do not show such a pattern, leading me to conclude that they may be merged higher or they represent the overt realization of Subjº, the position where the verb lands in Surmiran subject-initial contexts. In (51) the derivation:
56 The adverbs náttúrulega and vonandi can fill the class of the evaluative adverbs. In order to classify these adverbs, Cinque (1999: 84) assumes the same type of analysis as the logic tradition, that is, for p is a proposition, ''it is good/perfectly/bad thing that p'' (Cinque 1999:84 from Palmer 1986:12ff). Then, if I report here the sentence in (24a), Jón náttúrulega lýkur þessu einhvern daginn 'John naturally finishes this some day', I can turn it into "it is a natural thing that John finishes this some day"; the same pattern seems to work with vonandi 'hopefully', ''It is a hopeful thing that John finishes this someday''. The adverb auðvitað 'obviously' is one of the plausible candidates for generation in the specifier position of the Evidential Mood Phrase, according to Cinque (1999: 86). The adverbs sennilega and likelega, both 'probably', seem to be generated in the specifier of the Epistemic Mood Phrase (Cinque 1999: 86). The adverb kannski 'maybe' finds place in the specifier of the Irrealis mood (Cinque 1999: 88 – 89); the adverb ennþa 'still', according to Cinque (1999: 95) appears to be generated in the specifier of the Continuative Aspect Phrase.
(51) Icelandic V3 (Samo 2016)
a. [SpecSubj° Jón [Aux° - [ModEvaluative náttúrulega/vonandi [ModEvidential auðvitað [ModEpistemic
likelega / sennilega [ModIrrealis kannski [AspContinuative ennþa [Tense° lýkur [AspPerfective alltaf / never ]]]]]]]]]
b. [SpecSubj° Jón [Aux° hefur [ModEvaluative náttúrulega/vonandi [ModEvidential auðvitað
[ModEpistemic likelega / sennilega [ModIrrealis kannski [AspContinuative ennþa [AspContinuative ennþa [AspPerfective alltaf / never [Tense lokið]]]]]]
c.
Icelandic belongs to the generalized V2 languages, where the inflected verb can also move to the IP in embedded contexts. To be more specific, as reported in Wiklund et al.
(2007: 210), the verb obligatorily precedes alltaf 'always' in embedded contexts (as in main clauses).
To sum up, I propose that, unlike in SR varieties, the verb is not in a Spec-head configuration with the subject, but the verb moves to a lower functional projection. The verb targets a T position57 like in Italian (Belletti 1990) or French (Pollock 1989). Finally, as SR varieties, Icelandic has verb movement in embedded contexts.
3.3.3.4. Spec-Head in SubjP, but the subject targets a lower subject position.
A different syntactic structure is involved in SR varieties, for which it has been proposed that the subject can target a Topic position in the low IP area, as in Italian (Belletti 2004).
Frey (2004), Grewendorf (2005), and G. Bianchi (2012) propose that subjects in German may also target this lower functional projection. Evidence for subjects in Belletti’s (2004) lower Topic positions for SR varieties are shown in (52), where the past participle also precedes the subject as in Italian. SubjP is however activated, since the verb shows subject clitics as observed in section 3.3.3.2..
(52) Putér
a. Co dalungia füttan-e à Senece et a sa duonna avertas sü las avainas Then immediately were.3P to Seneca and to his wife opened up the veins in la bratscha
in the arms
‘‘Immediately, then, they slashed the wrists of Seneca and his wife.’
(Linder 1987: 160 from Fuß 2005: 197; 92) Vallader
b. Che effet ha=la gnü aint il pövel la nouva predgia?
Which effect has.CL3SF had on the people the new sermon?
‘Which effect had the new sermon on the people?’
(Linder 1987: 154)
Surmiran
c. An quell mumaint è-la riveda er la mama in this moment is.CL3SF arrived also the mother
‘In this moment, the mother arrived too’
(Linder 1987: 154) Sutselvan
d. Anzucuras e-la lura la mort gnida Eventually is.3F then the death come
‘And eventually, the death then came…’
(Linder 1987: 155) e. “Ah, nagn a tgea”, â-l getg igl fumegl
Go-1p to home, has.3SM said the servant
“Well, let’s go home”, said the servant..’
(Arquint-Felix; Linder 1987: 155) f. Dalùnga en-i svanidas las pusadas
immediately are.3P vanished the cutlery-pl
‘Immediately, the cutlery vanished’
In sum, V2 languages may differ according to the required movement of the subject or the verb, but the subject criterion in SubjP/SpecIP should be activated. I propose that the examples in sections 3.3.3. suggest that “canonical” subjects do not move to the CP, adopting a VITP analysis.
58 As noted by Fuß (2005) (27d) and (27e) are written by the the same author (J.C. Arquint, Cuorsda Rumantsch. Adataziun sutsilvana da A.-L. Felix. Cuera, 1958). “Thus, it might be possible that (27a–b) reflect the grammar of a single speaker that has not yet completed the reanalysis of clitics as Agr-morphemes on C. Alternatively, one might attribute the placement of the participle in (27) to influence from Italian, where the participle always precedes the subject in inversion contexts.” (Fuß 2005: 198).