This level is called Logical Form LF. Ouhalla, Partial wh-movement is unusual in that it seems that the wh-phrase has to move twice. So what is the reason behind this partial movement construction? Is partial wh-movement essentially different from ordinary long-distance movement or wh-in-situ, or do all constructions yield the same interpretation?
- The Battle of Britain : The Greatest Air Battle of World War II!
- A-bar syntax : a study in movement types - Semantic Scholar!
- Poetics (Clarendon Greek Text and English Commentary).
- Exploring Developmental Psychology.
- X-bar theory - Wikipedia.
In my paper, I examine various syntactic constraints on partial wh-movement and the nature of the scope-marker. I also compare and contrast the two major analyses of partial wh-movement in the current literature. Instead, the scope marker forms an indirect relationship with the embedded clause. In section 3 I will outline the two current analyses, the DDA and the IDA, and subsequently I will show how both of them handle these syntactic constraints, respectively.
I will show that the IDA is by far the more elegant solution for partial wh-movement in terms of a linguistically unified approach. This is shown in 7 McDaniel , [25a-d] :. Examples 7a-c show that the wh-phrase can move to any SpecCP as long as the SpecCPs above it are filled by was the highest being the scope marker and the SpecCPs below it are filled by the complementiser dass.
Note that all of the above examples merely vary in their syntactic structure while carrying the same meaning, expressed by their English counterpart in 7e. Therefore it seems that the wh-phrase always has matrix scope, regardless of which position it occupies at SS. This observation may indicate that the wh-phrase forms a kind of chain with the scope marker situated in its actual target position.
The intermediate scope markers would then form links between these two positions. Now consider 7d : According to McDaniel, amongst others, 7d is ungrammatical because the complementiser interferes in the chain formation process between the scope marker and the wh-phrase. Research conducted in my own speech community 7 has also shown that 8 is in fact grammatical for a substantial number of speakers.
The class of verbs that license partial wh-movement in German is assumed to be more or less identical with the class of bridge verbs 8. Bridge verbs are verbs of saying, thinking and believing, e.
They lexically select a non-interrogative complement clause, i. This is shown in 9 :. Recalling the partial wh-movement construction 4 , redisplayed in 10 :. Examples 12 and 13 show the ungrammaticality caused by preference predicates and strong factive predicates, respectively.
Note that ordinary long-distance movement is possible with these verbs 10 :. The verb may also allow the sentential expletives es or das when combined with a constituent clause 17 , but not with an interrogative clause Following 18 , partial wh-movement is also ruled out in combination with es and das as seen in 19 :.
However, long distance movement including sentential expletives is possible:. Therefore they are mutually exclusive Based on this observation it is not surprising that verbs which generally do not combine with es and das also do not allow partial wh-movement:. These verbs which also include e. Therefore the direct object position is not available for either es or the scope marker was.
To summarise, despite the syntactic and semantic constraints just mentioned, the class of verbs that licences partial wh-movement in German roughly coincides with the class of bridge verbs. Reis argues that negative predicates cause ungrammaticality with partial wh-movement constructions but not with long-distance movement. According to my own native speaker judgement, however, I find that partial wh-movement with negative predicates is acceptable when compared to the same constructions involving a verb which genuinely licenses partial wh-movement, such as meinen.
The ban on these verbs may be due to independent semantic reasons, which will not be further discussed here. Negation of the matrix predicate in long distance movement is, however, grammatical:. Consider the following examples:. Summarising this section, it is clear that negation in the matrix clause of a partial wh-movement construction is not possible, but is possible in the matrix clause of a long-distance construction. According to Rizzi, partial wh-movement constructions should show an asymmetry with respect to all weak islands, e. We have seen in sections 2. On the contrary, SS-movement of the wh-phrase out of a factive or negative island does not cause a problem.
We have seen that long-distance constructions such as 28 and 30 are grammatical. This account is based on syntactic constraints on LF-movement in general and also holds for negation asymmetries in constructions other than partial movement. On semantic grounds, Dayal has analysed the Negation Asymmetry in partial wh-movement in terms of D iscourse -linking This means that certain background knowledge must always be presupposed in negative questions, i.
For 33a it is not necessary to know the set of individuals from which the possible values for who can be chosen. In 33b , however, this knowledge must be presupposed as otherwise, the context necessary for interpreting this question would be missing. Likewise the partial movement construction 28 , repeated here in 34 , is not D-linked and therefore ungrammatical:. Dayal captured these facts in the following observation Dayal , Property E :. This issue will be discussed in greater detail in section 4.
In the DDA, a direct relationship is formed between the scope and the wh-phrase in the embedded clause. This kind of analysis was first proposed by van Riemsdijk and then further developed by McDaniel The scope marker is the highest basic wh-phrase in a partial wh-movement construction, e. It has the nature of an expletive which is replaced by the real wh-phrase at LF, i.
The idea of a scope marker in partial wh-movement constructions in German was first developed by van Riemsdijk He compared the following examples 42a-e  in van Riemsdijk :. This means that an answer to a certain question must refer to all and only those wh-phrases that have matrix scope, i. A possible answer to all 42a-e is shown in 43 :.
- A-bar Syntax;
- Semantic Web Technologies for e-Learning, The Future of Learning, Volume 4.
- More Contemporary Cinematographers on Their Art.
Therefore, as von Stechow points out, all of 42 must have the same LF derivation von Stechow, , [5a] 13 :. This means that all of the syntactic alternatives displayed in 42 are merely structural variants at SS. It remains to be explained, however, what the actual motivation for allowing the various syntactic structures at SS could be. This observation poses some difficulty for the DDA and will be addressed again in section 4.
McDaniel claims that in German, partial wh-movement as well as full wh-movement, are restricted in the same way by a Subjacency-like constraint. Her analysis is based on the constraints of Quantifier Raising May, , whereby the wh-phrases must move through SpecCPs in successive-cyclical steps in order to avoid subjacency.
Since the scope marker in the highest SpecCP as well as the intermediate links are base-generated in their respective SpecCPs, McDaniel suggests that Subjacency must be a condition on representation rather than on movement. Thus the scope marker is a special type of wh-operator which nevertheless enters into a standard relationship with its variables, thereby following the usual conditions on chain formation processes. Fanselow suggests that partial wh-movement is quite similar to long distance movement in the sense that both rely on creating A-bar-chains. The two constructions can be paraphrased by the following chains, respectively:.
The two chains essentially share the same function, hence derive the same interpretation. While in 45a the wh-operator binds to variable traces, in 45b the second link is created by insertion of the expletive scope marker. In this chain, the semantic relationship between its components is established by coindexing. Starting from Hindi, Dayal has developed a different analysis in her approach to partial wh-movement. In fact, she prefers calling the structure a scope marking structure as, according to her, there is no partial movement involved as such.
Although there are two different lines of analysis within the IDA, its attempt to unify the phenomenon of partial wh-movement across all languages where it occurs is very appealing. As will be discussed in section 4, the DDA runs into severe problems in trying to account for the facts of partial movement in e. It would be surprising if, as Mahajan points out, phenomena that are so closely related should require different analyses.
In this section, I will first of all outline the general idea of how partial wh-movement is accounted for under the IDA.
Word Order Permutations in Greek
According to the the IDA, kyaa in Hindi or was in German are not scope markers although, following Dayal, I will continue referring to them in this way , but ordinary wh-phrases in their own right. The starting-point for this proposal lies in the fact that was in both 47 and 48 is a wh-word for clausal complements, i. As shown in 48 , the two wh-phrases do not form a direct relationship with each other. Instead, both form local dependencies within their own clauses as displayed by their traces.
The embedded CP functions as a clausal complement to the scope marker was. They are coindexed and form an indirect relationship or dependency with each other. At this point in the analysis, there are two different solutions as to how this indirect dependency should be established. Therefore, a parallel can be drawn between the partial wh-movement construction in 49 and a construction containing a sentential expletive with a clausal associate, shown in 50 :.
As shown in 49 , was originates as the direct object of the matrix verb. Since the scope marker is an expletive , Mahajan argues that it must be replaced by its clausal associate at LF, i. In this way, the true wh-phrase can take scope over the sentence.
They do, however, have the same semantic values and thus yield the same interpretation, i. In the analysis argued for by Dayal , the scope marker is a standard wh-phrase that is base-generated in argument position of the matrix verb This is shown in 51b :. As was is an argument and not an expletive, it cannot be replaced by the embedded CP at LF. Instead, the CP restricts the domain that was can take scope over. Although grammatical judgments of adverb orders are often delicate and notoriously difficult when more than two or three adverbs are involved, Cinque found a way of testing the order of adverbs, which he then applied systematically to adverbs in a number of languages.
His method consisted of comparing the order of two adverbs in a sentence, say A1 and A2. In most cases, one order is judged more acceptable that another. Now, the introduction of a third adverb, A3, leads one to expect three potential orders consistent with the order previously established, and not the combinatorialy possible six orders. One should now test the order of A3 and A1. A systematic comparison of pairs of adverbs across different primarily Romance languages, leads Cinque to a cline of over 30 adverbs.
Adjunctions to maximal projections typically do not show such rigid ordering constraints and even if a mechanism were devised to impose such an ordering, the second property of adverbial syntax demonstrates that such a device would fail, in principle. This second property is that a verb can appear between any two adverbs. The following set of data, adapted from Cinque op. Cinque concludes that the most economical structural expression of these facts is to take adverbs to be specifiers of projections, the heads of which can host a raised verb.
This is shown schematically in 9 , where X stands for a head position. The third property leads to a characterization of the heads of the functional categories of which the adverbs are specifiers of. Cinque shows that that the order of adverbs can be matched with the crosslinguistically stable order of heads such as inflectional affixes, functional particles, preverbs of different sorts and auxiliaries.
YP in 11 , the specifier of Top or Foc, contains the topicalized or focalized constituent. The heads of these projections, Topic 0 and Focus 0 are overtly realized in some languages. Subsequent work has refined this proposal by proposing a privileged position for fronted adverbs Rizzi a,b and distinguishing different kinds of hierarchically ordered topic phrases e. This is perhaps one of the reasons for the impact that this project is having on the field and its extension to the nominal domain see e. The accumulated results of these and other cartographic studies make a strong case for the universality of the functional hierarchy.
Two interrelated questions arise at this point: first, where do the contents of this functional structure come from? Second, what explains the particular order or hierarchy in which they appear? Both questions are rather difficult to answer because they relate not only to properties of the computational system itself but to its interface with other cognitive faculties.
It stands to reason that the choice of functional items is not an arbitrary, irreducible property of UG although familiar arguments from the poverty of stimulus militate in favor of the hypothesis that the functional features and their hierarchical arrangement are wired into the grammar and not learned or otherwise acquired through experience. Minimalism introduced a distinction between interpretable and uninterpretable features.
The latter drive computations: Movement or internal merge is triggered by the need to check and delete these features Chomsky or value them Chomsky Interpretable features, on the other hand, are those that can be understood and exploited by the modules which interface with the computational system, semantics, pragmatics and indirectly, the thought systems. The question of what is coded or articulated by grammatical features can be rephrased as following: Of the properties which enter into human thought and belief systems, which ones are represented as grammatical features?
If UG respects a principle of decompositionality, as he suggests Kayne b , whereby every interpretable syntactic feature projects a unique head, than may turn out to be a low estimate of the magnitude of syntactic features. The study of the feature inventory of UG requires a massive database compiled on the basis of detailed studies of particular grammars.
Movement - Linguistics - Oxford Bibliographies
As Kayne b has frequently argued, this is about the only way to uncover functional material which remains unpronounced, and hence difficult to sequence when one studies only a small set of languages. Whatever the actual number of features turns out to be, it will surely be as subset of the properties which enter into cognition, thought or communication. Just as there are certain participant roles which appear to lie outside the realm of thematic role coding viz. A close look at the structural maps which cartographic studies have drawn reveals some systematicity in the composition of the functional domain and the hierarchy of projections.
Ramchand In the Cinquean hierarchy in 10 , the two tense heads Past and Future are arrayed above all the aspectual heads. While the range of options is quite limited, the combinatorial possibilities which arise when they are evaluated for each feature engender an astronomical number of possible settings and hence of possible grammars. The switchboard metaphor of parameters endorsed by Chomsky in the Government and Binding framework see Boeckx for recent discussion may be retained, but construed somewhat differently: Every feature is endowed with its own switchboard, consisting of half a dozen or so binary options.
To the degree that this hierarchy feeds semantic computation, one expects a tight correspondence between the ordering of the rules of semantic composition and the cartographic sequence. This is largely true. However, the fairly transparent correspondence between the cartography of clause structure and its semantics does not lead to the conclusion that the hierarchy can simply be reduced to semantic considerations which is not to deny that some aspects of it may, viz. Cinque and Rizzi 52— Hinzen Cartographic work is not very explicit as to how the hierarchy of heads is formally guaranteed.
If it is derived by precise statements of selection e. Asp27 selects Asp26 , then it follows that the entire cartographic hierarchy must be merged in every full clause. This view receives some prima facie support from empirical studies of reduced or truncated structures. Rizzi argues that the mechanisms of truncation cannot pick and choose which projections to effect; the entire structure down from the point of truncation must be retained. Thus, the entire structure is and must be represented syntactically, but only certain features are activated and feed interface interpretation.
In part, this is a division of labor: Minimalism focuses on mechanisms of computation Merge and Search and the role of uninterpretable features, while the cartographic enterprise is primarily concerned with the inventory of interpretable features. Thus, phases CP and vP and their edges i.
VOS in Greek
It is far from clear how to integrate these notions into the structural maps of cartography, in which the clause is typically seen as a homogenous hierarchy of projections. Each should, in principle, have a specifier but which one corresponds to T? Which one of these lower heads should be taken to constitute the edge of vP? Perhaps an even more nagging problem is that of selection. Minimalism inherits from previous approaches the view that selection is carried out under sisterhood. Thus, C selects T and V selects C. How is selection satisfied, for example an indirect question, if the head bearing the interrogative feature is Foc or Int cf.
Rizzi and thus not a sister to V? Or take the familiar problem of how subjunctive features on an inflectional head can be selected by a higher predicate, given the number of intervening heads between V and the relevant mood head? Which C? Which T? The desirable goal of integrating the research agendas of Minimalism and Cartography requires, so it seems, modifications in the way structure, in the cartographic sense, is manipulated by the computational system. Why, one may ask, do these features typically fail to extend to the heads of left periphery?
Cartographic works have, for the most part, implicitly assumed delimited structures or spaces but have not provided a formal implementation of domains. The emergence of Cartography as a research topic is a natural development in a theory of grammar concerned with how interpretable features are structurally represented.
The question was and remains which is the right one. Cartography is not an alternative to Minimalism. His research interests are primarily in theoretical and comparative syntax. I wish to thank D. Basilico for suggesting this project, G. Cinque and two anonymous referees for helpful comments. Because of its limited scope, this review article does not do justice to the wealth of material which has been published in the realm of syntactic cartography. The list of references is very selective and I extend my apologies to the many researchers whose work is not cited.