Abstract
In this chapter, we want to introduce issues of bounding in syntax and the far more controversial issue of bounding in LF. Starting with groundbreaking work by Huang, Kayne, Pesetsky and others, the notion of extraction domain has gained increasing attention in linguistic theory over the past decade. The basic idea in this work is that extraction is only possible from domains which are governed, and that it is in addition constrained by tree geometric properties. Pesetsky (1982) proposed that paths of movement may overlap but must not intersect. More relevant for our concerns is Kayne (1983) where directionality of government is considered as a parametrical option in syntactic theory the chosen value of which (partially) determines extractability. The idea is that a head X can start a g(overnment) projection along which extraction from its minimal maximal projection XP can be licensed. This may take place when a g-projection can be built. In a language like English, where the head takes its complement to the right, a g-projection can be built along the lines of positions which are on the right branches with respect to a governing element. Thus, extraction from object position and positions inside an object position is predicted to be licit. Extraction from the subject position, however, is not possible because a g-projection cannot be built beyond the subject position, since the subject is on a left branch. With this machinery Kayne could derive hitherto unexplained subject/object asymmetries in English and in the Romance languages. Chomsky’s (1986a) Barriers-system is in this spirit insofar as every maximal category is taken to be a potential bounding node (a barrier) for syntactic movement and that there are several options for circumventing barrierhood. It differs from Kayne’s work, however, in not taking directionality into account. Chomsky’s theory is largely based on θ-government and the availability of intermediate landing sites to avoid the effects of barriers. The classical case of long movement from COMP to COMP is supplemented with an adjunction operation that can apply to XPs which are θ-marked but fail to be L-marked due to the fact that they are not selected by a lexical head. According to Chomsky, I selects and also θ-governs VP but is not a lexical head. VP is then a blocking category (BC) which will turn the next XP up, namely IP, into a barrier. Chomsky provides a mechanism by which the BC-status of VP can be suspended. This mechanism is adjunction to VP. As we shall see, these assumptions do not constrain derivations sufficiently; this becomes particularly clear in languages with a mixed system of governing heads, such as German and Dutch. The work that has assigned movement constraints based on directionality of government a central place in syntactic theory, and, in addition, deals with the intricacies of Dutch and German word order, is Koster’s (1987) theory of Domains and Dynasties. This is the reason why I will take Koster’s work as a starting point for the following discussion.
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
To be precise, Koster assumes DP, PP, AP and CP as bounding nodes, but not VP.
For the present introductory purpose I will not refer to the opacity factor. See Chomsky (1973) and subsequent work up to Chomsky (1981) for details.
The definition given here slightly changes Koster’s original definition (Koster, 1987: 174) in order to avoid a more thorough introduction to Kayne (1983).
This can clearly be seen in V2-clauses where for the computation of the dynasty the verb must appear in its base position.
Sometimes it is claimed that in Dutch tensed CPs cannot appear in the middle field (to the left of V). See Bennis and Hoekstra (1988). For SOV-languages this is certainly not universally true. Bennis and Hoekstra (1989) acknowledge this fact in a footnote. German tensed CPs can in principle appear in the middle field, although there are constraints. We will return to this issue.
With respect to adjuncts, see Huang’s (1982) Condition on Extraction Domain (CED). In Koster’s theory, the CED derives from the fact that phrases which are by their very nature ungoverned cannot enter a dynasty.
Koster (1987: 232)
Tancredi (1990) argues that particle and associated DP do not form a constituent at LF. In this sense, his analysis resembles the one in Jacobs (1983), the difference being that Jacobs did not assume the level of LF while Tancredi does. Tancredi takes particles of the relevant sort as propositional modifiers. This should, however, not be taken to mean that there is always adjunction to IP. If this were the case, we would expect scope ambiguity in cases such as: (i) a. that nobody saw only JOHN in this room b. daß niemand in diesem Zimmer nur HANS gesehen hat
The reading “Only John is such that nobody saw him in this room”, however, is not available. This indicates that each QP raises to its minimally available scope position, - in this case IP and VP respectively. “Propositional scope” means that VP is a complete functional complex including the subject, and as such attachment to VP attains propositional scope.
In section 2.3.1.3 we will turn to QPs in subject position.
The same is true in cases in which the particle attaches to a PP with argument status.
Recall that the particle makes a specific semantic contribution which is compositionally tractable, and that the “trigger” for movement must be built into its lexical semantics by virtue of the involvement of the universal quantifier for only and the existential quantifier for even.
In Chomsky’s (1993: 33) words, “derivations are driven by the narrow mechanical requirement of feature checking only, not by a ”search for intelligibility“ or the like”.
Notice that Chomsky ( 1993: 33) proposes that shorter derivations should be preferred to longer ones, but this can hardly be intended to hold for QR. First of all, QR cannot naturally be couched into the morphology-based checking system; secondly, each target of LF-movement yields a structure with a new interpretation. Therefore these cannot be compared along the lines of the economy metric.
According to Giuseppe Longobardi (p.c.), the same is true for Italian examples involving solo (“only”).
See Longobardi (1991: 187).
See Rizzi (1990) for a discussion of strategies which enable C to acquire features that allow it to serve as a proper head governor of the subject trace.
See Fanselow (1991: ch.2)
If this were the case, it would be hard to see why the scalar interpretation disappears in various contexts, e.g. (i) John can only be an employee r) Only an employee will handle this job (i) does not get the reading that it is possible that John is an employee and employees are low on some scale. It rather means that for all possible situations s, John is an employee in s. This is the reason why (i) is synonymous with the proposition that John must be an employee. Similar considerations hold for (ii).
Our goal is not to provide any new insights into these matters. Thus, we simply call the reader’s attention to the existence of these facts. Altmann (1976) is a rich source for the distinction between quantificational and scalar interpretation with respect to a larger number of German particles (which Altmann calls Gradpartikeln,’degree particles’).
See Bolinger (1972), Jackendoff (1977), Abney (1987) and Corver (1990) among others.
The contextual effect can be seen when we introduce alternative common noun meanings. This Maharajah possesses only a thousand elephants sounds less plausible than This Maharajah possesses only a thousand bottles of champagne. It may be a big thing even for a Maharajah to possess a thousand elephants, while his possession of a thousand bottles of champagne may not be so spectacular.
This is a simplified version of the structure Löbel (1990) proposes on the basis of a number of languages. See also Bhatt (1990: 77).
The same holds for German. While *Die nur DREI Mädchen is almost as bad as the English example, die nur WENIGEN Mädchen (“the only FEW girls”) is good. This suggests that drei and wenigen may not be in the same functional projection, and that wenigen is an XP, while drei is an X°. Of course, the only three girls is grammatical if three does not bear primary stress, as in The only three girls that I know are Susan, Wilma and Mary. The German correspondent here is not nur but the adjective einzig. We will turn to a discussion of adjectival quantifiers in the next chapter.
This will be refined in chapter 3.
This is not so in other cases of extraction from Deg-phrases as Corver shows. Extractions like in How many inches was the board e too long? are possible. As far as I can see, the difference must be connected to the fact that in these cases the Deg-head is lexicalized while it is empty in the ungrammatical cases.
See Bierwisch (1987) where a detailed theory of dimensional adjectives and their interaction with explicit or default values of SpecDegP (in Bierwisch’s terminology “Faktorphrase”) is developed.
I will come to the question how association with focus enters the picture.
They become ungrammatical when the focus is shifted to the noun: *…behind only three GIRLS. In this case, only cannot use the DP as a scope domain because the focus it associates with does not invoke a graded domain of quantification.
Notice the ungrammaticality of (i) *Schwere, hatte Heinz eine e, Verletzung
Rooth’s own cases are not exactly like (60b). They differ insofar as the particle is always attached to VP, not to PP as in our case. We will later show that this difference is inessential for the point to be made here.
More about this in chapter 6
The argument can also be made for English, although the data are more complicated due to the possibility of P-stranding in LF. Consider the following examples which make use of negative inversion (NI) and/or the NPI ever: i) To only WEAKLY gifted students Bill would never talk ii) ??To only WEAKLY gifted students would Bill talk iii) *To only WEAKLY gifted students Bill would ever talk iv) ??To only WEAKLY gifted students would Bill ever talk NI) ?To only MARY would Bill ever talk vi) Only to MARY would Bill ever talk If the scope of only is confined to the DegP, the absence of NI in (i) is expected. Raising the entire DP in LF should make (ii) grammatical. At LF only would then license NI, but the scope of only seems to be fixed at S-structure, and there is no motivation to assume an analysis like [only DP] which would force movement out of PP. (iii) is out because the NPI cannot be licensed by only. NI in (iv) invites an analysis in which [only DP] moves at LF, but as before only is interpretable in situ. Once no such possibility arises, the result gets better as seen in (v); yet the construction is still marked because only MARY has to undergo QR for only to license the NPI. The ideal case is provided by (vi) where NPI-licensing takes place at S-structure. Thanks to Steve Berman for discussion of these data.
Löbner (1987), in developing a theory of natural language quantification, shows that ordinary quantifiers such as some and every can be subsumed under a generalized notion of “scale”. Löbner calls this phase quantification. His theory allows a unified treatment of ordinary determiner type quantifiers and adverbial quantifiers such as already and still. In the present context we will be more conservative and rather treat the scalar use of only as a quantification of the usual kind. It is interesting, however, to see that there is this independently motivated attempt at a unification.
I use a German example because we have seen that there can be no movement from the PP in German.
This is basically what Jacobs (1983: 169ff.) suggests.
For (74b) this is, of course, only true to the extend that the speaker/hearer cannot mentally invoke a scale on which Alceste would naturally range low. It seems to me that this is very hard to do in this case. On the other hand, there is some improvement when common nouns are chosen which can conventionally invoke a scale. Consider: (i) John is a clever guy who can deal with all kinds of people. He can talk to the director of the factory and a. he can talk to only a WORKER (?consistent) b. he can talk only to a WORKER (contradictory)
For the time being I remain neutral about the status of topicalization. It has been argued that topicalization in English is adjunction to IP (see Lasnik and Saito, 1992) or movement to a specifier (see Chomsky, 1977, and more recently Müller and Sternefeld, 1993 ).
As I have already pointed out, what prevents direct interpretation of S-structure in this case is the fact that [PRT XP] is not a semantically usable QP like everybody. We will turn to this problem immediately.
See von Stechow (1993: 26f.).
This does not mean that the QR-analysis employed so far is the only possible option. In chapter 5, I will show in which sense functional head positions can be assumed to achieve what we achieve right now in terms of standard QR.
One problem remains, namely how the particle, which is an X° -element, can be licensed in an adjoined XP-position. For instance, only in (80) violates structure preservation if at LF the PP in the operator position vanishes. Thanks to Chris Tancredi for discussion of this point. I will propose an amendment of the present theory in chapters 5 and 6 which will overcome this problem.
The term “reconstruction” should be understood as shorthand for the copy-and-deletion operation sketched above.
Consider an example like: (i) The only thing you can do is quit yourob It is not quite clear how (i) can be brought within the purview of the two licensing mechanisms set up by (76) and (77). We will return to such cases in section 3.4 of the next chapter. (ii) and (iii) show German examples in which nur is used as a modal particle (speech act modifier) and as an adversative conjunction much like English but: (ii) Seid mir nur recht brav and anständig! be me NUR quite good and decent Try to be good and decent! (iii) Es ist ja gut daß du diese Theorie kritisierst, nur ich bin leider nicht it is good that you this theory criticize NUR I am unfortunately not VERANTWORTLICH dafür responsible for-it It is alright that you criticize this theory, but I am not really RESPONSIBLE for it The use of nur in (ii) is confined to imperatives, hypothetical conditionals and constituent questions. (see Abraham, 1991). In (iii) accent falls on a constituent that clearly does not associate with nur. This is never possible when nur is used as a focusing particle. It may be tempting to speculate about the common core of all these uses, but this would lead far afield into pragmatics and diachrony. For a comparison of these two uses, see Bayer (1991).
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1996 Springer Science+Business Media Dordrecht
About this chapter
Cite this chapter
Bayer, J. (1996). Movement to a Scope Position: Quantificational and Scalar Interpretation. In: Directionality and Logical Form. Studies in Natural Language and Linguistic Theory, vol 34. Springer, Dordrecht. https://doi.org/10.1007/978-94-017-1272-9_3
Download citation
DOI: https://doi.org/10.1007/978-94-017-1272-9_3
Publisher Name: Springer, Dordrecht
Print ISBN: 978-90-481-4623-9
Online ISBN: 978-94-017-1272-9
eBook Packages: Springer Book Archive