Abstract
In this chapter I offer an account of the word order of main clauses in Old French (OFr), focussing largely, though not exclusively, on verb-second effects. I hope to shed light on some longstanding issues in the analysis of OFr syntax, some of which have implications for later stages of the language as well. The chapter is divided into two main parts. Part I (sections 3.2., 3.3., and 3.4.) is devoted to issues of V2 in general, while Part II introduces an independent property of OFr, the A’ nature of Spec, AgrP, and its interaction with the V2 grammar (sections 3.5 and 3.6.). In sections 3.2. and 3.3. I propose (as in Vance (1988a) but contra Adams (1987a,b), Vance (1993), and Roberts (1993)) that while non-subject initial V2 clauses are CP’s, SVX clauses are IP’s. Arguments are given from both syntax and discourse. In section 3.4. I document evidence that in this otherwise strictly V2 language, the beginnings of a competition between CVS and CSV word order — i.e. the decline of V2 — can be seen in the variable behavior of fronted adverbial clauses. From a diachronic perspective, these first two points taken together make it possible to see in OFr the kind of solidly verb-second but potentially unstable system that has characteristically been attributed to Middle French (cf. Adams (1987a,b), Lemieux and Dupuis (1990, (1995)); however, the full development of a diachronic analysis is reserved for Chapters 6 and 7.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Notes
Here I am abstracting away from the cliticization of subject pronouns to C° in embedded clauses, which sometimes creates a surface order unique to subordinates (see section 4.2). Such a difference in word order has no bearing on the choice between IP and CP structures for main clause SVO.
It must be noted here that Adams (1987b) offers some ancillary arguments for the reanalysis of SVO as IP in late Middle French. See Chapter 7 for discussion.
In Early OFr, the postverbal subject pronoun could be separated from the finite verb by an object pronoun in certain types of constructions, such as direct yes/no questions. This word order was still possible in 13th century prose; see the discussion in Chapter 4, footnote 5.
For discussion of the possibility that OFr subject clitics adjoin to the left of IP, as argued for (object) clitics in a variety of languages by Taylor (1990), Pintzuk (1991), Fontana (1993), and Kroch and Taylor (1995), see Chapter 4, section 4.2.
If Early OFr turns out to be a symmetric V2 language, then the same observation holds true but the projection at issue is IP, not CP.
This conclusion is based on an examination of the examples in the Queste and in Skarup’s (1975) chapters 6 and 7.
See Franzen (1939: Chapter VII) for additional discussion of fronted subordinate clauses and their effect on a following main clause.
Roberts (1993) claims that subjects cannot be in Spec, TP because Spec, TP is not an A-position. As far as I can see, his claim is motivated only by the need to prevent postverbal subjects in English when an expletive fills Spec, IP: (i) *There has a man loved Mary. An analysis under which (i) is ruled out by the Case Filter would seem to be equally plausible; case is generally assigned by Infl to the left in English. Roberts rejects this analysis because of the existence of residual V2 constructions in which case assignment appears to be to the right: (ii) Who has Mary loved? However, it is also possible that the NP Mary is assigned nominative case leftward by the trace of the verb in Infl, as in the analysis presented in section 3.6.
The empty category in Spec, IP may be either an intermediate trace of the subject or, as argued in section 3.6., the trace of the XP in Spec, CP.
An alternative means of handling this type of word order puzzle is given for Italian by Rizzi (1991), who proposes the following adjacency condition between a subject generated on the right of VP and T°: (i) Postverbal subjects are adjacent to T° or to the past participle morphology. Such a system is implicitly assumed for Old French by Roberts (1993), who hypothesizes that T° is the assigner of nominative case in the inversions under consideration here. Rizzi (1990b) does not spell out how this condition would be fulfilled or specify at what level it applies. Presumably, NP and PP VP-complements would move to the right of the VP, leaving the VP-final subject linearly adjacent to T° although not structurally adjacent. A heavy subject would be exempt from the adjacency requirement and could thus remain VP-final.
Adams (1988), Dupuis (1989), and Roberts (1993) suggest that C° may be the assigner of nominative case in some constructions in medieval French but do not adopt case-assignment by C° as a general characteristic of Old and Middle French or of V2 languages.
A possible exception involves languages that are in the process of losing the verb second constraint; cf. Chapter 6.
The type of chain government of underlying direct objects discussed here cannot be the only means of licensing nominative case on the postverbal arguments of unaccu- sative and passive verbs in OFr, however, since they are also possible in embedded clauses where the chain government condition, as formulated here, does not hold. See Chapter 4.
The decision to formulate this property in terms of chain government grew out of conversations with Rex A. Sprouse and Beatrice Santorini.
Thanks are due Beatrice Santorini (p.c.) for this suggestion. My use of Rizzi’s typology differs from that of Santorini (1995) in that I consider that the “dominant” head (in Santorini’s terms) is not always projected, resulting in default case licensing by a lower head.
Discussions with Rex A. Sprouse, Laurent Dekydtspotter, Jan Wouter Zwart, Beatrice Santorini, Richard Janda, Christer Platzack, Anders Holmberg, Sten Vikner, Charlotte Reinholtz, and Rajesh Bhatt helped clarify many aspects of this section. None of these people necessarily agrees with the outcome, however.
For analyses of ModF subject-clitic inversion in which V is claimed to move only as far as 1°, and not (overtly) to C°, see for example Noonan (1989), Drijkoningen (1990), and Sportiche (to appear).
Kroch and Taylor further claim that, in Old English, topics and wh-phrases move to Spec, CP but finite verbs move only as far as Agr°. This is their answer to a problem posed by Pintzuk’s (1991) work, i.e. that the position of Old English pronominal clitics argues for its status as a symmetric V2 language while the lack of V2 inversion in embedded clauses in the texts simultaneously argues against this conclusion. This account differs from the one proposed here for Old French, in which we claim that verbs move to C° in main clause V2 inversions. That the two languages are different in exactly this respect is suggested by a comparison of subject clitics in the two languages: (i) Aelc yfel he maeg don (WHom, 4.62) each evil he can do (ii) Car vilanie ne feroit il pas au vaslet (Q 89,10) (for) evil neg would-do he not to-the young-man ‘for he would not do evil to the young man’ If subject clitics in both languages occur between C° and the left edge of IP, but the verb moves as far as C° only in OFr, then the above contrast is expected.
Several linguists have pointed out to me that the pronominal version of (108)b, given in (i), is much worse than the nonpronominal version: (i) *Have not they finished their homework? These judgments parallel the facts of the older language, in which pronouns were clitics on the verb. The retention of this contrast in Modern English, albeit in diluted form, may be reinforced by a tendency to interpret the scope of the negative differently in (108)b and (109)b.
Note that this fact is also problematic for a strict version of Santorini’s (1989) or Diesing’s (1990) analyses of Yiddish, in which postverbal subjects are assumed to be in Spec, VP. If such subjects are in Spec, TP, as proposed here, the problem can be solved if the negative adverb can be shown to be adjoined to VP rather than TP. Beatrice Santorini (p.c.) suggests an alternate possibility, that the negative element nit in Yiddish is not a typical adverb but rather that subject-negation order in Yiddish is determined by scope assignment.
For example, Bobaljik and Jonas (1996) discuss the status of Spec, TP in Icelandic. First, they note evidence that the negative ekki may indeed occur lower than Spec, TP, given the assumption that object raising in (i) involves movement out of the VP: (i) Jolasveinarnir borucfu buingin [VP ekki tj the. Christmas, trolls ate the.pudding not They also argue that postverbal subjects in Transitive Expletive Constructions may appear in Spec, TP: (ii) Pa£ boruu sennilega margir jolasveinar bjugun there ate probably many Christmas, trolls the.sausages Furthermore, Bobaljik and Jonas claim that the following Germanic languages allow Transitive Expletive Constructions and thus license Spec, TP as a subject position: Yiddish, Frisian, Dutch, German, (one dialect of) Faroese, and Icelandic. It appears then that we have independent evidence that this position is licensed in the two languages for which we make a clear prediction, Yiddish and Icelandic, and that it is not licensed in the mainland Scandinavian languages, also as predicted. Significant issues remain, however, which require additional research before an adequate comparative study may be made. First, Bobaljik and Jonas claim that Spec, AgrP is also licensed as a (referential) subject position in TEC’s. This claim, if true, runs counter to the predictions of the current analysis; further study is needed to see if their claim holds under my assumptions about clause structure. Second, the inclusion of German and Dutch in the list of languages allowing TECs indirectly implies for us that they have V to I movement. Recall from Chapter 2 that this property of German and Dutch cannot be directly observed but must be determined on theory-internal grounds; a den Besten-style analysis postulates V to I movement in these languages while a Minimalist analysis such as Zwart (1993) polstulates no V to I movement. Bobaljik and Jonas, who adopt a Minimalist perspective, avoid making the claim that German and Dutch have V to I movement like Yiddish and Icelandic by linking the possibility of TECs directly to verbal inflection, which is rich (by their criteria - but see footnote 29) in German and Dutch. Our account, on the other hand, would be forced to a non- Minimalist stand on the issue if the possibility of TECs were found to correlate directly with the status of Spec, AgrP.
See Suner (1994) for an alternative point of view.
The exceptional behavior of some dialects of Norwegian with respect to this generalization has been discussed by e.g. Schwartz and Sprouse (1996) and Holmberg and Platzack (1995:25). The Troms0 dialect behaves like Kronoby Swedish, while the Halingdalen dialect has strong verbal agreement features but no V to I movement. Holmberg and Platzack are able to account for the Halingdalen facts under their analysis.
Interestingly, however, Rohrbacher’s specific proposal for determining the presence of strong Agr in a language may make a more accurate prediction for the history of French than do other inflection-based accounts. Under Rohrbacher’s criteria, Old and Middle French would be considered to have rich agreement paradigms because (a) they make a three-way distinction in at least one number (the plural) of the most common paradigms and (b) the infinitive is distinct from both these (plural) first and second person forms. However, once the /R/ ending of the -er infinitives is lost in late Middle French, the infinitive falls together with the second person plural. If it could be shown that subject clitics became agreement markers at that time, it would be possible to maintain that French has always been a language with rich agreement, thus avoiding the incorrect prediction that V to I (and, in my system, the nature of Spec, AgrP) should be unstable. I will not pursue this avenue here for two reasons. First, it would be impossible to show definitively that late Middle French possessed the relevant kind of subject clitics; Julie Auger (p.c.) notes that the written evidence for her account of Popular French dates back no further than the present century. Second, I find the definition of strong Agr provided by Vikner (1995b) (discussed in the text) more intuitive than Rohrbacher’s.
Schwartz and Sprouse (1996) argue that there is no direct relationship between the form of verbal or nominal paradigms and the presence/absence of movement. According to Zwart’s (1993) Minimalist analysis of V2 effects, both German and Dutch must be languages without V to I movement (except as an intermediate step in movement from V to C). Yet only Dutch has the predicted non-rich verbal agreement, according to either Vikner’s (1995a) or Rohrbacher’s (1994) criteria. These same two languages present problems for the movement of NP’s; Dutch has no overt case marking and German has quite rich case-marking, but both languages have overt movement of direct objects to Spec, AgrOP. These facts suggest that the correlation between morphological richness and movement should be left at a very abstract level.
But cf. Drijkoningen (1990) for an approach in which the pre verbal subject position is empty.
Roberts, on the other hand, finds support from the analysis of Medieval Italian dialects proposed by Poletto (1991, 1995).
Bailard (1982) points out that (121) is possible with a heavier subject: (i) Il dit qu’ont été acceptés tous les candidats qui se sont présentés ce jour-là. he says that-have been accepted all the candidates who refl are presented this day-there ‘He says that all the candidates who applied that day have been accepted.’ In our terms, (i) is an instance of Heavy Inversion; the preverbal position is occupied by the trace of the subject adjoined to the right of IP.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1997 Springer Science+Business Media Dordrecht
About this chapter
Cite this chapter
Vance, B.S. (1997). Main Clause Word Order in Old French. In: Syntactic Change in Medieval French. Studies in Natural Language and Linguistic Theory, vol 41. Springer, Dordrecht. https://doi.org/10.1007/978-94-015-8843-0_3
Download citation
DOI: https://doi.org/10.1007/978-94-015-8843-0_3
Publisher Name: Springer, Dordrecht
Print ISBN: 978-90-481-4886-8
Online ISBN: 978-94-015-8843-0
eBook Packages: Springer Book Archive