• Keine Ergebnisse gefunden

4. Analysis

4.1 Previous research

4.1.2 On “Stylistic Fronting”

4.1.2.2 Syntactic properties

Recall the definition of “Stylistic Fronting” given by Maling (1980/1990): she defines occurrences of fronting in embedded clauses in Icelandic, where elements appear to be moved to unfilled subject positions, as “Stylistic Fronting”. As we have seen in the introductory

136 The matter is taken up in the chapter on the analysis of our proper data.

137 Analyses of the type of Cardinaletti (2003) for Italian, who observes the fronting does not create a marked word order but has a literary or formal flavour, are not taken into consideration.

remarks to the present sub-chapter, the studies on “Stylistic Fronting” in Medieval French focus on different data and thus feature in different syntactic environments, e.g. in embedded and in main contexts. Accordingly, the basic syntactic properties that are claimed to characterize

“Stylistic Fronting” vary considerably. Bearing this problem in mind, here is an overview in the present section of the syntactic features distinctive of “Stylistic Fronting”. The review of the current research concentrates on factors relevant to “Stylistic Fronting” in Medieval:

whenever useful, there is a recourse to findings on other languages.

Clause-boundedness

Since Maling’s (1980/1990) article, one of the important properties of “Stylistic Fronting”, as distinct from Topicalization, is the fact that the former cannot occur over a long distance, i.e.

outside the respective clause. With respect to the clause-boundedness of “Stylistic Fronting” in Medieval French, one notices that this criterion seems to be of less relevance to the different authors. Cardinaletti and Roberts (1990/2002) and Labelle and Hirschbühler (in press) do not address the issue, Salvesen (2011) mentions it only briefly without discussing it, while Labelle (2007) notes that she cannot find any relevant evidence. Mathieu (2013) takes the issue not to be relevant to the discussion at hand. However, in his article from 2006, he states that “Stylistic Fronting” is clause-bounded. Finally, Fischer (2010) finds no evidence for “Stylistic Fronting”

implying movement outside an embedded clause, as shown by (45) retaken here as (47), in contrast to occurrences exhibiting topicalization as in (48); examples taken from Fischer (2010:

121).

(47) Une espee rasauda qui brisieei est ti en deus moitiez a sword sharp which brokeni is in two halves

‘a sharp sword that is broken into two halves’ (1238, 13, conperc) (48) Car enemisi pense il bien que ce soit ti

because enemies believe he well that it be

‘because he believes that there are enemies’ (La queste de Saint Graal 112)

For the sake of completeness, “Stylistic Fronting” in Medieval French can even be combined with long-distance movement, compare (49), taken from Labelle and Hirschbühler (in press, their (68)).

(49) [en sarraguce]i sai ben qu’alerj m’estoet tj ti

[in Sarraguce]i know well that.go me.is-necessary I know well that I must go to Sarraguce’ (Roland, 23.283)

While aler is stylistically fronted to the left edge of the embedded clause, en sarraguce is topicalised and moved out of the embedded clause to the left periphery of the whole complex clause.

Locality conditions

According to Maling (1980/1990), “Stylistic Fronting” in Icelandic seems to be governed by the following accessibility hierarchy in (50), taken from Maling (1980/1990: 81).

(50) ekki (‘not’) > predicative adjective > past participle/verbal particle

Further research on Icelandic has criticised this hierarchy based on the following observations (see for instance the synopsis in Holmberg 2006: 539-540). First, not only negation, but also sentence adverbs can block the movement of other elements. Second, participles other than those of vera ‘be’ and hafa ‘have’ do undergo “Stylistic Fronting” instead of predicative adjectives contained in a small clause. Third, the equal accessibility of a verb and a verb particle are assumed to represent the more general idea that a head and its complement are equally accessible. As a consequence, the above mentioned observations were reconsidered in light of the work on locality conditions on movement, namely the Head Movement Constraint and Relativized Minimality (Jónsson 1991) and the Minimal Link Condition (Holmberg 2000).

Recent work on Medieval French is essentially based on the latter, which is retaken in (51), compare Chomsky (1995: 355-356).

(51) Minimal Link Condition (MLC)

A feature F attracts the closest feature that can check F. Closeness is defined in terms of c-command:

in a configuration [α … β … γ] where α c-commands β and γ, β is closer than γ to α, if β asymmetrically c-commands γ.

The Minimal Link Condition predicts that the closest element is stylistically fronted and hence explains the above shown observations on Icelandic (for more details, see Holmberg 2006).

With respect to Medieval French, once again, not all authors address the issue (Cardinaletti and

Roberts 1990/2002). Salvesen (2011) more generally comments on the accessibility hierarchy but does not relate it to her Medieval French data. Fischer (2010) states that “Stylistic Fronting”

in Medieval French respects the Minimal Link condition. Her examples (2010: 119-120), retaken here in (52), (53) and (54) illustrate different observations in this matter.

(52) Que hautementi avoit ti commenciee chevalerie that highly have started knighthood

‘that he has greatly started knighthood’ (1225ca, queste)

As (52) shows, the adverb hautement is stylistically moved but not the participle commenciee.

Fischer (2010) points out that this is always the case in contexts with an adverb between the finite verb and the infinitive or between the finite verb and the past participle. Furthermore, she indicates that due to their equidistance to the triggering head, the past participle and the complement can equally undergo SF (53) and (54) .138

(53) Qui nommeei est ti virgene marie who named is virgin Mary

‘who is called Virgin Mary’ (1227ca, mir)

(54) Qui [tel juge]i avez estably ti aussi comme en enseignement who [such judge] have established also like in education

‘who have established such a judge also for educating’ (1280, abe)

I do not share her assumption that the same effect can be seen with negation since the examples she provides do not show the effect of equidistance of negation and complements or past participles. Mathieu (2006) explains examples such as (54) by the Minimal Link Condition in (51) as well. In his data he observes that double “Stylistic Fronting” constructions are possible in Medieval French, namely a head and a phrase at the same time but never two heads or two phrases.

In contrast, Labelle (2007: 306) gives the examples in (55) and (56) in order to explain that in her data, there is no accessibility hierarchy.

138. Fischer (2010) does not compare (53) and (54). However, the examples, on which she has recourse to illustrate the equidistance do not contain fronting of a past participle instead of a complement.

(55) Huelin dist une novele qui a Gorm[un]d ne fut pas bele Huelin said a piece-of-news which to Gormund NEG was not good

‘Huelin brought news which didn’t please Gormund’ (Gormont, 239–240) (56) Se de ce vous volez desfendre, alez en tost voz armes prendre

if of this you want to.defend, go.2PL LOC soon your weapons take

‘if you want to defend yourself, go get your weapons now’ (Thebes1, p. 96) According to her, Medieval French does not obey the accessibility hierarchy since in (55) the PP a Gormund is fronted instead of the negative adverb pas. Similarly, in (56), the PP de ce is raised instead of the infinitive desfendre.

Mathieu (2013) contradicts Labelle (2007). With respect to (56) , he argues that the PP de ce and the infinitive desfendre are sisters. Hence the latter does not asymmetrically c-command the former and both are equidistant to the target position of “Stylistic Fronting”. Furthermore, he assumes, as illustrated in the following example in (57) (2013: 339), that a) Medieval French negation markers as pas, mie, and point are heads and not phrases, b) that the fronting of two elements targets two different positions, and c) that they do not compete with each other.

(57) Quant la pucele le salue, qui sa bouchei pasj when the young girl him salute who his mouth not

n’en palue tj ti ne ne li a neant costé.

neg.en turn-white neither NEG to-him has nothing cost

‘The young girl’s greeting which was not unpleasant did not cost him anything.’ (Chevalier à la Charrette, 1570-1573)

Regarding (55) then, one can assume that only the phrase has been moved not the head.

To this idea, Labelle and Hirschbühler (in press) object that Icelandic ekki ‘not’ blocks “Stylistic Fronting” of heads and phrases as illustrated in (58) - (61) , their (20)-(23).

(58) þeir sem hafa ekki búið í Ósló those that have not lived in Oslo (59) þeir sem ekki hafa búið í Ósló

those that not have lived in Oslo

(60) *þeir sem búið hafa ekki í Ósló those that lived have not in Oslo (61) *þeir sem í Ósló hafa ekki búið

those that in Oslo have not lived

Accordingly, in (60) , the fronting of the head búið ‘lived’ is ungrammatical, while in (61), the fronting of the PP í Ósló yields an ungrammatical sentence. Furthermore, Labelle and Hirschbühler (in press) note that the “Stylistic Fronting” of a participle is not blocked by a high adverb, compare (62) and (63), their (26) and (27).

(62) car recuvréi sunt veirement ti

since found are certainly

‘since they have certainly been found’ (Quatre livre, 18.590) (63) car demureti i unt trop ti

since stayed LOC have too-much

‘since they stayed there too long’ (Roland, 136.1832)

Finally, they observe that Medieval French also contrasts with Icelandic in the possibility of fronting secondary predicative adjectives (64) and (65).139

(64) Blanche avait la barbe

‘white had the beard’

(65) *madurinn sem rikur er talinn

‘the man that rich is considered’

Hence, they come to the conclusion that the accessibility constraints typical of Icelandic

“Stylistic Fronting” are not operative in Medieval French.

Subject gap

Maling (1980/1990) specifies that a subject gap is necessary for the application of “Stylistic Fronting”. For Icelandic, she states that there are three possible origins of a subject gap: a) the

139 Examples taken from their footnote 6.

extraction of the subject NP, i.e. by relativization, question formation, comparative clause formation or clefting; b) the usage of impersonal passives or lexically impersonal predicates;

or c the postposing of indefinite NP. For Medieval French, generally assumed to be a null-subject language (Vance 1997), the null-null-subject property represents a fourth possible origin of a subject gap. Hrafnbjargarson (2004) attenuates the idea of a strict subject-gap condition: he observes that “Stylistic Fronting” of a head to the right of a subject pronoun seems to be marginally allowed in Icelandic.140 For Medieval French, the existence of a subject-gap condition appears to be debatable. While some authors, as, for instance, Cardinaletti and Roberts (1990/2002) and Salvesen (2011), follow Maling’s (1980/1990) idea of a subject gap, other authors call it into question as a defining property.141

Fischer (2010, 2014) shows that “Stylistic Fronting” in Medieval French is attested together with subject pronouns and full subject DPs, compare (66) and (67), taken from Fischer (2010: 121-122).

(66) Et quant tu arouséi l avras ti d uille and when you gained it have of.the oil

‘and when you have gained oil’ (1300 ca, mace) (67) Et sa puissance moulti est ti povre

and his power very is poor

‘and his power is very poor (1200,bodo)

Consequently, Fischer (2010) concludes that the subject-gap condition is not active in Medieval French. Note that Fischer’s (2010) examples containing a subject and a stylistically fronted element all are of the order ‘subject – fronted element – finite verb’ (SXV).

For Mathieu (2006, 2013), the subject-gap condition is operative in Medieval French. However, recall that Mathieu (2006) assumes that two elements, namely a phrase and a head, can be stylistically fronted. With respect to the subject-gap condition, he sticks to the distinction

140 Hrafnbjargason (2004: 117) states that the speakers tend to more easily accept the coexistence of a subject pronoun and a stylistically fronted element if the former is cliticised onto the complementizer.

141 Labelle (2007) is not mentioned in what follows since she discusses the V2 in embedded clauses more generally and, with respect to the presence of a subject, gives only examples of postverbal subjects (Labelle 2007: 305-307). Since Mathieu (2013) comments about Labelle’s (2007) criticism on his article of 2006, he does not address the issue, either, and he is hence disregarded in what follows.

between phrases and heads. While the fronting of phrases is said to be blocked by the presence of an overt subject, the coexistence of a subject and a stylistically fronted head appears to be possible (68) , taken from Mathieu (2006: 251).

(68) L’an m’a conté ce poise moi que partiri vos volez The.one me.has told this saddens me that to.leave you want del roi ti

of.the king

‘I’ve been told – this saddens me – that you want to leave the king.’

(Chevalier à la Charrette 141-142)

In contrast to Fischer’s (2010) examples, the order here is ‘fronted element – subject – finite verb’ (XSV). This alludes to Labelle and Hirschbühler’s (in press) view on the subject-gap condition. In their data they find that “Stylistic Fronting” co-occurs with a preverbal subject in 11% of the cases. Both combinations are attested in their data but the type SXV is much more frequent (9% of the overall data). Furthermore, Labelle and Hirschbühler (in press) observe that the order XSV is almost exclusively restricted to pronominal subjects. Hence, they take the SXV order to be the unmarked order.

To sum up, the relevance of the three syntactic properties of “Stylistic Fronting” presented here have been analysed in different ways. While the clause-boundedness is uniformly taken to be relevant for “Stylistic Fronting”, the existence of locality constraints or the subject-gap condition is strongly debated. The discussion of syntactic properties is here limited to those three aspects, which appear to be subject to purely descriptive adjudication, i.e. presence and/or absence in the respective data. The next section deals with the different analyses that are themselves based on theoretical assumptions and hence might be subject to the different concepts used by the various authors.