Remarks on the Economy of Pronunciation *
|
|
|
- Bethany Holt
- 9 years ago
- Views:
Transcription
1 Remarks on the Economy of Pronunciation * Gisbert Fanselow, University of Potsdam Damir Ćavar, Technical University of Berlin 0. Introduction and Overview Although it is technically independent of it, the idea that syntactic movement is composed of two steps, a copying operation followed by a deletion operation (the C&D-theory of movement) as illustrated in (1)- became again popular with the rise of the Minimalist Program (Chomsky 1993). In one of the straightforward extensions of the C&D-approach, at least certain instances of so-called covert movement arise from the overt copying of a full phrase before SPELLOUT, followed by the deletion of the higher rather than the lower copy - an assumption that implies that spellout conventions regulate whether the target or the source position of the copying operation is realized phonetically (see, e.g., Bobaljik 1995, Groat & O'Neill 1996, Pesetsky 1997, 1998a, Roberts 1997 (for head movement), Sabel 1998, among others) as illustrated in (2) for Chinese. (1) Overt Movement (it does not matter) she likes who COPYING (it does not matter) who she likes who DELETION OF SOURCE (it does not matter) who she likes who (2) Covert Movement ta weishenme da meigeren COPYING he why hit everyone weishenme [ta weishenme da meigeren] DELETION OF TARGET weishenme [ta weishenme da meigeren] In this paper, we will discuss four constructions which we believe have a fairly simple analysis in a C&D-theory of movement only: true partial wh-movement as in Bahasa Indonesia (3), wh-copying (4), left-branch extractions/split constituents as in German (5), and (apparent) head movement. We agree with Pesetsky (1997, 1998a) in the conviction that the particular success of a C&D theory of movement (as compared to other models) hinges on its interaction with principles of sentence pronunciation in an optimality theoretic fashion 1. (3) a. Bill tahu Tom men-cintai siapa Bill knows Tom loves who b. Bill tahu [siapa yang Tom cintai] Bill knows who FOC Tom loves c. Siapa yang Bill tahu Tom cintai "who does Bill know that Tom loves?" (4) wen denkst du wen sie liebt who think you who she loves "who do you think that she loves" (5) was hat sie [t für Bücher] gelesen what has she for books read "what kind of books has she read?"
2 Gisbert Fanselow & Damir Ćavar Furthermore, the analysis of the constructions discussed here relies on the notion of cyclic optimization in syntax, as it was recently proposed by Müller (1999) and Heck & Müller (1999). The paper is organized as follows. We will introduce the basic idea of pronunciation economy approaches in the next section, with an analysis of partial wh-movement. It will be argued that partial wh-movement data are exactly the kind of construction one would expect to find if the C&D-theory of movement is correct. Certain syntactic differences among the languages that have partial movement will be analyzed in an optimality theoretic fashion. The economy aspect of the approach defended here will become clear in section 2, where we show that deletion in chains may be incomplete if certain constraints are ranked above the principle of pronunciation economy. In section 3, we show that semantic and phonological conditions may imply that deletion is distributed over both copies in a movement chain. One particularly promising aspect of this account is that it allows us to reduce head movement to phrasal movement without being confronted with notorious problems like missing freezing effects that arise in other approaches (e.g. Kayne 1998, Koopman & Sportiche 1999, Mahajan 1999). 1. True Partial Wh-Movement An integration of the C&D-theory into an OT-approach to syntax (see Grimshaw 1997, Pesetsky 1998a, Legendre (in press), among many others) involves the following ingredients: copying and deletion apply freely in the GEN-component of grammar, but the effects they have on grammatical outputs are determined by a number of principles on syntactic structure and sentence pronunciation. For concreteness, we will assume without further discussion that the creation of copies ("movement") is forced by the need to check features of an attracting head (as in Chomsky 1995, but nothing hinges on that assumption) 2, and we assume that movement is successive-cyclic. Chains that are created in this way will "originally" contain multiple copies of the same phonetic and semantic material. These copies may, but need not be subjected to a deletion operation, so that GEN will generate at least the candidates in (6) 3 for a question like who do you think she will invite? (6) a. who do you think who she will invite who b. who do you think who she will invite who c. who do you think who she will invite who d. who do you think who she will invite who e. who do you think who she will invite who f. who do you think who she will invite who g. who do you think who she will invite who h. who do you think who she will invite who In the standard case of movement, only one of these copies is actually pronounced. This follows from an interaction of the principles PRONECON and RECOV 4. PRONECON favors those structures in which the deletion of phonetic matrices in chains is maximized, but deletion is subject to recoverability, so that normally 5 exactly one copy will be retained in each chain. In other words, in most situations, only (6e-g) are potential winners. (7a) Pronunciation Economy (PRONECON) 6 *Phonetic Matrix (7b) Recoverability (RECOV) 7 The content of unpronounced elements must be recoverable from a local antecedent. 2
3 Economy of Pronunciation Ceteris paribus, this approach leads to the expectation that any copy in a chain may be the one that is spelled out, with all the others being deleted. So-called 'partial wh-movement' 8 as it can be found in Bahasa Indonesia (cf. (5), repeated here as (8), and Saddy 1991, 1992) or Malay (Cole & Hermon 1998) seems to bear this prediction out. In a wh-question, the wh-phrase may either appear in situ (8a), or be realized in its scope position (8c), but it can also show up in the specifier positions of any of the CPs that may intervene between the root position of the wh-phrase and its scope position, as (8b) illustrates ('true partial wh-movement'). (8) a. Bill tahu Tom men-cintai siapa Bill knows Tom loves who b. Bill tahu [siapa yang Tom cintai] Bill knows who FOC Tom loves c. Siapa yang Bill tahu Tom cintai "who does Bill know that Tom loves?" Cole & Hermon (1998) argue that partial wh-movement is not focus movement, see also Basilico (1998) for arguments that partial movement in Slave cannot be reduced to focus movement. The most straightforward analysis for (8) (considered but rejected in Cole & Hermon 1998) assumes that siapa has in fact been attracted to its scope position in all examples, forming the chain indicated in (9a). Due to the interaction of PRONECON and RECOV, all but one of the copies of siapa must not be pronounced. In the optimal state of affairs, any of the copies may be the one that is realized overtly, as the abstract structures (9b-d) illustrate, that (roughly) correspond to (8). (9) a. [ CP siapa. [ CP siapa siapa]] b. [ CP siapa. [ CP siapa siapa]] c. [ CP siapa. [ CP siapa siapa]] d. [ CP siapa. [ CP siapa siapa]] There is at least one argument for analyzing true partial wh-movement along these lines. As Saddy (1991, 1992) and Cole & Hermon (1998) observe, partially moved wh-phrases behave as if they are moved to the scope position at least in terms of island conditions: there must be no movement island between the partially moved wh-phrase and its scope position. Thus, a wh-phrase cannot be moved out of an adjunct clause in Malay, and partially moved wh-phrases must not occur within adjuncts either, as (10) (taken from Cole & Hermon 1998: 227, 236) illustrates. The same holds, e.g., for subject islands or for wh-islands: wh-phrases cannot be extracted out of such islands, and wh-phrases that seem to have undergone 'partial' wh-movement are not tolerated in these constructions either. (10) a. *apa (yang) Ali dipecat kerana dia beli t what (that) Ali was-fired because he bought b. *Ali dipecat apa (yang) kerana dia beli t "what is the thing such that Ali was fired because he bought it?" Such observations are explained straightforwardly if -as (9) suggests- 'partial' movement is in fact full wh-movement, involving a 'non-standard' deletion part though: the constellation in (11a) created by copying is ruled out by standard island theories. It does not matter then whether the uppermost or an intermediate copy of this (illicit) wh-chain fails to undergo deletion triggered by PRONECON. (11) a. *wh-phrase.. [ Island.. wh-phrase. wh-phrase..] b. *wh-phrase.. [ Island.. wh-phrase. wh-phrase..] c. *wh-phrase.. [ Island.. wh-phrase. wh-phrase..] Island facts thus favor the C&D-analysis of partial wh-movement. 3
4 Gisbert Fanselow & Damir Ćavar Saddy (1991) and Cole & Hermon (1998) point out, however, that argumental wh-phrases in situ can appear within syntactic islands (although they cannot be moved out of these islands), as (12) illustrates. (12) Ali dipecat kerana Fatimah fikir dia membeli apa Ali was-fired because Fatima thinks he bought what (12) illustrates, in fact, a fairly widespread property: unlike their adjunct counterparts, argumental whphrases in situ do not obey any island constraints in a number of languages (but not in all), among them Chinese (see, e.g., Huang 1981). If island conditions affect overt and covert movement in the same way - as they have to if the difference between the two movement types is one of pronunciation only- (12) cannot involve movement: rather, argumental wh-phrases in situ must be assumed to be bound by a (null) question operator in the appropriate Comp (see e.g. Aoun & Li 1993, Tsai 1994, Cole & Hermon 1998) 9. More precisely, the [+wh]-comp as in (13) may or may not have a syntactic feature F that attracts a wh-phrase 10. If the feature F is missing, the wh-phrase cannot move to the specifier position of Comp, but a binding relation can be established for wh-phrases that are 'referential' enough for being bound (i.e., for argumental wh-phrases). Then no island effects are to be expected 11. If the attracting feature F is present, the wh-phrase moves, so that island effects arise, irrespective of where the wh-phrase is spelled out in the end. (13) Comp [+wh].. wh-phrase Chinese illustrates the prediction that certain in situ wh-phrases can be island sensitive: wh-adjuncts can stay in situ, but they must not appear in islands. Note that adjuncts cannot be bound by an (argumental) question operator base generated in Comp. Therefore, wh-adjuncts can form a part of a question only if a chain is built up which links the adjunct to its scope position. Wh-adjuncts can thus be realized phonetically in situ only if a copy-chain (respecting islands) is built up to the scope position, in which the lowest copy surfaces after the deletions as forced by (7) 12. We therefore follow Cole & Hermon (1998) in making the assumption that two strategies for forming questions coexist in Malay at least: copying of the wh-phrase to its scope position, and the binding of wh-arguments in situ. See Pesetsky (1998b) for a related but slightly different view on English, German and Slavic questions. A phonetic sequence such as (14) in which an overt copy of the wh-phrase appears in its base position is thus ambiguous in our account (but not in Cole & Hermon 1998): buah apa may be bound by a [+wh]-comp, or it may be the copy of a chain link to the matrix Spec,C position that is spelled out phonetically. Given that the binding-in-situ strategy is, in general, more liberal than the formation of questions by movement, (nearly) all examples that are grammatical under a movement analysis are generatable with a binding analysis, too - so that the existence of an ambiguity is both hard to establish and also hard to refute. (14) Mary (mem)-beli buah apa di kedai Mary prefix-buy fruit what at shop "what fruit did Mary buy at the shop? This prediction of an ambiguity may come closer to what holds in Singaporean Malay than in Bahasa Indonesia. According to Saddy (1991), wh-phrases in situ in Bahasa Indonesia are special in that they always take widest possible scope with respect to other operators, and one can take this as an argument against the systematic structural ambiguity of wh-phrases in situ that we predict, but the relevant 4
5 Economy of Pronunciation judgments seem not to be shared by native speakers of Singaporean Malay (Cole & Hermon 1998:225), so that we may uphold our analysis at least for the latter language. As for the situation in Bahasa Indonesia, systematic differences between wh-phrases that are phonetically realized in situ and those that appear in other positions can be accounted for in the following way. Müller (1997) argues for a principle like (15) as one of the determinants of parametric variation among languages with respect to question formation (we have adapted the formulation to the needs of the system we develop here) (15) WH-IN-SPEC A wh-phrase must be phonetically realized in the specifier position of a CP WH-IN-SPEC blocks the phonetic realization of wh-phrases in situ whenever a chain has been formed which contains more than two members (and if at least one of the chain members is a specifier of CP). Thus, unless other principles override it, (15) implies that wh-phrases in situ are not part of a chain reaching Spec,CP, as required for Bahasa Indonesia. As long as RECOV is ranked above WH-IN-SPEC, however, wh-phrases that are bound by a null operator in Comp (rather than being part of a chain created by copying) can nevertheless be realized in situ, because their omission would violate recoverability. For languages like Chinese in which wh-phrases show up in root positions only, the effects of (15) must be counteracted by a further principle like STAY* that has been proposed in a somewhat different form in various works (see e.g. Grimshaw 1997, Müller 1997, Legendre et al. 1998, Ackema & Neeleman 1998) and that can be formulated as in (16) in our approach (16) STAY* (Nonstandard formulation) 13 If the phonetic matrix of α c-commands a member of the chain of β, then it c-commands the phonetic matrix of β. When STAY* >> WH-IN-SPEC, the Chinese type of question formation arises (no visible effects of copying whatsoever), when WH-IN-SPEC >> STAY*, the Bahasa Indonesia system comes into being (wh-phrases in chains are always displaced phonetically 14, but bound wh-phrases may be realized in situ). Finally, if there is a tie between STAY* and WH-IN-SPEC, the grammar of Singaporean Malay arises, in which wh-phrases that belong to movement chains may surface in situ and in derived positions. A further observation on Malay and Bahasa Indonesia discussed in Saddy (1991) and Cole & Hermon (1998) seems to be incompatible with our analysis and figures as the key argument against the spellout account of partial movement in Cole & Hermon (1998:251) 15. Transitive verbs in Bahasa Indonesia and Malay optionally combine with certain prefixes like meng. These must be absent, however, when a wh-phrase has moved across them, but they can be present when the wh-phrase is in situ. This rule holds for long wh-movement as well (see Cole & Hermon 1998: for details). Crucially, in partial wh-movement constructions, meng must be absent between the root position and the position of the phonetically realized wh-phrase only, it can appear between the overt position of the wh-phrase and the latter's scope position, as (17) indicates. (17) Ali (mem) beritahu kamu tadi apa yang Fatimah (*men)-baca Ali MENG told just now what that Fatimah MENG read "what did Ali tell you just now that Fatimah was reading" Whether this observation creates a problem for the C&D-analysis of 'partial' movement (as Cole & Hermon (1998) claim it does) or not, depends, of course, on the details of the rule system that governs 5
6 Gisbert Fanselow & Damir Ćavar the distribution of meng. (18) appears to capture the empirical facts, and if such statements are allowed as (language-particular?) constraints in grammars, particle distribution cannot even be used as an argument against a fully representational interpretation of a C&D analysis of partial movement. (18) *MENG, whenever MENG is c-commanded by the phonetic matrix of wh-phrase α and when it c-commands a trace of α. A more convincing analysis of the particle facts can be constructed, however, if we assume a notion of cyclic optimization, as proposed by Müller (1999) and Heck & Müller (1999). In standard OT, the generator component of grammar (GEN) constructs a set of syntactic objects from an input. These syntactic objects are the candidates for the evaluation procedure (EVAL) that selects optimal candidates, which are then grammatical sentences. In a cyclic model of optimization, this interaction of GEN and EVAL applies sequentially, building up and optimizing larger and larger syntactic objects. For concreteness, suppose that by applying merger and copying operation to lexical entries or syntactic objects previously formed, GEN generates a set of syntactic objects, until the elements so formed correspond to cyclic nodes (NPs and CPs) or to 'phases' in the sense of Chomsky (1998). These candidates are then subjected to the EVAL procedure, yielding an optimal candidate. The optimal candidates for the expression of cyclic categories/phases so formed may then be fed into the GEN component again, in order to generate even larger structures, until the level of cyclic nodes or phases is reached again, at which the EVAL procedure selects the optimal structure again. In such a system, the question of which of the copies created by movement can be retained, and which copies are deleted phonetically, poses itself each time the construction of alternative structural representations has reached the cyclic node level. Consider, then, a stage in a derivation in which a wh-phrase has been copied to a higher position, crossing an occurrence of meng in this context (=19a). Suppose that Σ is cyclic, so that optimization can and must start. Because of PRONECON, one of the two wh-phrase copies must disappear 16. If the upper copy loses its phonetic matrix (19b), nothing seems to have to happen to meng, i.e., it can be retained. In structures in which meng has been retained, the uppermost specifier position of CP therefore does not have a phonetic matrix, and it will not be able to re-acquire this phonetic matrix in later copying steps for more or less obvious reasons 17. Therefore, above a retained meng, no copy in a wh-chain originating lower than meng can have a phonetic matrix. Assume, however, that there is a principle that requiring that meng must be deleted (=19c) when the upper copy is retained phonologically. This can (and must) be checked locally in each cyclic domain relevant for optimization. Thus, the empirical generalizations that concern meng-distribution are very well compatible with a C&D-approach when it is executed cyclically 18. (19) a. [ Σ wh-phrase meng wh-phrase] b. wh-phrase meng wh-phrase c. wh-phrase meng wh-phrase A more straightforward version of this account assumes that cyclic wh-movement always targets the specifier positions of the relevant "phases", i.e., it assumes that cyclic wh-movement always passes through Spec,CP and the specifier position of a functional projection below the subject position but above VP (see Chomsky 1986, 1998; the relevant functional head might e.g. be AGR-O relative to some earlier versions of the minimalist program, or the "outer specifier of vp" as in Chomsky 1998). The ban on the use of the meng-prefix in situations in which is has been "passed" by overt movement can then be reduced to (20), which bears an obvious similarity to the doubly-filled Comp filter. 6
7 Economy of Pronunciation (20) *[ AGR-O-P WH-PHRASE [ AGR-O meng [ vp.]]], if WH-PHRASE has a phonetic matrix. When the wh-phrase has been copied to Spec,AGR-O, a phase is completed and the output must be optimized. If meng deletes, the wh-phrase in Spec,AGR-O may or may not be the one that retains its phonetic matrix. If meng fails to delete, the lower copy of the wh-phrase and not the one in the specifier position of AGR-O- must be the one that retains its phonetic matrix. The upper copy (being stripped of its phonetic features) is, however, the one that will undergo further (and therefore invisible) movement (see footnote 17 for a precise argumentation). Thus, the fact that meng is never "crossed" by overt movement is derived. That the wh-phrase is, apparently, never realized phonetically in Spec,AGR-O seems to follow from the interaction of STAY* and WH-IN-SPEC. STAY* only favors the root position of a wh-phrase, while WH-IN-SPEC disfavors the realization of wh-phrases in anything but Spec,CP. We now seem to run into a problem, however: if optimization applies cyclically, WH-IN-SPEC forces the phonetic realization of a wh-phrase in the lower Spec,CP-1 position when Σ is optimized in (21). Spec,AGR-O is then, left, with a wh-phrase lacking a phonetic matrix and this is the only one that can be copied to higher positions. Spec,CP-2 thus seems to never be able to acquire a phonetic matrix. (21) [ Σ Spec,CP-2... [ Σ Spec,AGR,O- Spec,CP-1 ]] Note, however, that we need a further principle anyhow in order to capture languages that neither follow the wh-strategy (Chinese) nor partial movement (Malay). Consider in this respect PARSESCOPE borrowed from Legendre et al. (1998), but adapted to our current needs: (22) PARSESCOPE (Nonstandard formulation) If α has scope over β, then the phonetic matrix of α c-commands the phonetic matrix of β Suppose PARSESCOPE is tied with WH-IN-SPEC, while STAY* is low (or tied). If α is a wh-phrase, the optimization of Σ in (23) will imply that α's phonetic matrix appears in Spec,AGR-O-1 in (23) only STAY* runs counter this conclusion, but STAY* has a lower rank. When Σ* is optimized, the phonetic matrix of α is moved even further, because PARSESCOPE and WH-IN-SPEC pull in the same direction. (23) [ Σ*** Spec,CP-2 [ Σ Spec,AGR,O-2 [ Σ Spec,CP-1 [ Σ Spec,AGR,O-1 α ]]]] The wh-phrase appears now in the lowest Spec,CP position. The derivation bifurcates when Σ** is reached: α climbs up phonetically if PARSESCOPE is given more weight, while its phonetic material stays in Spec,CP-1 when the tie is resolved towards WH-IN-SPEC. The former derivation will finally copy the phonetic material of α further to Spec-CP-2 (because the two constraints in question have the same implications for the last derivational step), the latter cannot but leave α phonetically at Spec,CP- 1. In other words, where there is a tie between WH-IN-SPEC and PARSESCOPE, partial wh-movement arises. When PARSESCOPE dominates WH-IN-SPEC, the phonetic material of the wh-phrase will be realized in the highest chain position under consideration (WH-IN-SPEC cannot block scope driven movement up to Spec,AGR-O) this characterizes languages with full and multiple wh-movement like Romanian and Bulgarian. The factorial typology leads us to expect that there are also languages in which WH-IN-SPEC dominates PARSESCOPE. Slave could be such a language: In Slave (Basilico 1997), the wh-in-situ strategy is less restrictive than overt movement, as (24a-b) show: the complements of so-called "indirect 7
8 Gisbert Fanselow & Damir Ćavar discourse verbs" are barriers for overt movement, but wh-in-situ is licensed. It is thus surprising that (24c) involving partial movement within the island is in fact grammatical, quite in contrast to what one would expect from the situation we found in Malay. (24) a. *?ode netá nimbáa enáih?á kenéhdzáh where your father tent 3pitch 3tried "where did your father try to pitch the tent?". b. Raymond Jane judeni ri yili kodisho Raymond Jane where Foc 3is 3knows c. Raymond [judeni ri Jane yili] kodisho "where does Raymond know Jane to be (=1875a,b of Rice 1989) Slave differs from Malay in a further respect: there are configurations in which partial movement is ruled out, while complete movement is not: complements of direct discourse verbs are transparent for movement, but disallow partial movement: (25) a. John beya judeni ráwozée sudeli John my son where 3opt.hunt 3wants1 "where does John want my son to hunt?" b. *John judeni beya ráwozée sudeli c. hodi nurse egháuhndá néndi where nurse 1opt-see-2sg 3told2 "where did the nurse tell you she would see you" Basilico (1997) suggests that complements of direct discourse verbs lack a Spec,CP node, so that (25b) cannot possibly arise. Thus, the grammaticality of (24c) is the only surprising property of Slave. We can understand the contrast between (24a) and (24c), however, if we assume that complements of indirect discourse verbs are transparent for movement, but that some constellation C of principles rules it out that any of the occurrences of wh-phrases that have been copied out of the complement CP could ever bear a phonetic matrix: wh-movement then has to be obligatorily "partial". We have already seen what this constellation of principles is: WH-IN-SPEC dominating PARSESCOPE inevitably prevents wh-phrases from leaving a Spec,CP position they have reached if wh-movement needs to additionally pass through an AGR-O-position in the next higher clause. The predictions of the factorial typology constructible from STAY*, PARSESCOPE and WH-IN-SPEC are therefore borne out. 2. Wh-Copying 2.0 Preliminary remarks on question formation in German. Before we discuss wh-copying in German, we start with a few remarks on the embedding of German in the principle system developed so far. German is not a wh-in-situ language, and does not allow partial wh-movement of the kind we find in Malay languages, at least not in simple questions. In multiple questions, only one wh-phrase appears in its scope position: (26) wen hat er wem gezeigt who acc has he who dat showed "who did he show to whom? *wen wem hat er gezeigt On obvious grounds, (26) can be analyzed in two ways: Taking up ideas proposed by Grewendorf (1999) and Sabel (1998), we may hypothesize that all wh-phrase move to their scope position, but that there is a principle that bans the spelling out of more than one wh-phrase per Spec,CP position. Alternatively, we may follow Müller (1997) in the assumption that there is a principle that bans the 8
9 Economy of Pronunciation (phrasal) movement of more than one wh-phrase to Spec,CP in German. The wh-phrase in situ would then have to be bound in situ (or undergo feature movement in the system of Pesetsky 1998b). When the two wh-phrases of a multiple question originate in different clauses, no uniform pattern emerges: in addition to the constellation in (27a), which closely mirrors the English counterpart and which characterizes standard German, there are dialects in which the multiple question cannot be formed in the way it is in (27a) 19. In such dialects, the lower wh-phrase either has to undergo 'partial' wh-movement to the specifier position of the complement clause (as in (27b), which is acceptable for at least some speakers in Potsdam and surroundings), or the lower wh-phrase must be the one that undergoes overt movement (as in (27c), blatantly violating superiority thereby). (27) 20 a.!wer denkt, dass sie wen liebt who thinks that she who loves "who thinks that she loves who" b.!wer denkt, wen sie liebt c.!wen denkt wer, dass sie liebt The latter two dialects thus resemble Iraqi Arabic (see Ouhalla 1996) and Hindi (Mahajan 1990) in the sense that (a) wh-phrases in situ cannot take scope out of the minimal finite clause they are contained in (unless they fill this clause's specifier position) and (b) that the distribution of wh-in situ is therefore more constrained than the distribution of wh-phrases that have undergone overt movement. If German wh-phrases in situ are not moved covertly, and are subject to additional binding requirements of the sort we find in Iraqi Arabic (Ouhalla 1996), the unavailability of a multiple question interpretation for (27a) in the relevant dialects is captured fairly easily, while it is less clear why covert movement should have to fulfill less liberal island conditions than overt movement, if the major difference between the two operations is one of the location of spellout. The dialects that rule out (27a) thus suggest that German wh-phrases in-situ do not involve covert movement. This is quite in line with the conclusions arrived at (for what he terms covert phrasal movement) by Pesetsky (1998b) on quite different grounds. Obviously, the dialects with the stricter binding restrictions on wh-in situ solve the pertinent problem in two ways, either by allowing it that partial movement moves the wh-phrase to a position in which it can be bound from outside (27b) or by moving the lower phrase directly to its scope position (as in (27c)). The latter strategy cannot help with triple questions for straightforward reasons: the lowest whphrase still is separated from its scope position by a finite clause boundary. (28) *wen denkt wer, dass sie wem vorgestellt hat who acc thinks who that she who dat introduced has "who thinks that she presented who to whom" On the other hand, (29) is fine in the dialects that tolerate partial movement, so we must assume that wem fulfills the locality requirements on binding because of the presence of wen in the next Spec,CP position. (29)!wer glaubt wen sie wem vorgestellt hat who believes who she who introduced has "who thinks that she presented who to whom" Let us now turn to the different strategies of forming long-distance questions in German: Like Romani or Frisian, Germanis fairly rich in this respect. 9
10 Gisbert Fanselow & Damir Ćavar (30) a.!wen denkst du dass sie liebt who think you that she loves b. wen denkst du liebt sie c. was denkst du wen sie liebt what think you who she loves d.!wen denkst du wen sie liebt who think you who she loves e.!/*wen denkst du was sie liebt "who do you think that she loves?" (30a) exemplifies standard long wh-movement of arguments which is grammatical in some, 21 but not all varieties of German (see e.g. Kvam 1983, Fanselow, Kliegl & Schlesewsky 2000). Extractions from so-called verb-second-complements (30b) are well-formed in all dialects of German 22. (30c) exemplifies so-called "wh-scope marking", and is often analyzed as involving partial wh-movement plus the insertion of a scope marker (see McDaniel 1989, Müller 1997, and the contributions to Lutz et al. (2000)). If Fanselow & Mahajan (1996, 2000) are correct, however, the constructions involve quite a different analysis. Presupposing that the latter approach is correct, we will ignore the construction in the rest of this paper. (30d) is, however, a construction of particular interest in the context of pronunciation economy: it appears as if more than one copy within a wh-chain formed by overt movement is spelled out phonetically (Copy-Construction, CC). Similar constructions exist in Frisian (Hiemstra 1986), Afrikaans (du Plessis 1987) and Romani (McDaniel 1989). We will focus our attention on this construction in this section, after brief comments on remaining options. McDaniel, Chiu & Maxfield (1995:741) state structures like (30e) are ungrammatical in German although they are used e.g. in Romani with the wh-marker so. Comparable constructions involving what exist in Child English, but it is indeed hard to find native speakers of German who accept (30e). Note that the "wh-scope marker" so appearing in the Romani counterparts of (30e) is homophonous with the complementizer in Romani, an observation that suggests a straightforward analysis of the construction: was/so/what is not a question word or a scope marker in (30e), but rather the agreeing form of a complementizer - which agrees with its specifier position hosting a (silent) copy of a whphrase created by overt movement of an element into an even higher clause. We assume that this analysis is correct. (31a,b) are taken from Anyadi & Tamrazian (1993), who locate speakers accepting these sentences in the Ruhr dialect of German. This does not appear entirely correct, but there are dialects which tolerate these structures 23. We will comment on (31) at the end of this section. (31) a.!welchem Mann glaubst du wem sie das Buch gegeben hat which man believe you who she the book given has "which man do you think that she has given the book to? b.!mit welchem Werkzeug glaubst du womit Ede das Auto repariert hat with which tool think you what-with Ede the car repaired has "with which tool do you think that Ede has repaired the car?" 2.1. Some Facts about the CopyConstruction Let us turn, then, to the Copy Construction CC, and see how it fits into our analysis. The CC is characterized by a number of interesting generalizations, two of which are fairly standard. First, no copy may appear in the root position of the wh-chain. 10
11 Economy of Pronunciation (32) *wen denkst du wen sie [ VP wen liebt] who think you who she who loves "who do you think she loves" Overt copies show up in Spec,CP only. If infinitive clauses have no Spec,CP position in German, the non-wellformedness of (33) is explained immediately. (33) *wen versuchst du wen zu küssen who try you who to kiss *wen batest du mich wen zu küssen who asked you me who to kiss Keeping an overt copy in a root position (in addition to the copy in the scope position) does not only imply violations of PRONECON, it incurs a further violation of WH-IN-SPEC. As long as there is no reason to keep a copy there (and there is none), (33) (34) always give way to alternative structures in which the lowest copy is not spelled out. A second generalization concerns the nature of all overt copies of the wh-chain (but the lowest one): they may not be syntactically complex: (34) *wessen Studenten denkst du wessen Studenten wir kennen whose student think you which student we know "whose student do you think that we know?" *wieviel Studenten denkst du wieviel Studenten wir kennen how many student think you how many students we know "how many student do you think that we know" (35) womit denkst du womit er sie verletzt hat with what think you with what he her hurt has "with what do you think that he has hurt her" (36) (?*)mit was denkst du mit was er sie verletzt hat with what think you with what he her hurt has "with what do you think that he has hurt her" While CCs that involve wh-phrases that consist of a single word only are perfect in all dialects that allow the CC at all, the situation differs radically when the wh-phrase is syntactically complex: in the CC ungrammaticality arises in quite a number of dialects/idiolects as soon as the upper copy contains two or more words (see (34)). (35) - (36) form a nice minimal pair in this respect - the structures do not differ in meaning but just in the fact that womit is a single word, in contrast to mit was. For some (most?) speakers, a contrast exists between (35) and (36) - with the ungrammaticality of the former type being rather mild only. There is practically nobody who would go beyond (36) in terms of the complexity of the copied wh-phrase, though. It has been assumed (Fanselow & Mahajan 1996, Höhle 1996) that this anti-complexity restriction affects all copies in the same way, but this claim overlooks the greater flexibility we observe for the lowest copy: (37) wen denkst du wen von den Studenten man einladen sollte who think you who of the students one invite should "which of the students do you think that one should invite?" (38) wieviel sagst du wieviel Schweine ihr habt how many say you how many pigs you have "how many pigs do you say that you have?" 11
12 Gisbert Fanselow & Damir Ćavar That there is a syntactically complex wh-phrase in the specifier position of the complement clause is obvious in (38), and can be argued for easily in the case of (37) as well: note that the boldface material precedes the pronoun man, to the left of which it is normally impossible to scramble PPs. That the boldfaced material of the lower copy forms one constituent (and not two, with the PP being scrambled to second position) is also obvious in those dialects of German that allow a mixing of long movement and CC, so that constellations like (39) may arise, in which the underlined copy wen von den Studenten can have reached the intermediate clause only by mono-constituental wh-extraction, since there is no long scrambling in German. (39) wen denkst du wen von den Studenten sie sagte dass man einladen sollte who think you who of the students she said that one invite should "which of the students do you think she said that one should invite?" (40), on the other hand, shows that it is not sufficient for grammaticality that one copy only in the chain is syntactically complex: (40) *wen von den Studenten denkst du wen man einladen sollte which of the students think you who one invite should Finally, in those dialects which have little problems with (36), (41) is perfect as well. PPs are strict islands for movement in German, so aus Konstanz could not possible have ever left an wen aus Konstanz by standard movement. Thus, there is no alternative to an analysis of (41) in which the lower Spec,CP position is occupied by an wen aus Konstanz, and the upper one by an wen. (41) an wen denkst du an wen aus Konstanz man das schicken darf at who think you at who from Constance one that send may "at which person from Constance do you think one is allowed to send that?" The complexity restriction thus does not apply to the lowest copy. The complexity restriction holding for upper copies renders wh-copying ungrammatical whenever a wh-phrase cannot be split or "separated", as it is e.g. the case for which-phrases. (42) a. welches denkst du welches er nehmen wird which think you which he take will "which one do you think he will take?" b. *welches denkst Du welches Schweinderl er nehmen wird which think you which piggie he take will "which piggie do you think he will take? c. *welches Schweinderl denkst Du welches Schweinderl er nehmen wird The CC obeys stricter locality restrictions than standard long overt movement, as Höhle (1996) and others have observed, see (43) corresponding examples with simple long movement would be grammatical. What we get is exactly analogous to the intervention effect Beck (1996) and Pesetsky (1998b) identify for German wh-in situ: no operator may intervene between the copies of the whphrase. (43) *wen glaubt keiner wen sie liebt who believes nobody who she loves "who does nobody believe that she loves" *wen glaubt jeder wen sie liebt who believes everyboy who she loves "who does everybody believe that she loves" 12
13 Economy of Pronunciation 2.2. Three analyses There have not been too many proposals for an analysis of the CC, but Inge Hiemstra's (1986) theory of the construction (and its Frisian counterpart) is certainly outstanding in many respects. Published nearly ten years before Chomsky (1995), her contribution preempts insights of much work in featural movement theory in a number of respects. The central idea of her analysis of (44) is that when a structure requires wh-movement, this may be carried out as either - movement of the wh-feature alone - the pied piping of the ϕ-features aligned with the wh-feature - the pied piping of the whole phrase bearing the wh-feature. The resulting system is, thus, quite reminiscent of a movement theory generally adopted later in the mid-nineties. The first two options for effecting movement must then be complemented by a theory of spell-out for the displaced feature complexes. According to Hiemstra, it is the most unmarked lexical element bearing the relevant features that will realize the feature complex in question. A single [+wh]- feature is therefore realized as was (=44a), the feature complex [+wh, 3rd ps., acc] as wen (=44b). (44) Pure feature movement: a. was denkst du wen sie eingeladen hat what think you who she invited has Pied-piping of ϕ-features b. wen denkst du wen sie eingeladen hat Pied-piping of full phrase: c. wen denkst du dass sie eingeladen hat "who do you think she has invited" If Fanselow & Mahajan (1996, 2000) are correct, was in (44a) is a sentential wh-expletive originating in the object position of the matrix clause. The parallel between (44a) and (44b) would thus be a spurious one. It is also not too clear to what extent we want to consider wieviel as in (38) or an wen as in (41) to be mere spellouts of complexes of ϕ-features. For welche, one would have to give an answer to the question of why it allows ϕ-featural copying only if what is left behind is a single word item. Thus, a more promising version of her approach would seem to have to move closer to the ideas developed in Chomsky (1995), in assuming that the minimal element that can be moved overtly is the word carrying the attracted feature (=wen, wieviel, welche). We can then analyze the CC as a sequence of two types of movement steps, the first one involving the pied-piping of the phrase dominating the attracted word, the second one confining itself to the overt displacement of the word-level category. With the exception of those dialects that can prepose a minimal PP like an wen in the CC, this approach is descriptively correct, but it leaves open some questions: why is the wh-word that moves in the second step not deleted in the phrasal copy, as in (45) (which would be grammatical if the second occurrence of wieviel had been kept phonetically)? (45) *wieviel denkst du wieviel Schweine sie sagt dass wir haben how many think you how many pigs she says that we have "how many pigs does she say that we have" Fanselow & Mahajan (1996,2000) concentrate on the development of an account for (44a), but add a sketch of an analysis of the CC, too. Their analysis makes use of the fact that the Comp position cannot be empty phonetically in German embedded clauses 24 (except in indirect question complements), and they assume that the copies of the wh-phrase occupying the lower SpecC position 13
14 Gisbert Fanselow & Damir Ćavar cliticize onto Comp whenever this position is not filled in another way. For wh-phrases of the size of a single word, this approach works smoothly, but they did not take into account the principled availability of more complex constructions like (37) and (38), for which it is hard to believe that the complete wh-phrase occupies the Comp position. Pesetsky's (1997, 1998a) crucial insight concerning uneconomical pronunciations of wh-chains is that they typically arise in contexts where standard movement would violate island conditions. In fact, one often finds the CC in dialects that do not allow long movement of arguments. In a form slightly adapted to the general approach we pursue here, the relevant Island Constraints takes the form (46): (46) ISLAND *α... [ Σ β... ] where α, β belong to a single chain, α or β are unpronounced, and Σ is an island. Suppose in the dialects allowing CCs, CPs are (or, can be) barriers for extraction. In the first derivational step for a long distance question, the wh-word will be copied to Spec,C at some stage: (47) wen dass du wen eingeladen hast who that you who invited have Whether dass may be preserved in the overt presence of phonetic material in SpecC is, partially, a function of the rank of the Doubly Filled Comp Filter in German it is violable in some but not all varieties of German (see e.g. Heck 1997, for a pertinent OT analysis), and we will not go into this issue here. When it comes to the optimization of the pronunciation of (47), a number of principles comes into play. In addition to PARSESCOPE introduced above, LEFTEDGECP taken over from Pesetsky (1998a:341) 25 seems relevant: (48) LEC The first pronounced word in a CP must be the complementizer projecting that CP. If PARSESCOPE has a high rank in German (as it seems to be the case), PARSESCOPE >> LEC guarantees that wen can be kept in the initial position of (47) whenever (47) represents a complete complement question or forms part of a larger movement structure. PRONECON implies that one of the two occurrences of wen disappears, if PARSESCOPE >> LEC, it is the higher copy that must be retained. We thus end up with (49). (49) wen dass du wen eingeladen hast who that you who invited have Suppose that the head and the specifier of a CP/a phase (but no other elements) are accessible in the next optimization cycle. (50) represents the AGR-OP or vp of a matrix clause that will end up as a matrix question. If Σ is not interpreted as an island, PARSESCOPE implies that the upper copy of wen is retained, and LEC implies that the lower copy of wen should disappear. (50) wen denkst [ Σ wen dass du t eingeladen hast] who think 2sg who that you invite have If Σ is interpreted as an island, ISLAND blocks the deletion of the lower copy if ISLAND >> LEC. In fact, we might capture the dialectal variation we find in German by assuming that Σ=CP is always counted as a barrier and that the ranking of ISLAND and LEC is not fixed in the same way in the various dialects of German: the copy construction arises when ISLAND >> LEC, while we get long movement when LEC >> ISLAND, as tables 1 and 2 show. 14
15 Economy of Pronunciation (51) a. wen denkst wen dass du eingeladen hast b. wen denkst wen dass du eingeladen hast c. wen denkst wen dass du eingeladen hast d. wen denkst wen dass du eingeladen hast Table 1 PARSESCOPE ISLAND LEC!51a * 51b *! 51c *! 51d *! Table 2 PARSESCOPE LEC ISLAND 51a *!!51b * 51c *! 51d *! A cyclic application of the principles discussed so far, together with the assumption that the specifiers and heads remain accessible for optimization from outside, thus yields the CC under the ranking given in table 1. Why is it that upper copies must be non-complex? We can derive this from the principles PRONECON and PARSESCOPE if we interpret them properly. In showing how, we may confine our attention to the derivational step linking a wh-phrase in Spec,C to its first landing site in the matrix clause. In (53), we ignore one candidate structure to which we will return later. (52)... wen von den Studenten denkst wen von den Studenten dass who of the students think who of the students that --- (53) a. --- wen von den Studenten denkst wen von den Studenten dass --- b. --- wen von den Studenten denkst wen von den Studenten dass --- c. --- wen von den Studenten denkst wen von den Studenten dass --- Presupposing the ranking ISLAND >> LEC, we need to consider only those candidates that retain phonetic material both in the lower SpecC and in the first matrix clause landing site position. Obviously, PRONECON is violated more often in (53a) than in (53b,c) it has three more words than the two other candidates. If PRONECON is ranked below ISLAND (so that copying is possible at all), the former principle will still block (53a). The decision between (53b,c) seems to follow from PARSESCOPE, if we interpret the principle as applying to semantic units, and if we make the fairly standard assumption that the restrictor of a whquantifier should not appear in the scope position of the operator, i.e., if (54a) is preferred over (54b) (see e.g. Chomsky 1993, Fox 1996). (54) a. wh-x. (Pred(x)). b. wh-x, x a Pred..x. 15
16 Gisbert Fanselow & Damir Ćavar The higher the restriction of a wh-operator is moved in a tree, the more violations of PARSESCOPE arise relative to it, so that (53b) is favored over (53c). The core properties of the CC thus seem derived. Note, however, that an account of (53b) vs. (53c) in terms of PARSESCOPE makes the incorrect prediction that a wh-phrase that can be split up must be so. This is false, as (55) shows (55) a. wen von den Studenten kennst du who of the students know you wen kennst du von den Studenten b. was für Studenten kennst du was for students know you was kennst du für Studenten "what kind of students do you know! c. wieviel Studenten kennst du denn how many students know you then wieviel kennst du denn Studenten "how many students do you know" We therefore need a principle that penalizes structures that have been contiguous at level L but cease to be so at level L'. (56) CONTIGUITY IN SYNTAX (CIS) The phonetic material corresponding to a constituent must be spelled out in one position only CIS disfavours separation, whereas PARSESCOPE requires it. When the two constraints are tied, the constellation we find in (55) arises 26. The tie with PARSESCOPE implies a fairly high rank for CIS, in particular, it dominates PRONECON. Therefore, we must understand CIS in a such way that it is satisfied when there is at least one copy of a phrase that is pronounced in an unsplit fashion - otherwise, the CC would be ruled out because it would always imply a CIS violation. We must make sure, however, that the tie between PARSESCOPE and CIS does not imply that complex wh-phrases (which always contain a restrictor that should be left in situ) do not have to move at all (because the ParseScope violation by the operator part is always counterbalanced by the PARSESCOPE violation of the restrictor). This is effected by the principle WH-IN-SPEC introduced above. Consider, then, the consequence of CIS for the two crucial movement steps the one from the root position to Spec,C, and the subsequent step mapping the wh-phrase into the matrix clause. As table (3) shows, we correctly predict the distribution of grammaticality in the first movement step of (57) (57) wen von den Studenten du wen von den Studenten einlädst who of the students you who of the students invite (58) a. *wen von den Studenten du wen von den Studenten einlädst b. wen von den Studenten du wen von den Studenten einlädst c. wen von den Studenten du wen von den Studenten einlädst d. *wen von den Studenten du wen von den Studenten einlädst 16
17 Economy of Pronunciation Table 3 PARSESCOPE /CIS ISLAND WH-IN-SPEC PE LEC 58a *(restrictor) * ****! 58b *(restrictor)! 58c *(contiguity) 58d * (operator) * If the derivation proceeds with (58c), nothing new happens: a wh-phrase consisting of a single word cannot violate CIS. If (58b) is chosen, we proceed as discussed in the context of (53a-c) (repeated here as (59a-c), but it is also obvious that we deliberately ignored a candidate above: (59d). Nevertheless, (59b) is still optimal: CIS is respected by the lower copy, we do not move the restrictor of the whoperator further up, and the wh-operator moves to a position c-commanding its scope. (59) a. --- wen von den Studenten denkst wen von den Studenten dass --- b. --- wen von den Studenten denkst wen von den Studenten dass --- c. --- wen von den Studenten denkst wen von den Studenten dass --- d. --- wen von den Studenten denkst wen von den Studenten dass --- Table 4 PARSESCOPE /CIS ISLAND WH-IN-SPEC PE LEC 59a *!(restrictor) **** *! 59b * * 59c *!(restrictor) * * 59d *!contig * * Therefore, we have derived the fact that the copy construction allows complex overt wh-phrases in the lowest SpecC position only. As we have remarked above, this restriction can be minimally violated in certain dialects in which a PP may be copied, cf. (36), repeated here as (60) (60) (?*)mit was denkst du mit was er sie verletzt hat with what think you with what he her hurt has "with what do you think that he has hurt her" Given what we have seen so far, the optimal candidate SHOULD be one that "strands" the preposition in the lower copy. For the constraint that makes (60) possible by overriding PRONECON, a natural formulation comes to mind: Note that the copying operation moves a PP category upwards, and one may assume that phonetic material that does not contain a preposition in the head position cannot be a phonetic realization of a PP: (61) LEFT EDGE PP (LEP) The leftmost element realized phonetically in a PP must be the preposition from which that PP was projected If LEP is ranked above of PRONECON, the prepositional head may be retained in a CC. Our final task in describing the CC is a discussion of the intervention effect. As (62) shows (see also Beck 1996), an operator like sentential negation cannot intervene between a wh-phrase and its restrictor the problem can be solved in various ways, by respecting contiguity (62a) or by scrambling the wh-phrase in front of the operator (62c) before it is split up. 17
18 Gisbert Fanselow & Damir Ćavar (62) a. was für Bücher hat er nicht gelesen what for books has he not read b. *was hat er nicht für Bücher gelesen c. was hat er für Bücher nicht gelesen According to Pesetsky (1998b), the relevant anti-intervention constraint implies that a wh-operator must not be separated from its restrictor by a further operator. This applies to those copy constructions in a straightforward way in which a restrictor is indeed stranded, by the same effect shows up in (63), i.e. when the wh-phrase consists of a single word only. (63) fits into Pesetsky's proposal if we assume that, semantically, there is a restrictor part present for wen as well, and that this restrictor part is left behind with the lowest visible copy of the phrase. Alternatively, the intervention constraint might be reformulated so that it requires that links in a chain in which both elements contain either phonetic or semantic material must not be separated by an operator. (63) *wen glaubt keiner wen sie liebt who believes nobody who she loves "who does nobody believe that she loves" *wen glaubt jeder wen sie liebt who believes everyboy who she loves "who does everybody believe that she loves" The only potential problem that arises, then, is related to ineffability. A sufficiently high rank of the intervention constraint will be able to block the copy construction in the situations where this is called for, so that the winning competitor is a long movement construction. The same consequence arises for constellations in which the wh-phrase must not be split up (which-phrases, or PPs, in certain dialects). This is an acceptable result for those dialects in which the copy construction co-exists with long movement, but it does not capture the ineffability effect that can arise for long distance dependencies when a dialect forbids long movement and an intervention effect rules out the copy construction at the same time. A standard solution (see Legendre et al. 1998) would be to rank the intervention constraint higher than faithfulness constraints concerning scope assignments for the intervening operators Some related issues As we have mentioned above, certain varieties of German (e.g. Lower Rhine area, Bavarian Suabia) allow to construct questions in the form given in (64). (64) welchen Mann denkst du wen er kennt which man think you who he knows "which man do you think he knows" We have little to say about this construction, except for the observation that it is not likely to be a subcase of a CC. While the lower occurrence of a wh-element is a non-complex one, it does not copy the wh-operator of the upper wh-phrase. It is rather the minimal spellout of the wh-features that should be present in the lower Spec,C position, as (64) and (65) show 27 : (65) a. wieviel Bier denkst du was er trinkt how much beer think you what he drinks "how much beer do you think that he drinks?" b. *wieviel Bier denkst du wieviel er trinkt 18
19 Economy of Pronunciation One simple analysis would analyze wen and was as agreeing forms of the complementizer. This would be consistent with the observation that (35b) is judged worse than (35a) (repeated here as (66)), if we assume that womit makes a bad [+wh]-complementizer. (66) a.!welchem Mann glaubst du wem sie das Buch gegeben hat which man believe you who she the book given has "which man do you think that she has given the book to? b.!mit welchem Werkzeug glaubst du womit Ede das Auto repariert hat with which tool think you what-with Ede the car repaired has "with which tool do you think that Ede has repaired the car?" Alternatively, the contrast in (66) might be caused by the fact that German dialects tend to not block standard long movement when PPs are affected, so that (66b) might be blocked by a candidate involving long movement. We might also consider wen, was, and womit as spellout forms for ϕ features of a wh-phrase that has lost its original phonetic content. In a dialect that ranks ISLAND over LEC, the insertion of "expletive" phonetic material spelling out wh-ϕ-features is an alternative means of avoiding an ISLAND-violation. Suppose that FI blocks the use of expletive material, and suppose that PRONECON works in such a way that it blocks the repetition of phonetic material only. Then if FI >> PRONECON, we still get the CC, but if the ranking is reversed, the situation in (64) - (66) arises. We do not wish to commit ourselves to this analysis, though. German dialects allows at least two further examples of "uneconomical pronunciation". When the highest verbal element of a clause is topicalized as in (67), the problem arises that the second position should be filled by exactly this element, too, given that the V/2 constraint is not violated in German. The problem may be solved by expletive insertion (67b) or by uneconomical pronunciation (as in (67a)). Retention of two copies is confined to modals, though. (67) a. können kann ich nicht can can I not "I am not ABLE to" b. können tue ich nicht can do I not c. schlafen tue ich nicht sleep do I not Sleeping is not what I do" d. *schlafenschlafe ich nicht sleep sleep I not In a simple split construction (68) (see also next section), the phonetic material belonging to a single constituent is distributed over two places in the sentence without any repetitions, but there are two exceptions to this property of split XPs. In those dialects in which a PP can enter the split construction, the preposition must appear in both positions in which the PP is spelled out partially (68b) 28. Given the high rank of LEP, this is not unexpected. Likewise, Riemsdijk (1987) observes that the indefinite article may (and sometimes) must be repeated in split noun phrases a fact we can relate to the observation that singular count noun phrases may never be realized phonetically without an initial determiner. (68) a. teure Bücher habe ich viele expensive books have I many "I have many expensive books" 19
20 Gisbert Fanselow & Damir Ćavar b. In Schlössern habe ich noch in keinen gewohnt in castles have I yet in no lived c. *In keinen habe ich noch in Schlössern gewohnt d. einen amerikanischen Wagen kann ich mir nur einen grünen leisten an American car can I me only a green afford "I can only afford a green American car" Thus, the split construction supports the idea that the economisation of pronunciation is quite in general subject to other constraints. 3. A few remarks on the split construction and so-called head movement A detailed analysis of the split construction is beyond the scope of the present paper, and would mostly repeat what is said in Ćavar & Fanselow (1997, 2000). The following remarks are meant to prepare the discussion of a further advantage of a pronunciation economy account: it gives a straightforward analysis of so-called head-movement. That constructions apparently involving rightward movement might (at least in some contexts) have to be re-analyzed as resulting from the stranding of phonetic material in a leftward movement operation was proposed, e.g., by Kayne (1994) or by Wilder (1995). It seems also obvious that the "stranding" of β (say, a relative clause) in the process of moving Σ (say, a DP) in (69) could be the result of an incomplete deletion 29 in the source position of movement, following by an erasure of β in the target position, due to PRONECON. (69). [ Σ α β]...[ Σ α β ] [ Σ α β]...[ Σ α β ] [ Σ α β]...[ Σ α β ] [ Σ α β].. In OT terms, the candidate set for the pronunciation of a movement chain (70) is simply enlarged by allowing (free) partial deletion in the copies created by movement. (70) --- X Y Z --- X Y Z ---- (71) a. X Y Z --- X Y Z b. X Y Z --- X Y Z c. X Y Z --- X Y Z d. X Y Z --- X Y Z e. X Y Z --- X Y Z f. X Y Z --- X Y Z g. X Y Z --- X Y Z h. X Y Z --- X Y Z Candidate (71a) violate PRONECON three times, while (71b) and (71c) satisfy this constraint and represent full overt and full covert movement, respectively. Candidates (71d,e) represent split constituents (the pronunciation of the copies created by movement is distributed over two places) - they imply a CIS violation that is justified only when higher constraints like PARSESCOPE are fulfilled thereby. (71e) implies a (presumably fatal) RECOV violation, whereas the PRONECON violation in (71g) is acceptable only if a constraint like LEP forces it. Additional constraints may come into play: the PARMOV constraint of Müller (1998a) will imply that the c-command relations between the phonetic occurrences of X,Y, and Z should not change when the phrase is split up by free deletion - this is respected in (71c,d), but not in (71h). 20
21 Economy of Pronunciation Ćavar & Fanselow (1997, 2000) extend this approach that we have presupposed above for left branch extractions to the split construction that one finds in German (68a), Russian, Polish, Croatian and many other languages, and that has so far resisted a satisfactory treatment. For constructions such as (68a), the obvious alternative to analyzing the split construction in terms of partial deletion is simple extraction. But notice that split constituents do not respect standard islands for movement. E.g., PPs are islands for standard movement in Croatian and Polish, yet, PPs may be split up freely, as the examples in (70a-b) illustrate. Likewise, German PPs are extraction islands, but split constructions arise, as (68b) and (70c) show. See Fanselow (1988, 1993), Kuhn (1998), van Geenhoven (1996), and in particular Ćavar & Fanselow (2000) for a detailed discussion of the shortcomings of simple movement accounts. (70) a. na kakav je Ivan krov skočio (Croatian) on what-kind has Ivan roof jumped? "on what kind of roof has Ivan jumped?" b. na jaki Marek dach skocił (Polish) on what-kind Marek roof jumped c. mit was hast Du für Problemen nicht gerechnet with what have you for problems not reckoned "what kind of problem have you failed to take into account" The analysis offered in Ćavar & Fanselow (2000) can be re-cast in OT terms in the following way: the split construction arises because CIS is ranked below (or rather tied to) PARSESCOPE as applied to specific pragmatic-semantic features. Thus, for Croatian, Polish, or German it can be shown that a DP or PP is split up just in case its phonetic material is linked to at least two different pragmatic (focus) or semantic (wh-) features. Thus, suppose that krov bears a focus feature and na kakav a wh-feature in (70a), and suppose that focus features have to be checked in a focus position in Croatian. If PARSESCOPE is ranked higher or at least as high as CIS, the PP na kakav krov need not or cannot be realized in a single structural position it is split up 30. Up to now, we have only considered cases in which semantic constraints (but see note 30) require that a phrase is linearized discontinuously after movement. It is natural to suspect that conditions relating to the phonological interface may have the same effect. Consider (71) in this respect. It is a notorious fact that stressed particles have to be stranded (71b,c) in German (and Dutch) when a verb-second clause is being formed, while unstressed particles go along with the verb (71e,f), (71) a. dass er das Buch zu lesen an-fängt that he the book to read begins b. er fängt das Buch zu lesen an c. *er anfängt das Buch zu lesen d. dass er das Buch zu lesen ver-sucht that he the book to read tries e. *er sucht das Buch zu lesen ver f. er versucht das Buch zu lesen Depending on one's assumptions concerning the internal structure of [an+fangen], (71b) violates what was called the "Lexical Integrity Principle" in earlier versions of the generative approach, which would rule out an operation like movement that affects a part of a lexical item only. If we want to avoid such a violation, we could assume that anfangen is structured as [ V an [ V fangen]], and that head movement affects the smallest item only to which it is applicable. But particles must be stranded in constructions involving movement to Comp only: in certain varieties or German (but not in Dutch, see 21
22 Gisbert Fanselow & Damir Ćavar Roberts 1993) the particle must not be stranded in contexts of non-finite verb incorporation (see 72-73). (72) dass er es zu lesen hätte anfangen müssen that he it to read had begin must "that he should have begun to read it" (73) *dass er es zu lesen an hätte fangen müssen In the system proposed here, (71) (73) constitute no particular problem. The principle OPW introduced in (74) blocks the appearance of more than one prosodic word in the position to which the finite verb moves. If OPW >> CIS, a particle that constitutes a second prosodic word must not be realized in the landing site of verb movement, but a stressless particle will, on the other hand, appear in second position together with the verb proper, because the CIS-violation which the stranding of the particle incurs is not justified by the avoidance of a clash with OPW in this case. (74) ONEPROSODICWORD (OPW) The second position of the clause may host a single prosodic word only. In other words, a full lexical category is copied to second position in (71b) and (71e), but whether its complete phonetic material may show up there depends on conditions of the syntax-phonology interface. Note that OPW is also justifiable on the basis of the observation that pronouns that are adjoined to that second position must in fact be phonological clitics, because otherwise, (74) would be violated (see Ćavar 1999 for a related view). Recourse to partial deletion and OPW does not only allow us to respect the Lexical Integrity Hypothesis. It also helps avoiding excorporation analyses (Roberts 1991) for cases like (75): there is good reason to believe that a verbal complex anrufen hätte können is formed in the derivation of (75), in which anrufen first incorporates into können, with the complex so formed moving up further to the finite auxiliary (this may explain, e.g. the need to replace the past participle gekonnt by the infinite können, see Schmid 1999 for an analysis). In standard analyses, hätte then has to be excorporated out of this verbal complex when it moves to second position, but we can avoid this conclusion by assuming that the whole verbal complex anrufen können hätte is moved to second position, in which, due to OPW, only the highest prosodic word may be spelled out in second position, however. (75) er hätte Maria anrufen können he had Mary call up could "he could have called Mary up" In fact, Roberts (1997) comes quite close to a similar line of reasoning in his discussion of Romance restructuring constructions, for which he assumes that chains created by overt movement are spelled out in the higher position, unless this violates a constraint *[ W WW] which bans two "syntactic" words dominated by a single X category. For obvious reasons, a position characterized by (74) may even be targeted by full phrasal copying without any change in the phonetic effect: whatever is copied to second position can find a single word spellout there only; the remaining part must be realized in situ. We may therefore hypothesize that movement is always phrasal copying: the overt appearance of the movement of a head in isolation arises when additional constraints imply that little more (or less) than the head of the copied phrase may appear phonetically in the landing site. For many instances of apparent head movement, OPW is too strong (recall that verb inversion need not strand a particle in German), so it can and must be complemented by a principle like (76), with proper choice of S. 22
23 Economy of Pronunciation (76) NoPhrasalSpec NPS The phonetic realization of specifier S must not contain material that forms a phrase. Under this perspective, there is an analysis for apparent verb movement to Infl in French (77a) in which the whole VP has been moved to the innermost specifier position of Infl (77b-c). The upper copy of the VP must then be reduced phonetically to its head if NPS must be respected by the innermost specifier of Infl. Likewise, the head must be eliminated phonetically in the lower copy because of PRONECON. (77) a. il embrasse souvent Marie he kisses often Mary b. il... souvent embrasse Marie c. il embrasse Marie souvent embrasse Marie d. il embrasse Marie souvent embrasse Marie It thus seems clear that such a re-analysis of head movement is a possible extension of our approach, so that the question arises whether it is also a step that is called for. Several proposals (Kayne 1998, Koopman & Sportiche 1999, Mahajan 1999) have indeed been made that imply that at least certain instances of head movement should be re-interpreted as phrasal movement. Some of the obvious advantages of the resulting systems are: Phrasal movement can always be carried out in such a way that it extends the phrase marker (=it targets the root node) that is being constructed. The movement of a head to another head position does not fulfill this extension requirement on obvious grounds: when head F moves out of ZP to H in [H ZP], it does not land at the root node dominating H and ZP. One unwelcome consequence of this is that a head H moved to G does not c-command its root position, quite in contrast to what one would normally assume to hold for movement. Head movement violates most generalizations one would like to defend in the context of a movement theory. In a feature driven theory of movement, it is not obvious what factor determines whether a given feature must be eliminated by head movement or by phrasal movement. In the configurational definition of phrasal levels (Speas 1990, Chomsky 1995), an element E is a maximal projection if its mother node is not a projection of E. If a head H adjoins to head G, neither G nor G' are projected from it. Therefore, the head H should become a maximal projection by movement. Even if this consequence was tenable, it would induce a violation of the chain uniformity condition (Chomsky & Lasnik 1991), because the trace of the head is not maximal. We therefore seem to possess excellent evidence for the elimination of head movement, but the approaches that are discussed in this context face at least one potential problem. Typically, the visible displacement of a single word W is re-analyzed as the remnant movement of a WP out of which all material but W has been extracted. The derivation of (77a) would therefore proceed as in (78) (see, e.g., Mahajan 1999): (78) il souvent [ VP embrasse Marie] il souvent Marie [ VP embrasse Marie] il souvent Marie [ VP embrasse t ] il [ VP embrasse t ] souvent Marie [ VP embrasse t ] il [ VP embrasse t ] souvent Marie t The movement operations that "evacuate" the VP or any other XP before "head"-movement are typically unmotivated in terms of feature checking or semantic considerations, which may (but need not) create a problem in minimalist accounts but not necessarily in OT, since the relevant STAY* 23
24 Gisbert Fanselow & Damir Ćavar violations may be called for by the need to respect a higher ranked NPS or OPW. But we know since Wexler & Culicover (1980) that movement has a freezing effect on phrases P in the sense that P becomes an island after movement, as demonstrated impressively in Müller (1998a). "Head" movement constellations do not induce such island effects, however: (79) über wen liest sie eine Geschichte t about who read she a story "who does she read a story about" In (79), the verb has been proposed. If this implies phrasal fronting in the sense that eine Geschichte über wen has been extracted out of VP before the rest of VP (=liest) was preposed, then eine Geschichte über wen should have become an island for movement, which it has not: über wen can still be moved out of this phrase Our account avoids this problem because the splitting up of the VP into a preposed verb and a stranded rest is effected by pronunciation laws (and not be remnant movement)- there is no particular reason for why the object should become an island thereby. A comparison of attempts to guarantee that certain movement have no freezing effects (Müller 1999) and our approach may thus help to identify the proper way of eliminating head movement. References Ackema, P. & A. Neeleman WHOT?. In: P. Barbosa, D. Fox, P. Hagstrom, M. McGinnis & P. Pesetsky (eds). Is the Best Good Enough?. Cambridge, Mass.: MIT-Press Anyadi, S. & Tamrazian, S Wh-movement in Armenian and Ruhr German. UCL Working papers in Linguistics 5: Aoun, J. & A. Li Wh-elements in situ: Syntax or LF?. Linguistic Inquiry 24: Barbiers, S Remnant stranding and the theory of movement. Paper presented at the Workshop on Remnant Movement, Feature Movement and their Implications for the T-Model. Potsdam, July Basilico, D Wh-movement in Iraqi Arabic and Slave. The Linguistic Review 15:4. Beck, S Quantified structures as barriers for LF movement. Natural Language Semantics 4: Bobaljik, J Morphosyntax. The syntax of verbal inflection. Doctoral dissertation, MIT. Ćavar, D. & G. Fanselow Split Constituents in Germanic and Slavic. Paper, presented at the International Conference on Pied Piping, Jena. Ćavar, D. & G. Fanselow Discontinuous constituents in Slavic and Germanic languages. Ms. Ćavar, Damir Aspects of the Syntax-Phonology Interface. Doctoral dissertation, University of Potsdam. Chomsky, N Barriers. Cambridge, Mass.: MIT-Press. Chomsky, N A minimalist program for linguistic theory. In: K. Hale & S.J. Keyser (eds). The View from Building 20. Cambridge, Mass.: MIT-Press Chomsky, N The minimalist program. Cambridge, Mass.: MIT-Press. Chomsky, N Minimalist Inquiries: The framework. Ms., MIT: Chomsky, N. & H. Lasnik Principles and Parameters Theory. In: J. Jacobs, A.v. Stechow, W. Sternefeld & Th. Vennemann (eds). Syntax. An International Handbook of Contemporary Research. Berlin: de Gruyter. 24
25 Economy of Pronunciation Cole, P. & G. Hermon The typology of wh-movement. Wh-questions in Malay. Syntax 1: du Plessis, H Wh movement in Afrikaans. Linguistic Inquiry 8: Fanselow, G Aufspaltung von NPn und das Problem der "freien" Wortstellung. Linguistische Berichte 114: Fanselow. G Die Rückkehr der Basisgenerierer. Groninger Arbeiten zur Germanistischen Linguistik 36: Fanselow, G Partial Movement. SynCom Project. Fanselow, G. & A. Mahajan Partial Movement and Successive Cyclicity. In: U Lutz & G. Müller (eds). Papers on Wh-Scope Marking. Arbeitspapier 76 des Sonderforschungsbereich 340. Stuttgart & Tübingen, Fanselow, G. & A. Mahajan Towards a minimalist theory of wh-expletives, wh-copying, and successive cyclicity. In: U. Lutz, G. Müller & A. v. Stechow (eds). Wh-scope marking. Amsterdam:_ Benjamins. Fanselow, G.. R. Kliegl, & M. Schlesewsky "Long" Movement in Northern German: A Training Study. Manuscript, University of Potsdam. Fox, D Condition C effects in ACD. MIT Working Papers in Linguistics 27: van Geenhoven, V Semantic Incorporation and Indefinite Descriptions. Doctoral dissertaion, Universität Tübingen. Grewendorf, G Multiple wh-fronting, Ms., Universität Frankfurt. Grimshaw, Jane Projection, heads, and optimality. Linguistic Inquiry 28: Groat, E. & J. O'Neill Spellout at the LF-interface. In: W. Abraham, S.D. Epstein, H. Thraínsson & J.W. Zwart (eds). Minimal ideas Heck, F Komplementierer und ihre Spezifikatoren, Ms., Univ. Tübingen. Heck, F. & G. Müller Repair is Local. Paper, presented at the Workshop on Conflicting Rules in Phonology and Syntax. Potsdam, December Hiemstra, I Some aspects of wh-questions in Frisian. NOWELE 8: Hinterhölzl, R Licensing movement and stranding in the West Germanic OV languages. Höhle, T German w...w-constructions. In: U Lutz & G. Müller (eds). Papers on Wh-Scope Marking. Arbeitspapier 76 des Sonderforschungsbereich 340. Stuttgart & Tübingen, Huang, C.-T. J Move wh in a language without wh-movement. The Linguistic Review 1: Kayne. R The antisymmetry of syntax. Cambridge, Mass.: MIT-Press. Kayne, R Overt vs. Covert Movement. Syntax 1: Koopman, H. & A. Szabolcszi.Verbal Complexes. Ms. UCLA Kuhn.J Resource Sensitivity in the Syntax-Semantics Interface and the German Split NP Construction. Ms. Kvam, S Linksverschachtelung im Deutschen und Norwegischen. Tübingen: Niemeyer. Legendre, G., P. Smolensky, & C, Wilson When is less more?. Faithfulness and minimal links in wh-chains. In: P. Barbosa, D. Fox, P. Hagstrom, M. McGinnis & P. Pesetsky (eds). Is the Best Good Enough?. Cambridge, Mass.: MIT-Press Legendre, G. in press. An introduction to optimality theoretic syntax. In: G. Legendre, J. Grimshaw, & S. Vikner (in press). Optimality Theoretic Syntax. Cambridge, Mass.: MIT-Press. Lutz,U., G. Müller, & A. v. Stechow. 2000, Wh-scope marking. Amsterdam: Benjamins, 25
26 Gisbert Fanselow & Damir Ćavar Mahajan, A The A/A-bar-distinction and movement theory. Doctoral dissertation, MIT, Cambridge, Mass. Mahajan, A Against head movement in syntax. Ms. UCLA. McDaniel, D Partial and multiple wh-movement. NLLT 7: McDaniel, D., B. Chieu & T. Maxfield Parameters for wh-movement types: Evidence from child English. NLLT 13: Müller, G Partial Wh-Movement and Optimality Theory. The Linguistic Review 14: Müller, G. 1998a. Order preservation, parallel movement, and the emergence of the unmarked. Ms. ROA Müller, G. 1998b. Incomplete category fronting. Dordrecht: Kluwer. Müller, G Shape conservation and remnant movement. NELS 30. Ouhalla, J Remarks on the binding properties of wh-pronouns. Linguistic Inquiry 27: Pesetsky, D Optimality theory and syntax: movement and pronunctiation. In: D. Archangeli & D.T. Langendoen (eds). Optimality Theory: An Overview. Blackwell: Oxford. Pesetsky, D. 1998a. Some Optimality Principles of Sentence Pronunciation. In: P. Barbosa, D. Fox, P. Hagstrom, M. McGinnis & P. Pesetsky (eds). Is the Best Good Enough?. Cambridge, Mass.: MIT- Press Pesetsky, D. 1998b. Phrasal Movement and Its Kin. Ms., MIT. Reis, M Wer glaubst du hat recht? On so-called extractions from verb-second clauses and verbfirst parenthetical constructions in German. In: U. Lutz & J. Pafel (eds). On extraction and exptraposition in German. Amsterdam: Benjamins. Rice,K A Grammar of Slave. Berlin: Mouton de Gruyter. van Riemsdijk, H Movement and regeneration. In: P. Benicà (ed.). Dialect Variation and the Theory of Grammar. Dordrecht: Foris. Roberts, I Excorporation and Minimality. Linguistic Inquiry 22: Roberts, I Restructuring, Head Movement, and Locality. Linguistic Inquiry 28: Roberts, I Have/Be Raising, Move F and Procrastinate. Linguistic Inquiry 29: Sabel, J Principles and Parameters of wh-movement. Habilitation thesis. Frankfurt/M. Saddy, D Wh-scope mechanisms in Bahasa Indonesia. MIT Working Papers in Linguistics 15: Saddy, D A versus A-bar- movement and wh-fronting in Bahasa Indonesia. Ms., University of Queensland. Schmid, T Optional and Obligatory IPP Constructions in Westgermanic. Paper, presented at the second workshop on Optimality Theory Syntax. Speas. M Phrase Structure in Natural Language. Dordrecht: Kluwer. Tsai, D On economizing the theory of A'-dependencies. Doctoral dissertation, MIT. Wexler, K. & P. Culicover, Formal principles of language acquisition. Cambridge, Mass.: MITpress. Wilder, C Rightward movement as leftward deletion. U. Lutz & J. Pafel (eds). On extraction and extraposition in German. Amsterdam: Benjamins Wilder, C Some Properties of Ellipsis in Coordination. In: Studies on Universal Grammar and typological variation. Eds. Artemis Alexiadou & T. Alan Hall, Amsterdam: Benjamins. 26
27 Economy of Pronunciation * Parts of this paper have been presented at the Workshop on Conflicting Rules in Phonology and Syntax at the University of Potsdam (Dec 15-16, 1999), and at the Linguistics Colloquium at the University of Leipzig. We would like to thank the audiences for useful comments and criticism. Particular thanks for helpful hints and for support in various respects related to this article go to Joanna Błaszczak, Caroline Féry, Susann Fischer, Gereon Müller, and Douglas Saddy. We also like to thank Artemis Alexiadou, Hans-Martin Gärtner, Anoop Mahajan, Matthias Schlesewsky, Peter Staudacher, and Chris Wilder. Research for this paper was supported by the grant INK 12/B1 "Innovationskolleg Formale Modelle kognitiver Komplexität" financed by the Federal Ministry of Education and Research and administered by the German Research Foundation. The paper was written while Damir Ćavar was employed by the University of Potsdam. 1 However, we do not share Pesetsky's view that the optimality theoretic aspect of syntax is confined to such principles of sentence pronunciation. 2 In an OT-framework, one expects that the link between the (uninterpretable) features of an attracting head and the creation of copies can be violated in two ways: there should be movement that does not check such features, and there should be uninterpretable features that are not checked by movement/copy formation. The former assumption helps at least in analyzing non-terminal movement steps in cyclic extractions, see also Chomsky (1998) for an approach in which the strict connection between the triggering of movement and feature checking is given up. 3 We confine our attention here to candidates that are wellformed with respect to conditions such as subjacency, in which no superfluous movement has taken place, in which every necessary movement step has been carried out, etc., i.e. we confine our attention to the effects of deletion on chains that are fully grammatical in every other respect. 4 Given that PRONECON never seems to outrank RECOV, it is more adequate to collapse the two principles into a single one that just bans the realization of phonetic material that is predictable from the syntactic environment. This principle would also be more in the spirit of Pesetsky (1998a). 5 That there will be phonetic material in one copy only follows from (7) only if we make a further assumption: the phonetic material must not be scattered over the various copies in the chain. See below for an elaboration of this point. 6 PRONECON is an obvious extension of Pesetsky's (1998a:344) TEL-principle that penalizes the use of a phonetic matrix for function words. 7 8 See Pesetsky (1998a:342) for a slightly different version of this principle. In a considerable number of languages (see Fanselow 2000 for an overview), a further option exists which is also discussed under the heading 'partial wh-movement': (i) was denkst du wie die Rosen riechen what think you how the roses smell "how do you think the roses smell?" Wie seems to have matrix scope in (i), yet it is moved to the specifier position of the complement clause only. The apparent scope position of the wh-phrase is filled by a different wh-element (was "what"). Malay and Bahasa Indonesia lack this element in the scope position, at least overtly. There are various proposals for the proper analysis of (i) (see e.g. the contributions to Lutz, Müller & v. Stechow 2000). If Fanselow & Mahajan (1996, 2000) are correct, both was and wie are in fact in their scope positions, i.e., the partial nature of wh-movement in (i) is only apparent. 9 Alternatively, the lack of island effects for argumental wh-phrases in situ can also be explained by assuming that they are subject to featural attraction in the sense of Chomsky (1995), Pesetsky (1998b), if featural attraction (or "Agreement", see Chomsky 1998) is not constrained by subjacency, as Pesetsky (1998b) argues. In such an account, there are two types of "covert" movement: standard phrasal copying followed by the deletion of the phonetic material in the landing site (=the topic of the present paper), and featural attraction An EPP-feature in the sense of Chomsky (1998), or a D-feature, as argued by Fanselow & Mahajan (2000). More precisely, it is island effects related to subjacency that one does not expect. Intervention effects as discussed by Beck (1996) or Pesetsky (1998b) are not excluded in this way. Furthermore, the binding of wh-phrases in situ may be subject to binding conditions, as Ouhalla (1996) points out. That the binding option is restricted to wh-phrases in situ is a consequence of the fact that wh-phrases can appear in the specifier position of a CP only if they have been attracted to that position. In other words, if the Comp-position to which the wh-phrase is semantically linked has an attracting feature, this feature must be checked by copying, so that island effects become automatically relevant. 12 The feature attracting wh-phrases to Spec,CP can be optionally absent in languages with "overt wh-movement" (as seems to be the case for French matrix Comps) and in languages without it (Chinese). German is a language with "overt" wh-movement in which the wh-attracting feature of Comp cannot be absent (if we ignore echo questions). Hindi, on the other hand, seems to be a wh-in-situ language in which wh-in-situ phrases must not appear in islands (see Mahajan 1990). In the system presupposed above, this array of facts suggests that the attracting feature of Comp is again always present (wh-arguments cannot be simply bound by an operator, just as in German). Therefore, there seem to be languages with (German) and without (Hindi) overtly visible wh-movement which require the wh-attracting feature of 27
28 Gisbert Fanselow & Damir Ćavar Comps to be present obligatorily. The only option that appears to be unrealized is a language in which the attracting feature of Comp is obligatorily absent (in such a language, adjunct questions could not arise at all). 13 Movement that is string-vacuous in the strictest possible sense is thus not penalized by STAY* if the principle relates to the realization of phonetic matrices. This may be a welcome result for various kinds of "evacuation" operations necessary in the context of remnant movement that are not feature driven. See e.g. Müller (1999) for a DISCUSSION. 14 Violations of STAY* must be assumed to not be cumulative, because otherwise wh-phrases would then move to the lowest SpecC position only. 15 That island effects can be captured straightforward in our system has been shown above the pertinent argument Cole and Hermon bring forward in this respect applies to a particular formulation of what they call "Multiple Spellout" theories only, but not to the system we develop here. 16 More precisely, the GEN component produces candidates in which all copies in a chain retain their phonetic matrices, and others in which all but one have lost it (and still many further candidates), and only the latter have a chance of winning the competition because they violate PRONECON as little as is possible in the light of RECOV. We will continue to use (slightly misleading) derivational formulations or optimization for pronunciation, however, because they are simpler. 17 More precisely, we assume that 'deletion' means that the phonetic information of (part of) a copy is marked as unpronounced. In principle, such a phonetic matrix could be promoted to the status of being pronounced in later derivational steps, but such syntactic objects are not likely to be optimal candidates. To see why, consider the following derivation. Suppose that β = [α 1...α 2...] has been formed, with β being cyclic, so that the optimal pronunciation must be selected. Suppose β* = [α 1...α 2...] is optimal. Then β* may be subjected to further operations, but note that only the specifier (and perhaps the head) of β* is accessible to such further copying or deletion steps (Chomsky 1998), that is in particular, the phonetic matrix of α 2 can no longer be affected - it cannot be marked as unpronounced in further derivational steps. Thus, when the object γ = [α 0... [ β* α 1...α 2...]...] is subjected to EVAL, α 0 must lose its phonetic matrix as long as chains are realized in one position only (=the standard case), because the decision that α 2 is pronounced can no longer be retracted. In other words, α 0 has a chance for phonetic realization only if other principles that RECOV crucially override PRONECON. See below for such a case. 18 Wilder (1994) shows that deletion operations operating from left (retained) to right (deleted) are subject to intervention effects from certain heads. Meng-distribution fits this proposal, because the phonetic presence of meng excludes the transformation of...xp... meng...xp... to...xp... meng...xp..., so that we are left with...xp... meng...xp... as the only option At least this holds for questions asking for lists. "!" marks that the structure is acceptable is some versions of German only. Long movement is e.g. certainly acceptable in all dialects spoken to the South of the Main river (the so-called whitesausage equator), but also in Eastfalian and in some dialects in Schleswig-Holstein. 22 According to Reis (1995), however, this construction is parenthetical in nature and therefore does not involve movement at all. 23 It is hard to find speakers from the Ruhr area (Dortmund, Bochum) who accept and use these constructions, but they seem fine to at least some speakers from the Lower Rhine area (around Wesel, Kleve) from Eastern Westfalia (around Bielefeld) and in Bavarian Suebia (Kempten). Thanks to Susanne Anschütz, Daniel Büring, Sascha Felix, Peter Gebert, Barbara Lenz, Sandra Muckel, Barbara Stiebels and in particular to Susann Fischer for their help in this respect. 24 Recall that in German verb-second complement clauses like (i), the embedded Comp position is filled by the verb attracted to that position. (i) ich denke er hat sie geküsst I think he has her kissed "I think he has kissed her" 25 Pesetsky (1998a) later revises this first version of the LEC. The reformulation is not relevant for the type of data we discuss in the main body of the article, though. 26 Given the cyclic nature of optimization, there seems little hope for an attempt to guarantee that the tie between CIS and PARSESCOPE is not resolved differently in different movement steps. When a phrase is split, it will not be put together in later derivational steps for obvious reasons, but on the other hand, one expects a phrase to be able to split at later derivational steps as well. (ii), modeled after corresponding Dutch data in Barbiers (1999) suggests that at least for some versions of German and some wh-phrases, the expectation is borne out. In addition, (ii) corroborates the view that long movement passes through an AGR-O position, as Barbiers observes. (i) was für Frauen hast denn du gedacht dass er einladen will what for women have ptc you thought that he invite wants (ii) was hast denn du für Frauen gedacht dass er einladen will 28
29 Economy of Pronunciation (iii)?was hast denn du gedacht dass er für Frauen einladen will We are obliged to Susann Fischer for helping us getting informants' judgments here. We owe this observation to Josef Bayer, p. c. This has been proposed recently by Hinterhölzl (1999) Caroline Féry (p.c.) suggests the following alternative explanation for split XPs that avoids the assumption of specific focus positions: XPs are split because of their suboptimal phonological properties. Notice that two prominent accents should not be adjacent in a string. If a noun phrase has two independent foci, it must realize two prominent accents. The splitting of the phrase avoids a situation in which these two accents would be too close to each other. 29
Movement and Binding
Movement and Binding Gereon Müller Institut für Linguistik Universität Leipzig SoSe 2008 www.uni-leipzig.de/ muellerg Gereon Müller (Institut für Linguistik) Constraints in Syntax 4 SoSe 2008 1 / 35 Principles
Semantics and Generative Grammar. Quantificational DPs, Part 3: Covert Movement vs. Type Shifting 1
Quantificational DPs, Part 3: Covert Movement vs. Type Shifting 1 1. Introduction Thus far, we ve considered two competing analyses of sentences like those in (1). (1) Sentences Where a Quantificational
Successive Cyclicity, Long-Distance Superiority, and Local Optimization
Successive Cyclicity, Long-Distance Superiority, and Local Optimization Fabian Heck and Gereon Müller University of Stuttgart and University of Tübingen 1. Introduction It is not a priori clear how intermediate
Constraints in Phrase Structure Grammar
Constraints in Phrase Structure Grammar Phrase Structure Grammar no movement, no transformations, context-free rules X/Y = X is a category which dominates a missing category Y Let G be the set of basic
Superiority: Syntax or Semantics? Düsseldorf Jul02. Jill devilliers, Tom Roeper, Jürgen Weissenborn Smith,Umass,Potsdam
Superiority: Syntax or Semantics? Düsseldorf Jul02 Jill devilliers, Tom Roeper, Jürgen Weissenborn Smith,Umass,Potsdam Introduction I. Question: When does a child know the grammaticality difference between
Derivational Optimization of Wh-Movement
Derivational Optimization of Wh-Movement Fabian Heck & Gereon Müller Universität Leipzig 1. Introduction An idea that was originally proposed by Bresnan (1971; 1972) has been resurrected in recent syntactic
Structure of Clauses. March 9, 2004
Structure of Clauses March 9, 2004 Preview Comments on HW 6 Schedule review session Finite and non-finite clauses Constituent structure of clauses Structure of Main Clauses Discuss HW #7 Course Evals Comments
The compositional semantics of same
The compositional semantics of same Mike Solomon Amherst College Abstract Barker (2007) proposes the first strictly compositional semantic analysis of internal same. I show that Barker s analysis fails
Melanie Klepp. 1. Introduction 1.1. The was-w-construction
Much ado about was: Why German directly depends on indirect dependency This paper sets out to challenge the widely held assumption that the Direct Dependency Approach (DDA) is the most suitable analysis
Lecture 1: OT An Introduction
Lecture 1: OT An Introduction 1 Generative Linguistics and OT Starting point: Generative Linguistics Sources: Archangeli 1997; Kager 1999, Section 1; Prince & Smolensky 1993; Barbosa et al. 1998, intro.
bound Pronouns
Bound and referential pronouns *with thanks to Birgit Bärnreuther, Christina Bergmann, Dominique Goltz, Stefan Hinterwimmer, MaikeKleemeyer, Peter König, Florian Krause, Marlene Meyer Peter Bosch Institute
Comprendium Translator System Overview
Comprendium System Overview May 2004 Table of Contents 1. INTRODUCTION...3 2. WHAT IS MACHINE TRANSLATION?...3 3. THE COMPRENDIUM MACHINE TRANSLATION TECHNOLOGY...4 3.1 THE BEST MT TECHNOLOGY IN THE MARKET...4
Linear Compression as a Trigger for Movement 1
Linear Compression as a Trigger for Movement 1 Andrea Carlo Moro 1. Beyond Explanatory Adequacy : How the World Shapes Grammar A new challenge has been addressed in generative grammar in a recent paper
Right Node Raising and the LCA
1 Right Node Raising and the LCA CHRIS WILDER 1 Constituent sharing in coordination In (1), a typical right node raising (RNR) sentence, the object the book is a constituent shared by the verbs of both
Double Genitives in English
Karlos Arregui-Urbina Department Linguistics and Philosophy MIT 1. Introduction Double Genitives in English MIT, 29 January 1998 Double genitives are postnominal genitive phrases which are marked with
Extended Projections of Adjectives and Comparative Deletion
Julia Bacskai-Atkari 25th Scandinavian Conference University of Potsdam (SFB-632) in Linguistics (SCL-25) [email protected] Reykjavík, 13 15 May 2013 0. Introduction Extended Projections
Is there repair by ellipsis?
Is there repair by ellipsis? Craig Sailor University of Groningen [email protected] Carson T. Schütze UCLA [email protected] Draft: December, 2014 Written for The book of syntactic questions 100 ideas
Noam Chomsky: Aspects of the Theory of Syntax notes
Noam Chomsky: Aspects of the Theory of Syntax notes Julia Krysztofiak May 16, 2006 1 Methodological preliminaries 1.1 Generative grammars as theories of linguistic competence The study is concerned with
ACKNOWLEDGEMENTS. At the completion of this study there are many people that I need to thank. Foremost of
ACKNOWLEDGEMENTS At the completion of this study there are many people that I need to thank. Foremost of these are John McCarthy. He has been a wonderful mentor and advisor. I also owe much to the other
Appendix to Chapter 3 Clitics
Appendix to Chapter 3 Clitics 1 Clitics and the EPP The analysis of LOC as a clitic has two advantages: it makes it natural to assume that LOC bears a D-feature (clitics are Ds), and it provides an independent
The finite verb and the clause: IP
Introduction to General Linguistics WS12/13 page 1 Syntax 6 The finite verb and the clause: Course teacher: Sam Featherston Important things you will learn in this section: The head of the clause The positions
[Refer Slide Time: 05:10]
Principles of Programming Languages Prof: S. Arun Kumar Department of Computer Science and Engineering Indian Institute of Technology Delhi Lecture no 7 Lecture Title: Syntactic Classes Welcome to lecture
Statistical Machine Translation
Statistical Machine Translation Some of the content of this lecture is taken from previous lectures and presentations given by Philipp Koehn and Andy Way. Dr. Jennifer Foster National Centre for Language
Non-nominal Which-Relatives
Non-nominal Which-Relatives Doug Arnold, Robert D. Borsley University of Essex The properties of non-restrictive relatives All non-restrictive relative clauses include a wh-word. There are no that or zero
Constituency. The basic units of sentence structure
Constituency The basic units of sentence structure Meaning of a sentence is more than the sum of its words. Meaning of a sentence is more than the sum of its words. a. The puppy hit the rock Meaning of
Varieties of specification and underspecification: A view from semantics
Varieties of specification and underspecification: A view from semantics Torgrim Solstad D1/B4 SFB meeting on long-term goals June 29th, 2009 The technique of underspecification I Presupposed: in semantics,
Optimality-Theoretic Syntax Gereon Müller (Universität Leipzig) Comparing Frameworks, Utrecht Institute of Linguistics (OTS) September 24-26, 2009
Optimality-Theoretic Syntax Gereon Müller (Universität Leipzig) Comparing Frameworks, Utrecht Institute of Linguistics (OTS) September 24-26, 2009 Contents 1. Model of Grammar 1 2. Evidence for OT Analyses
IP PATTERNS OF MOVEMENTS IN VSO TYPOLOGY: THE CASE OF ARABIC
The Buckingham Journal of Language and Linguistics 2013 Volume 6 pp 15-25 ABSTRACT IP PATTERNS OF MOVEMENTS IN VSO TYPOLOGY: THE CASE OF ARABIC C. Belkacemi Manchester Metropolitan University The aim of
Syntactic Theory. Background and Transformational Grammar. Dr. Dan Flickinger & PD Dr. Valia Kordoni
Syntactic Theory Background and Transformational Grammar Dr. Dan Flickinger & PD Dr. Valia Kordoni Department of Computational Linguistics Saarland University October 28, 2011 Early work on grammar There
19. Morphosyntax in L2A
Spring 2012, April 5 Missing morphology Variability in acquisition Morphology and functional structure Morphosyntax in acquisition In L1A, we observe that kids don t always provide all the morphology that
TD 271 Rev.1 (PLEN/15)
INTERNATIONAL TELECOMMUNICATION UNION STUDY GROUP 15 TELECOMMUNICATION STANDARDIZATION SECTOR STUDY PERIOD 2009-2012 English only Original: English Question(s): 12/15 Geneva, 31 May - 11 June 2010 Source:
SQL INJECTION ATTACKS By Zelinski Radu, Technical University of Moldova
SQL INJECTION ATTACKS By Zelinski Radu, Technical University of Moldova Where someone is building a Web application, often he need to use databases to store information, or to manage user accounts. And
SYNTAX: THE ANALYSIS OF SENTENCE STRUCTURE
SYNTAX: THE ANALYSIS OF SENTENCE STRUCTURE OBJECTIVES the game is to say something new with old words RALPH WALDO EMERSON, Journals (1849) In this chapter, you will learn: how we categorize words how words
The syntactic positions of adverbs and the Second Language Acquisition
September 2010, Volume 7, No.9 (Serial No.81) Sino-US English Teaching, ISSN 1539-8072, USA The syntactic positions of adverbs and the Second Language Acquisition ZHANG Zi-hong (Department of Foreign Language
Inflation. Chapter 8. 8.1 Money Supply and Demand
Chapter 8 Inflation This chapter examines the causes and consequences of inflation. Sections 8.1 and 8.2 relate inflation to money supply and demand. Although the presentation differs somewhat from that
Features, θ-roles, and Free Constituent Order
1 Features, θ-roles, and Free Constituent Order Gisbert Fanselow This paper pursues two different but related goals. On the one hand, it shows that free constituent order (at least the type one finds in
proceedings 2003/9/19 17:30 page 29 #33
proceedings 2003/9/19 17:30 page 29 #33 3 Surprising Specifiers and Cyclic Spellout Gisbert Fanselow University of Potsdam Contents 1 Introduction....................................... 29 2 German main
On directionality of phrase structure building
This is an extended version of a paper that should appear on Journal of Psycholinguistic Research (2014) October 2013 On directionality of phrase structure building Cristiano Chesi (NETS, IUSS Pavia) Abstract
CHAPTER II THE LIMIT OF A SEQUENCE OF NUMBERS DEFINITION OF THE NUMBER e.
CHAPTER II THE LIMIT OF A SEQUENCE OF NUMBERS DEFINITION OF THE NUMBER e. This chapter contains the beginnings of the most important, and probably the most subtle, notion in mathematical analysis, i.e.,
Phrase Structure Rules, Tree Rewriting, and other sources of Recursion Structure within the NP
Introduction to Transformational Grammar, LINGUIST 601 September 14, 2006 Phrase Structure Rules, Tree Rewriting, and other sources of Recursion Structure within the 1 Trees (1) a tree for the brown fox
or conventional implicature [1]. If the implication is only pragmatic, explicating logical truth, and, thus, also consequence and inconsistency.
44 ANALYSIS explicating logical truth, and, thus, also consequence and inconsistency. Let C1 and C2 be distinct moral codes formulated in English. Let C1 contain a norm N and C2 its negation. The moral
Quotes from Object-Oriented Software Construction
Quotes from Object-Oriented Software Construction Bertrand Meyer Prentice-Hall, 1988 Preface, p. xiv We study the object-oriented approach as a set of principles, methods and tools which can be instrumental
Overview of the TACITUS Project
Overview of the TACITUS Project Jerry R. Hobbs Artificial Intelligence Center SRI International 1 Aims of the Project The specific aim of the TACITUS project is to develop interpretation processes for
OPRE 6201 : 2. Simplex Method
OPRE 6201 : 2. Simplex Method 1 The Graphical Method: An Example Consider the following linear program: Max 4x 1 +3x 2 Subject to: 2x 1 +3x 2 6 (1) 3x 1 +2x 2 3 (2) 2x 2 5 (3) 2x 1 +x 2 4 (4) x 1, x 2
Syntax: Phrases. 1. The phrase
Syntax: Phrases Sentences can be divided into phrases. A phrase is a group of words forming a unit and united around a head, the most important part of the phrase. The head can be a noun NP, a verb VP,
A (Covert) Long Distance Anaphor in English
A (Covert) Long Distance Anaphor in English Christopher Kennedy and Jeffrey Lidz Northwestern University 1. Introduction The empirical focus of this paper is the distribution of strict and sloppy interpretations
Pricing in a Competitive Market with a Common Network Resource
Pricing in a Competitive Market with a Common Network Resource Daniel McFadden Department of Economics, University of California, Berkeley April 8, 2002 I. This note is concerned with the economics of
Data Analysis 1. SET08104 Database Systems. Copyright @ Napier University
Data Analysis 1 SET08104 Database Systems Copyright @ Napier University Entity Relationship Modelling Overview Database Analysis Life Cycle Components of an Entity Relationship Diagram What is a relationship?
What s in a Lexicon. The Lexicon. Lexicon vs. Dictionary. What kind of Information should a Lexicon contain?
What s in a Lexicon What kind of Information should a Lexicon contain? The Lexicon Miriam Butt November 2002 Semantic: information about lexical meaning and relations (thematic roles, selectional restrictions,
COMPARATIVES WITHOUT DEGREES: A NEW APPROACH. FRIEDERIKE MOLTMANN IHPST, Paris [email protected]
COMPARATIVES WITHOUT DEGREES: A NEW APPROACH FRIEDERIKE MOLTMANN IHPST, Paris [email protected] It has become common to analyse comparatives by using degrees, so that John is happier than Mary would
Discussion of Capital Injection, Monetary Policy, and Financial Accelerators
Discussion of Capital Injection, Monetary Policy, and Financial Accelerators Karl Walentin Sveriges Riksbank 1. Background This paper is part of the large literature that takes as its starting point the
16.1 MAPREDUCE. For personal use only, not for distribution. 333
For personal use only, not for distribution. 333 16.1 MAPREDUCE Initially designed by the Google labs and used internally by Google, the MAPREDUCE distributed programming model is now promoted by several
Complex Predications in Argument Structure Alternations
Complex Predications in Argument Structure Alternations Stefan Engelberg (Institut für Deutsche Sprache & University of Mannheim) Stefan Engelberg (IDS Mannheim), Universitatea din Bucureşti, November
Electronics for Analog Signal Processing - II Prof. K. Radhakrishna Rao Department of Electrical Engineering Indian Institute of Technology Madras
Electronics for Analog Signal Processing - II Prof. K. Radhakrishna Rao Department of Electrical Engineering Indian Institute of Technology Madras Lecture - 18 Wideband (Video) Amplifiers In the last class,
CHAPTER 7 GENERAL PROOF SYSTEMS
CHAPTER 7 GENERAL PROOF SYSTEMS 1 Introduction Proof systems are built to prove statements. They can be thought as an inference machine with special statements, called provable statements, or sometimes
6.042/18.062J Mathematics for Computer Science. Expected Value I
6.42/8.62J Mathematics for Computer Science Srini Devadas and Eric Lehman May 3, 25 Lecture otes Expected Value I The expectation or expected value of a random variable is a single number that tells you
Interpretation of Test Scores for the ACCUPLACER Tests
Interpretation of Test Scores for the ACCUPLACER Tests ACCUPLACER is a trademark owned by the College Entrance Examination Board. Visit The College Board on the Web at: www.collegeboard.com/accuplacer
Lectures, 2 ECONOMIES OF SCALE
Lectures, 2 ECONOMIES OF SCALE I. Alternatives to Comparative Advantage Economies of Scale The fact that the largest share of world trade consists of the exchange of similar (manufactured) goods between
7 Conclusions and suggestions for further research
7 Conclusions and suggestions for further research This research has devised an approach to analyzing system-level coordination from the point of view of product architecture. The analysis was conducted
How the Computer Translates. Svetlana Sokolova President and CEO of PROMT, PhD.
Svetlana Sokolova President and CEO of PROMT, PhD. How the Computer Translates Machine translation is a special field of computer application where almost everyone believes that he/she is a specialist.
The Role of Dispute Settlement Procedures in International Trade Agreements: Online Appendix
The Role of Dispute Settlement Procedures in International Trade Agreements: Online Appendix Giovanni Maggi Yale University, NBER and CEPR Robert W. Staiger Stanford University and NBER November 2010 1.
Last time we had arrived at the following provisional interpretation of Aquinas second way:
Aquinas Third Way Last time we had arrived at the following provisional interpretation of Aquinas second way: 1. 2. 3. 4. At least one thing has an efficient cause. Every causal chain must either be circular,
Basic Set Theory. 1. Motivation. Fido Sue. Fred Aristotle Bob. LX 502 - Semantics I September 11, 2008
Basic Set Theory LX 502 - Semantics I September 11, 2008 1. Motivation When you start reading these notes, the first thing you should be asking yourselves is What is Set Theory and why is it relevant?
Modeling Guidelines Manual
Modeling Guidelines Manual [Insert company name here] July 2014 Author: John Doe [email protected] Page 1 of 22 Table of Contents 1. Introduction... 3 2. Business Process Management (BPM)... 4 2.1.
Building a Question Classifier for a TREC-Style Question Answering System
Building a Question Classifier for a TREC-Style Question Answering System Richard May & Ari Steinberg Topic: Question Classification We define Question Classification (QC) here to be the task that, given
Data quality in Accounting Information Systems
Data quality in Accounting Information Systems Comparing Several Data Mining Techniques Erjon Zoto Department of Statistics and Applied Informatics Faculty of Economy, University of Tirana Tirana, Albania
Methodological Issues for Interdisciplinary Research
J. T. M. Miller, Department of Philosophy, University of Durham 1 Methodological Issues for Interdisciplinary Research Much of the apparent difficulty of interdisciplinary research stems from the nature
Pronouns: A case of production-before-comprehension
Faculty of Arts University of Groningen Pronouns: A case of production-before-comprehension A study on the the comprehension and production of pronouns and reflexives in Dutch children Petra Hendriks Jennifer
Free/Libre and Open Source Software: Survey and Study FLOSS
Free/Libre and Open Source Software: Survey and Study FLOSS Deliverable D18: FINAL REPORT Part 0: Table of Contents and Executive Summary International Institute of Infonomics University of Maastricht,
Late Merger of Degree Clauses
Late Merger of Degree Clauses Rajesh Bhatt and Roumyana Pancheva In this article, we propose that degree heads and degree clauses form a constituent not at the point where the degree head is merged, but
CAS LX 500 A1 Language Acquisition
CAS LX 500 A1 Language Acquisition Week 4a. Root infinitives, null subjects and the UCC Root infinitives vs. time Here are those Danish graphs again. Ooo. Consistent. Syntax at age two Root infinitives
Extraposition, the Right Roof Constraint, Result Clauses, Relative Clause Extraposition, and PP Extraposition
Extraposition, the Right Roof Constraint, Result Clauses, Relative Clause Extraposition, and PP Extraposition Mark R. Baltin revised version to appear in The Syntax Companion New York University First
Development models. 1 Introduction. 2 Analyzing development models. R. Kuiper and E.J. Luit
Development models R. Kuiper and E.J. Luit 1 Introduction We reconsider the classical development models: the Waterfall Model [Bo76], the V-Model [Ro86], the Spiral Model [Bo88], together with the further
Resumption by Buffers: German Relative Clauses
Resumption by Buffers: German Relative Clauses Gereon Müller Abstract In a local derivational (phase-based) approach to syntax, instances of resumption in German (longdistance) relativization constructions
Chapter 5. Phrase-based models. Statistical Machine Translation
Chapter 5 Phrase-based models Statistical Machine Translation Motivation Word-Based Models translate words as atomic units Phrase-Based Models translate phrases as atomic units Advantages: many-to-many
6.852: Distributed Algorithms Fall, 2009. Class 2
.8: Distributed Algorithms Fall, 009 Class Today s plan Leader election in a synchronous ring: Lower bound for comparison-based algorithms. Basic computation in general synchronous networks: Leader election
Abstraction in Computer Science & Software Engineering: A Pedagogical Perspective
Orit Hazzan's Column Abstraction in Computer Science & Software Engineering: A Pedagogical Perspective This column is coauthored with Jeff Kramer, Department of Computing, Imperial College, London ABSTRACT
Mathematical Induction
Mathematical Induction In logic, we often want to prove that every member of an infinite set has some feature. E.g., we would like to show: N 1 : is a number 1 : has the feature Φ ( x)(n 1 x! 1 x) How
Unit 2.1. Data Analysis 1 - V2.0 1. Data Analysis 1. Dr Gordon Russell, Copyright @ Napier University
Data Analysis 1 Unit 2.1 Data Analysis 1 - V2.0 1 Entity Relationship Modelling Overview Database Analysis Life Cycle Components of an Entity Relationship Diagram What is a relationship? Entities, attributes,
Settlers of Catan Analysis
Math 401 Settlers of Catan Analysis Peter Keep. 12/9/2010 Settlers of Catan Analysis Settlers of Catan is a board game developed by Klaus Teber, a German board game designer. Settlers is described by Mayfair
ASSESSMENT CENTER FOR IDENTIFYING POTENTIAL PROJECT MANAGERS: A CHANCE FOR SYSTEMATIC HUMAN RESOURCE DEVELOPMENT
ASSESSMENT CENTER FOR IDENTIFYING POTENTIAL PROJECT MANAGERS: A CHANCE FOR SYSTEMATIC HUMAN RESOURCE DEVELOPMENT Dipl. Psych. Ingo Heyn, ALLIANZ LEBENSVERSICHERUNGS-AG, Germany, 1999 Paper for the 6th
SUPPLEMENTARY READING: A NOTE ON AGNATION
1 SUPPLEMENTARY READING: A NOTE ON AGNATION Introduction The term agnate, together with its derivative agnation, was introduced into linguistics by the American structuralist H.A. Gleason, Jr. (Gleason
This accuracy is open to question for the following reasons:
AN EXPERIMENT TO COMPARE ALTERNATIVE VERIFICATION PROCEDURES FOR MORTALITY MEDICAL CODING Kenneth W. Harris and Dwight K. French National Center for Health Statistics I. Background One of the most difficult
Gerry Hobbs, Department of Statistics, West Virginia University
Decision Trees as a Predictive Modeling Method Gerry Hobbs, Department of Statistics, West Virginia University Abstract Predictive modeling has become an important area of interest in tasks such as credit
Information, Entropy, and Coding
Chapter 8 Information, Entropy, and Coding 8. The Need for Data Compression To motivate the material in this chapter, we first consider various data sources and some estimates for the amount of data associated
Points of Interference in Learning English as a Second Language
Points of Interference in Learning English as a Second Language Tone Spanish: In both English and Spanish there are four tone levels, but Spanish speaker use only the three lower pitch tones, except when
WHAT ARE MATHEMATICAL PROOFS AND WHY THEY ARE IMPORTANT?
WHAT ARE MATHEMATICAL PROOFS AND WHY THEY ARE IMPORTANT? introduction Many students seem to have trouble with the notion of a mathematical proof. People that come to a course like Math 216, who certainly
Word order in Lexical-Functional Grammar Topics M. Kaplan and Mary Dalrymple Ronald Xerox PARC August 1995 Kaplan and Dalrymple, ESSLLI 95, Barcelona 1 phrase structure rules work well Standard congurational
DATA QUALITY DATA BASE QUALITY INFORMATION SYSTEM QUALITY
DATA QUALITY DATA BASE QUALITY INFORMATION SYSTEM QUALITY The content of those documents are the exclusive property of REVER. The aim of those documents is to provide information and should, in no case,
