• Aucun résultat trouvé

Summing up: Towards a typology of criterial V2s?

Chapter 4 Criterial V2 and embedded contexts

4.3. Summing up: Towards a typology of criterial V2s?

In order to overcome the theory-internal drawbacks of mainstream analyses described in chapter 2, I proposed an ideal mechanism for V2 languages adopting Cartographic guidelines.

In Chapter 3, V2 is observed as the sum of residual V2s (Rizzi 1991): every criterial position requires a Spec - Head configurations with the inflected verb. In other words, movements are motivated by the satisfaction of scope-discourse semantics features.

Embedded clauses, “canonical” subjects and expletive subjects challenged the idea of a CriterialV2. The lack of the V2 pattern in embedded clauses (in the languages which do not allow it) is considered to be related to the nature of the complementizer. If the complementizer is first merged in the IP° and then moved to Force° (presumably, moving through all activated criterial positions) the language does not allow V2 in embedded clauses; if the complementizer is directly base-generated in Force°, the activated criterial positions should be activated through the movement of INFL from the IP. Data from Cimbrian and Colloquial German seems to point towards such a direction.

The presence of the verb to the “right” of the sentence is due to headedness.

According to the discussion in chapter 3 and chapter 4, variation among V2 languages is based on three parameters, as given in (43).

(43) Three variables in V2 languages

(i) The height of the V movement (T°, AgrS°, Subj°) in subject-initial contexts.

(ii) Headedness of inflection/verbal FPs (Kayne 1994; Haider 2010).

(iii) In embedded clauses, the locus of generation of the complementizer.

These three elements can be considered in order to build a typology of V2 languages.

However, I would like to speculate on a further element, which should be investigated in a criterial model of V2.

As stated in chapter 3, I conjecture V2 as the sum of residual V2s. Residual V2s are Spec-Head configurations involving the inflected verb and an activated criterial position.

One further variable at play in different V2 languages could be a pure syntactic element, which may be in line with the initial proposal in the Residual V2 hypothesis.

I propose that a further element involved in the micro-parametric variation of V2 languages is which Criterial positions require a Spec-Head configuration.

(44) A further variable

Functional projections requiring Spec-Head configurations.

I can restate (43) in (45).

(45) Four variables in V2 languages

(i) The height of the V movement (T°, AgrS°, Subj°) in subject-initial contexts.

(ii) Headedness of inflection/verbal FPs (Kayne 1994; Haider 2010).

(iii) In embedded clauses, the locus of generation of the complementizer.

(iv) Functional projections requiring Spec-Head configurations.

As observed with the data in Icelandic with V3 adverbs (section 3.3.3.3.), the verb does not create a Spec-Head configuration with the Subject. The same patterns are not possible in German, namely the verb should create a Spec – Head configuration in subject-initial contexts.

A very rough sketch of micro-parametric variation between V2 languages can be drawn following this proposal, as given in (46).

(46) FP Creating Spec – Head configurations

German: ForceP IntP TopP FocP ModP SubjP Icelandic: ForceP IntP TopP FocP ModP

What if a Left-Peripheral item (Topic, Focus, Mod) does not require a Spec-Head configuration in a V2 language? The ideal mechanism predicts that we should encounter some V3 orders since the ultimate landing site of the inflected verb will be in the highest functional projection requiring the Spec – Head configuration. Such speculations should be investigated in light of work concerning the loss of V2 (Haeberli 2002; Poletto 2014).

For example, if we imagine that ModP does not require Spec-Head configurations, we should expect an asymmetry between subject-initial and object-initial sentences as given in (47).

(47) Ideal V3 orders if ModP does not require Spec-Head configuration Subject-Initial sentences: V3

[ForceP [... [SpecMod Yesterday [Mod° - [SpecSubjP the student [Subj° read [IP the book]]]]]]]

Object-Initial sentences: V2

[ForceP[…[SpecFocusThe book [Foc° read [SpecMod yesterday [Mod°- [SpecSubj the student [IP ]]]]]

The proposal can fit with studies on the loss of V-to-C movement (cf. Roberts 2007).

I, therefore, investigated V3 orders in Standard German and in the attested literature in order to ascertain if certain syntactic elements have ceased to trigger a Spec – Head configuration. Furthermore, V3 orders serve as a diagnostic tool adopted by mainstream analyses in order to propose a typology of V3 orders.

Chapter 5

On V3 orders: a Qualitative Analysis of V3 orders involving Frame Setters and Hanging

Topics. Evidence from Standard German.

As briefly sketched in chapter 2, the Left Periphery adopted by mainstream analyses (1a, b) slightly differs from the one proposed in Rizzi (1997, 2001, 2004), Rizzi & Cinque (2016) and Rizzi & Bocci (2017) as shown in (1) and (2).

(1) a. [FrameP [ForceP [TopicP [FocusP [FinP]]]]]

(Wolfe 2015 based on Benincà & Poletto 2004, Poletto 200272) b. [HT [Deixis [Force [Topic [Focus [Fin]]]]

(Wolfe 2016)

(2) [Force [Top* [Int [Top* [Foc [Top* [Mod [Top* [Qemb [Fin [IP [...]]]]]]]]]]]]

(Rizzi & Cinque 2016:146)

In comparing (1) and (2), some differences can be observed:

(3) The Left Periphery of mainstream analyses

(i) In (1a, b), the Left Periphery is split into layers and not into functional projections.

Benincà & Poletto (2004: 53) display evidence in order to “identify two different fields in CP: a higher Topic field hosting nonoperator elements, and a lower Focus field hosting operator-like elements”. Indeed, they do not posit a Topic position lower than FocusP (or Focus layer).

72 The LP adopted by mainstream analyses is a smaller version of the LP which has been proposed for Italian and Italian varieties of Rhaeto-Romance by Benincà & Poletto (2004), in (i):

(i) [Hanging Top [Scene Setter [ForceP [TopP LD Topic [TopP List Intepretation Topic [FocP Contrastive Focus [FocP Informational Focus [FinP [FinP TP]]]]]]]

(Benincà & Poletto 2004)

(ii) In (1a, b), a layer of base-generated items exists higher than ForceP73; this layer is given the label FrameP, a projection unifying two different functional projections, the one for Hanging Topics (henceforth HT) and the one for Deictic elements, based on the functional projection SceneSetterP (Benincà & Poletto 2004). The order between these two layers seems to be fixed, as displayed by evidence from Italian, as given in (4).

(4) Italian

a. Mario, nel 1999, gli hanno dato il premio Nobel.

Mario in the 1999 to.him have given the Prize Nobel

b. ??Nel 1999, Mario, gli hanno dato il premio Nobel.

in the 1999 Mario to.him have given the Prize Nobel

c. *Sul giornale Mario ne hanno parlato malissimo.

on the newspaper Mario of him have spoken very badly

(Benincà & Poletto 2004: 67; 46)

A LP involving FrameP is fundamental for the analysis in Wolfe (2016). The typology is based on V3 orders among V2 languages (cf. chapter 2); strict V2 languages can admit V3 orders only if the leftmost element is base generated in a position higher than Force.

In (5), XP2 has been moved out from the clause and its trace in SpecFinP is able to block other elements from the IP. The only way to admit V3 orders is to posit that XP1 is generated higher than Force, in FrameP.

73 On the basis of the analysis of bridge verbs, which are usually claimed to have one additional CP layer, Benincà & Poletto (2004: 61) propose that “bridge verbs select a "full CP layer" with all CP projections available, [whereas] non-bridge verbs select only a smaller portion of the whole CP structure, pruning the CP projections where objects and some adverbial classes are assigned contrastive Focus, while permitting the lower CPs to be filled”.

(5) V3 orders in strict V2 languages

This chapter investigates the quality of the items base-generated in FrameP, whereas chapter 6 will be dedicated to those items that may not be externally merged.

This chapter is highly inspired by recent work by Haegeman & Greco (forthcoming), who looked closely at the differences in the types and distribuitions of scene setters in Standard Dutch and West Flemish. They find that in West Flemish the use of FrameP is common, even though speakers tend to be reluctant in questionnaires.

Frame Setters (following Haegeman & Greco, forthcoming) “provide temporal and/or modal restrictions to the circumstances of evaluation of the proposition expressed by the main clause (following Lewis 1975)”. According to Freywald et al. (2013: 12), the leftmost element “fulfills the function of providing an interpretational frame or anchor for the following statement, first, in terms of time, place, condition (in the case of adverbials meaning ‘from now on’, ‘yesterday’, ‘every year’, ‘if you are in school’ and so on), or second, more abstractly, in terms of discourse linking (as is the case in certain uses of the equivalents of ‘then’ and ‘afterwards’).” As for the nature of FrameP, Haegeman & Greco claim that the item is base-generated.

I will adopt the map in (1b) in which this functional layer has been split into Deixis and Hanging Topics. The former will be further split into (i) Scene Setters, adopting the traditional label in cartographic works by Benincà & Poletto (2004), in section 4.1., and (ii) a general Deixis layer, involving Speech-Act/Speaker oriented adverbs and peripheral

CPs (in the sense of Haegeman 2006 and following works) in section 4.2. Finally, Hanging Topics are investigated in section 4.3.

5.1. On FrameP: Scene Setters, Time and Space.

As the label suggests, FrameP is believed to host all those items creating a frame, especially from a temporal and spatial point of view. In the cartographic literature, the label has also been given as Scene Setter, as in Benincà & Poletto (2004). These items are considered to be base-generated74, since “embedded V2 never has a position for Scene

74 Cross-linguistically, these items believed to be externally merged in the LP do not seem to function as expected. For example, in Italian (I do not mark degrees of acceptability, but only grammatical vs.

*agrammatical):

(i) Italian

a. Ieri la mamma ha detto che l’imbianchino verrà domani.

Yesterday the mum has said that the housepainter will-come tomorrow b. La mamma ieri ha detto che l’imbianchino verrà domani.

c. La mamma ha detto ieri che l'imbianchino verrà domani.

If domani is “highlighted”, (a’), (b’) and (c’) are all ungrammatical (even if the subject targets a lower Topic position, Belletti 2004), plausibly in terms of featural Relativized Minimality (Rizzi 1990, 2004;

Starke 2001).

a’. *Domani ieri (la mamma) ha detto (la mamma) che l’imbianchino verrà.

b’. *Domani (la mamma) ieri ha detto (la mamma) che l’imbianchino verrà.

c’. *Domani (la mamma) ha detto ieri (la mamma) che l’imbianchino verrà.

If domani is focussed, (a’’), (b’’) and (c’’) are acceptable. As already proposed for adverbials in Rizzi (2004), the intervener in terms of locality is bypassed.

a’’. DOMANI ieri (la mamma) ha detto (la mamma) che l’imbianchino verrà, (non dopodomani).

b’’. DOMANI (la mamma) ieri ha detto (la mamma) che l’imbianchino verrà, (non dopodomani).

c’’. DOMANI (la mamma) ha detto ieri (la mamma) che l’imbianchino verrà, (non dopodomani).

If we assume that SceneSetters are generated higher than Force, we expect that a sentence like (ii) should be grammatical, but, according to my judgment, it is completely out.

(ii) *[FrameP Ieri [FocusP DOMANI [IP (la mamma) ha detto (la mamma) [CP2 che l'imbianchino verrà]]]]

(non dopodomani)

However, if ieri is “highlighted” in a lower position of the LP, such as ModP (Rizzi 2004), the judgment improves, as in (iii) = (ib’’).

(iii’) [FocusP DOMANI [ModP ieri [IP ha detto la mamma [CP2 che l'imbianchino verrà]]]] (non dopodomani)

The same patterns works for German:

(iv) Standard German

a. Morgen hat Mama gesagt _____ kommt er Tomorrow has Mom said ______ comes he b. *Morgen hat Mama gestern gesagt ____ kommt er

Setting, which is only available in root contexts. This also makes sense from the semantic point of view, as "setting the scene" is an operation done at the beginning of the utterance, not in an embedded context” (Benincà & Poletto 2004: 66)

However, some scene setters should undergo movement, since both temporal and locative items75 are able to trigger V2 structures, as observed in chapter 2 and chapter 3.

(6) Temporal and Locative V2 Temporal V2

Nach 1320 kehrte Giotto nach Florenz zurück After 1320 turned Giotto to Florence back Giotto came back to Florence after 1320'

Locative V2

Auf der linken Seite befindet sich ein kleines Fresko von Sano di Pietro.

On the left side arranged self a small fresco of Sano di Pietro.

'A small fresco by Sano di Pietro is on the left side'

On the other hand, it should be recalled that items like ‘yesterday’ do not create a V2 structure in certain V2 languages, such as Cimbrian.

(7) Cimbrian

c. MORGEN hat Mama gestern gesagt ____ kommt er (nicht uebermorgen) d. *Gestern, MORGEN hat Mama gesagt ____ kommt er

e. *Gestern, SCHNELL hat Mama gesagt _____ kommt der Maler f. *Am Telephon, MORGEN hat Mama gesagt ____ kommt er

g. Am Telephon morgen wird Mama sagen dass der Maler nächste Woche kommt.

75 English, a residual V2 language, shows a locative inversion (cf. Rizzi & Shlonsky 2006), but no temporal inversion, Rizzi p.c.) with something different involved.

a. Into the room walked my brother Jack (St 31) b. On the table was put a valuable book

c. Down the stairs fell the baby. (Rizzi & Shlonsky 2006: 2; 1 from Stowell 1981)

As for embedded clauses in those languages allowing embedded V2, temporal items trigger V2 structures only when they are focused, as in Badiot:

(8) Val Badia Rhaeto-Romance

a. Al m a dit c DUMAN va-al a Venezia.

he me has told that tomorrow goes-he to Venice 'He told me that he is going to Venice tomorrow.'

b. *Al m a dit c duman va-al a Venezia.

he me has told that tomorrow goes-he to Venice

(Benincà & Poletto 2004: 66; 44)

This section investigates when temporal or locative items occur as the leftmost item in V3 contexts. In the recent literature such a pattern have been investigated in studies on microvariation between Standard German and Urban Vernacular varieties of German, like Kiezdeutsch (see Alexiadiou et al. 2017; Lowell Sluckin 2017, 2018 and Walkden 2017, which excludes an analysis in terms of syntactic transfer):

(9) Kiezdeutsch

morgen ich geh arbeitsamt tomorrow I go job.centre

‘Tomorrow I will go to the job centre.’

(Wiese 2009: 787)

Finally, temporal items can both trigger V2 and occur optionally in V3 structures in Swedish as proposed by Bohnacker (2006: 454), in (10).

(10) Swedish

a. sen gick han then went he b. sen han gick

then he went

(Bohnacker 2006: 454)

Mainstream analyses adopt Poletto’s (2002) and Benincà & Poletto’s (2004) idea that in V3 orders the leftmost item is base generated in the Left Periphery76 in a dedicated functional layer called FrameP, hosting HangingTopicP and SceneSettersP. Using English as a meta-language.

(11) Setting the Frame

a. [FrameP Yesterday [SpecForceP we [Force° visited [IP a museum]]]]

b. [FrameP Last Summer [SpecForceP we [Force° visited [IP a museum]]]]

c. [FrameP In Siena [SpecForceP we [Force° visited [IP a museum]]]]

d. [FrameP In 2002 in Siena [SpecForceP we [Force° visited [IP a museum]]]]

e. [FrameP When we went to Siena, [SpecForceP we [Force° visited [IP a museum]]]]

76 V3 orders without postulating base generation:

Leftmost item in FrameP

[FrameP XP1 [Topic/FocusP XP2 [Topic°/Focus° V3]]]

“Big XP” movement [TopicP [XP1 [XP2]] [Topic°/Focus° V3]]

Or

Different Verbal Landing Site

the bottleneck-effect is “qualitatively” enlarged a. [TopicP XP1 [IP [SubjP XP2 [Subj/T ° V3]]]

b. [TopicP XP1 [IP [TopicP XP2 [Topic ° V3]]]

5.1.0. Materials and Methods

A list of V3 orders to be observed is in (12). XP1 represents the leftmost element and XP2 the element close to the inflected verb in V3 structures.

(12) XP1 XP2 V3

Temporal Adverb Subject Inflected Verb Temporal PP Non-Subject

Temporal CP Locative Adverb Locative PP77

In detail, the category Subject/Non-subject is further split in sub-types, as given in (12’).

(12’) Subtypes of subjects, non-subjects and inflected verb

Type of item Label

Subject  DP subjects SUBJDP

Pronominal Subjects SUBJPRO

Impersonal subjects (German man) SUBJIMP

Expletive subjects (German es) SUBJEXPL

Non-Subject  DP Argument/Object OBJDP Pronominal Argument/Object OBJPRO

DP/PP Oblique OBLDP/PP

Pronominal Oblique OBLPRO

Wh-bare items WHBARE

Wh-complex items WHCOMPLEX

Why WHY

Focus FOCUS

77 It is impossible to observe locative CPs, since they are either temporal or peripheral (cf. Haegeman 2006) CPs.

Force/Imperative FORCE

Speaker oriented/Speech act adverbs ADVSPEACT

Higher adverbs of Cinque’s hierarchy ADVHIGH

Lower adverbs of Cinque’s hierarchy ADVLOW

Temporal Adverb TEMPADV

Auxiliary Predicative AUXPRED

Auxiliary Passive AUXPASS

As for Standard German, grammaticality judgments were asked, thus the table shall show if the structure is grammatical or ungrammatical. As for other V2 languages, I could only rely on the attestation within corpora or literature78.

(13) Methods

Standard German Other V2 languages

V3 Structure */ok attested

78 The literature review involved for  West Germanic Standard German (S. Müller 2013; Meinunger 2004), Kietzdeutsch (Wiese 2009; Walkden 2017), Cimbrian (Bidese et al. 2016, Grewendorf & Poletto 2011, 2012), Bavarian (Grewendorf, & Weiß 2014), West Flemish and Standard Dutch (Haegeman &

Greco forthcoming). Diachronically: Old English (Haeberli 2002).  Scandinavian Norwegian (Nilsen 2003, Wiklund et al. 2007), Tromsø-Norwegian (Westergaard, Øystein,, Lohndal 2012), Swedish (Bohnacker 2006), Icelandic (Þrainsson 2007), Urban Vernaculars of Danish, Norwegian and Swedish (Walkden, 2017).  Non-Germanic Kashmiri (Bhatt 1999, Manetta 2011), Italian Rhaeto-Romance (Poletto 2002; Casalicchio & Cognola 2016), West/Central Grisons Swiss Romansh (Corpus SMS4Science http://sms.linguistik.uzh.ch/, Anderson 2005, 2016), Engadine Valleys Swiss Romansh (Oetzel 1994, Fuß 2005).

The table in (14) summarizes the results. The reader is referred to appendix for Standard German, whereas the analysis of V3 orders in V2 languages other than Standard German is given in section 5.1.2.

(14) V3, Time and Space: Standard German and attestation in other V2 languages

As in (14’), these orders are all ungrammatical in German, but some of them are attested in other V2 languages. Non-German V2 languages are marked in grey, since no exhaustive work can be done here: the research is based on the existing literature, which I shall discuss in section 5.1.2. The next subsection deals with all the cases discussed in the literature for German V3 orders, but, as I shall show, they turn out to be “superficial79” V3 orders.

5.1.1. On German superficial V3 in the literature involving FrameP

This section is mainly based on work in S. Müller’s (2013) corpus analysis, in which he discusses V3 orders. Naturally, I do not discuss the clear cases of superficial V3 orders

79 The adjective superficial is in inverted commas because it is intended to be a pun, because 'superficial', according to the Longman dictionary (Longman Dictionary of Contemporary English, (1978, 2003) Pearson Education Limited, page 1665), means ''seeming to have a particular quality, although this is not true or real'' and, as a wider spread usage, ''not studying or looking at something carefully and only seeing the most noticeable things''.

involving an XP embedded in a XP, but those which can be considered borderline, giving (as close as possible) a cartographic explanation.

5.1.1.1. Temporal Adverb– Subject – Verb

(a) TEMPORALADV SUBJDP-AUXPASS

S.Müller’s (2013) only occurrence with a Temporal Adverb preceding a subject in a V3 order is in (15).

(15) Standard German

[Jährlich] [14 Millionen Tonnendes Treibhausgases Kohlendioxid] würden durch [Annually] [14 million tons of greenhouse gases carbon dioxide] would by die Windparks der Atmoshäre erspart

the Wind.parks of the Atmoshere spared

“Annually 14 million tons of greenhouse gases carbon dioxide would be spared by the wind parks of the Atmoshere”

(S. Müller 2013: 5; 3a)

In example (15) jährlich ‘annually’ is an adverbial of the XP, therefore I consider it embedded in the DP.

5.1.1.2. Temporal Adverb– ¬Subject – Verb

(b)TEMPORALADV OBJDP-MODAL

(16) Standard German

[Übermorgen] [das Spiel gegen Kaiserslautern] würde ich gern live sehen Day-after-tomorrow the game against Kaiserslautern would I gladly live watch

‘I would like to see the game against Kaiserslautern live’

(S. Müller 2013: 10; from Abb 1994: 21)

The only occurrence in S.Müller’s (2013) corpus can be found in (16). It can be conceived as only one XP, with übermogen as a DP-internal element (David Gerards p.c.) of das Spiel gegen Kaiserslautern, roughly meaning the ‘match of the day after tomorrow against 1. FC Kaiserslautern’, a German football team.

(c)TEMPORALADV LOCATIVEADV -MODAL

(17) Standard German

Möglichst lange innen wollte sie laufen, so hatte sie es sich vorgenommen [embCP]

As long as inside wanted she run, so had she it planned

“As long as she wanted to run inside, she had planned”

(S. Müller 2013: 15, 33) (d)TEMPORALADV LOCATIVEPP-LEXVERB

(18) Standard German

Derzeit nur am Mac funktioniert die Hardwarekalibrierung via DCC/CI-Signal Currently only on the Mac works the hardware calibration via DCC / CI signal

über das VGA- oder DVI-Kabel via the VGA or DVI cable

“Currently only on the Mac, the hardware calibration via DCC / CI signal works via the VGA or DVI cable”

(S. Müller 2013: 15; 31) The best analysis concerning (17) and (18) is to propose that an entire chunk of XPs (or a cluster of adverbs as will be shown in 26, 27, 28) made of the items in Schweikert’s (2005) Temporal and Locative slots, creating a V2 structure. Schweikert’s (2005; see also Takamine 2010 for Japanese) propose an enrichment of Cinque’s (1999) hierarchy, adding a hierarchy of complements.

(19) Schweikert (2005)’s hierarchy

TEMPORAL > LOCATIVE > COMITATIVE > BENEFACTIVE > REASON >

SOURCE > GOAL > MALEFACTIVE > INSTRUMENTAL > MATTER > MANNER (Schweikert 2005: 132)

My hypothesis can be summarized in (20). A chunk of XPs belonging to the functional projection containing circumstantial elements (in 20, CircP) moves to the LP.

(20) Movement of CircumstantialP containing the Schweikert hierarchy

[TopP Temp+LocativeXP [Top°INFL [IP SubjXP [CircP <[Temp TempXP [Loc LocXP]]]>]]]]

Further evidence for Schweikert’s hierarchy comes from examples from West Flemish, here shown in (21) and (22).

(21) West Flemish

In Gent ze weundige by eur broere.

In Ghent she lived with her brother

‘When she was in Ghent, she lived with her brother.’

(Haegeman & Greco forthcoming: ex. 10d) (22) West Flemish

*In Gent ze weundige verleden joar.

In Ghent she lived last year

‘In Ghent she lived last year.’

(Haegeman & Greco forthcoming: 10a) Sentence (22) might be out because the locative complement crosses a higher temporal adverb, whereas, by eur broere ‘with her brother’ is lower than the locative, and thus sentence (21) is grammatical. A sentence like (22) can be compared to (23), a typical case of locality violations in terms of fRM for Cinque’s hierarchy (1999) of adverbs. French adverb Calmement ‘quietly’ is lower than probablement ‘probably’ and therefore it cannot cross it.

(23) French

a. Jean s’est probablement dirigé calmement vers la sortie.

‘Jean has probably moved quietly toward the exit.’

b. *Calmement, Jean s’est probablement dirigé vers la sortie.

‘Calmly, Jean has probably moved toward the exit.

(Rizzi 2004: 235, 33)

5.1.1.3. Temporal PP – Subject – Verb Such an order is not attested in the literature.

5.1.1.4. Temporal PP– ¬Subject – Verb (a) TEMPORALPPOBJECTDP-LEXVERB

(24) Standard German

Zum zweiten Mal die Weltmeisterschaft errang Clark 1965 For the second time the world-championship reached Clark 1965

Zum zweiten Mal die Weltmeisterschaft errang Clark 1965 For the second time the world-championship reached Clark 1965