my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for...

25
1 A multidominance approach to appositional constructions Herman Heringa (University of Groningen) * Abstract This paper presents a multidominance analysis for appositional constructions. It is argued that appositions are related to their anchors in two ways. First, these two elements are coordinated. Second, the apposition functions as the predicate of the anchor. The predication relation is argued to have the structure of a full copular CP, parenthetically related to the anchor. In this secondary clause, the anchor is represented by a (silent) pronoun, while the apposition functions simultaneously as the predicate of the copular CP and as the second conjunct to the anchor. This analysis is used to explain the case marking patterns on appositions in several languages. It is proposed that both the coordinative structure and the predicative structure assign a case feature to the apposition. Which of these features is spelled out is decided by language-specific rules in PF, also depending on the semantic type of the apposition. 1. Introduction Appositional constructions, exemplified in (1), have been the subject of a long and interesting debate, going back at least to Hockett (1955). (1) a. John, my neighbour, is a nice guy. b. The winner, Mary, was very glad. In (1a), John is called the anchor, and my neighbour the apposition. The anchor is always the first element and the apposition is therefore a postmodifier of the anchor. I take the modification in this construction to be non-restrictive, which means that so-called close appositions, such as the phrases in (2), are excluded from the discussion, since they will need a different analysis. (2) a. his brother John b. the noun house Being non-restrictive, appositions get a separate intonation contour from the matrix clause in which they appear. They get a lower tone and after the apposition the main contour continues. In written language, this is usually represented by comma's or other punctuation marks, such as dashes. The debate on non-restrictive appositional constructions is about the relation between its two parts, the anchor and the apposition. Since syntacticians assume only two types of relations between constituents, coordination and subordination, the question is which of the two is used here. Various linguists have argued for either of the two options, while others argued for a third type of relation, dedicated to exactly this type of construction. In this paper, I will explore a fourth possibility, namely that the relation * I want to thank Aysa Arylova, Alexander Grosu, Veronika Hegedüs, Gunnar Hrafn Hrafnbjargarson, Anikó Lipták, Evguenia Markovskaya and Radek Šimík for providing me with data from their native languages. Thanks go also to Caroline Heycock, Mark de Vries and the audiences of BCGL3, Brussels, Tabudag 2008, Groningen, the OC colloquium in Nijmegen, and the SSRG meeting in Edinburgh for their useful comments and suggestions. This paper is based on research financially supported by the Netherlands Organisation for Scientific Research (NWO).

Transcript of my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for...

Page 1: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

1

A multidominance approach to appositional constructions

Herman Heringa (University of Groningen)∗ Abstract This paper presents a multidominance analysis for appositional constructions. It is argued that appositions are related to their anchors in two ways. First, these two elements are coordinated. Second, the apposition functions as the predicate of the anchor. The predication relation is argued to have the structure of a full copular CP, parenthetically related to the anchor. In this secondary clause, the anchor is represented by a (silent) pronoun, while the apposition functions simultaneously as the predicate of the copular CP and as the second conjunct to the anchor. This analysis is used to explain the case marking patterns on appositions in several languages. It is proposed that both the coordinative structure and the predicative structure assign a case feature to the apposition. Which of these features is spelled out is decided by language-specific rules in PF, also depending on the semantic type of the apposition. 1. Introduction Appositional constructions, exemplified in (1), have been the subject of a long and interesting debate, going back at least to Hockett (1955). (1) a. John, my neighbour, is a nice guy. b. The winner, Mary, was very glad. In (1a), John is called the anchor, and my neighbour the apposition. The anchor is always the first element and the apposition is therefore a postmodifier of the anchor. I take the modification in this construction to be non-restrictive, which means that so-called close appositions, such as the phrases in (2), are excluded from the discussion, since they will need a different analysis. (2) a. his brother John b. the noun house Being non-restrictive, appositions get a separate intonation contour from the matrix clause in which they appear. They get a lower tone and after the apposition the main contour continues. In written language, this is usually represented by comma's or other punctuation marks, such as dashes. The debate on non-restrictive appositional constructions is about the relation between its two parts, the anchor and the apposition. Since syntacticians assume only two types of relations between constituents, coordination and subordination, the question is which of the two is used here. Various linguists have argued for either of the two options, while others argued for a third type of relation, dedicated to exactly this type of construction. In this paper, I will explore a fourth possibility, namely that the relation

∗ I want to thank Aysa Arylova, Alexander Grosu, Veronika Hegedüs, Gunnar Hrafn Hrafnbjargarson, Anikó Lipták, Evguenia Markovskaya and Radek Šimík for providing me with data from their native languages. Thanks go also to Caroline Heycock, Mark de Vries and the audiences of BCGL3, Brussels, Tabudag 2008, Groningen, the OC colloquium in Nijmegen, and the SSRG meeting in Edinburgh for their useful comments and suggestions. This paper is based on research financially supported by the Netherlands Organisation for Scientific Research (NWO).

Page 2: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

2

between the anchor and the apposition involves both coordination and subordination at the same time. In particular, I will argue that these constructions make use of a special type of coordination, specifying coordination (cf. De Vries 2006), and that the relation of subordination is expressed by predication, where the anchor is the subject and the apposition the predicate (cf. Doron 1994). This results in a secondary message, saying that the anchor is the apposition. The utterances in (1) implicitly contain the messages in (3). (3) a. John is my neighbour. b. The winner is Mary. In order to combine the relations of coordination and predication in one structure, I will use the idea of multidominance, which makes it possible that one element, in this case the apposition, is dominated by more than one node. One of the dominating nodes will then be in a relation of coordination with the anchor, while another is in a predication relation with that same anchor. Finally, I will present additional evidence for this approach on the basis of case marking patterns from various languages. 2. Coordination This section discusses the idea that appositional constructions involve a relation of coordination between the anchor and the apposition. Parallels between appositional and coordinate constructions have been noted by several authors, such as Kraak & Klooster (1968), Quirk et al. (1985), and De Vries (2006, 2009). A first argument for analysing appositional constructions as coordination comes from apposition markers. These markers can be used to make the relation between the anchor and the apposition more explicit. For now, it is important to note that the coordinators and and or can be used as apposition markers. This implies that the main types of coordination, conjunction and disjunction are represented. This is not only true for English and other germanic languages, but also for a slavic language like Czech. Examples (4a/b) come from Quirk et al. (1985:1311/12). The examples in (5) are Radek Šimík's (pc). (4) a. The United States of America, or America for short... [English] b. You could cut the atmosphere with a knife, and a blunt knife at that. c. Piet gaat een auto, en wel een Porsche, kopen. [Dutch] Piet goes a car and indeed a Porsche buy ‘Piet will buy a car, namely a Porsche.’ d. Marie leest een boek over lexicale semantiek, oftewel woordbetekenis. Marie reads a book on lexical semantics or.indeed word.sense (5) a. Spojené Státu Americké, neboli Amerika ... [Czech] United States Amer.-ADJ or America b. Vid�l jsem n�co krásného, a to zlat d�m. saw AUX-1SG-PAST something beautiful and it golden house ‘I saw something beautiful, namely a golden house.’

Page 3: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

3

Another indication for the idea that appositional constructions involve coordination is the fact that the apposition marker forms a constituent with the apposition, just like the coordinator forms a constituent with the second conjunct. This is not only clear from the intonation pattern, but also, for example, from extraposition. If the apposition or the second conjunct is extraposed, the marker has to be extraposed as well, which shows that it belongs to the same constituent; see (6). (6) a. Bill saw a nice butterfly yesterday, namely a red admiral. b. * Bill saw a nice butterfly namely yesterday, a red admiral. c. Bill saw John yesterday, and Mary. d. * Bill saw John and yesterday, Mary. For more arguments to treat appositional constructions in terms of coordination, I refer to Heringa (2007) and the references there. For now, I conclude that appositional constructions can be analysed as a special type of coordination. This means that there are (at least) four semantic types of coordination, illustrated below. (7) a. the Netherlands and Belgium (additive) b. the Netherlands or Belgium (disjunctive) c. not the Netherlands, but Belgium (adversative) d. the Netherlands, or Holland (specifying) The main difference between the traditional types of coordination and the type that is found in appositional constructions is the relation between the referents of the conjuncts. Whereas conjuncts usually refer to different entities, an apposition and its anchor may have the same referent. The apposition, like an appositive relative clause, says something more about the anchor. We will call this specifying coordination. Though the types of coordination differ semantically, their syntax is the same. Following Munn (1987) and Johannessen (1998), I assume a coordination phrase, CoP, with the coordinator as the head and the conjuncts as the specifier and the complement of this head. (8) This section is about the coordination phrase, or the CoP.

To distinguish specifying coordination from the other types, the colon is used as a symbol representing a relation of specification. The colon phrase was introduced by Koster (2000), who argues for an analysis of extraposition in terms of specifying coordination. In the case of apposition, the apposition marker (here or) is the head of the colon phrase. In the next section, I will argue that the relation between the anchor and the apposition not only involves coordination, but also predication.

Anchor :'

Apposition : (App. Marker)

:P

DP the CoP

: or

:P

:' DP the coordination

phrase

Conjunct1 Co'

Conjunct2 Co(ordinator)

CoP

Page 4: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

4

3. Predication Section 1 introduced the idea that appostitional constructions involve a secondary message, stating something like '[anchor] is [apposition]'. Several authors (Berckmans 1994, Dever 2001, Potts 2003, ao) have argued in favor of this idea. The most extensive discussion comes from Potts (2003, 2007). He argues that this message is really distinct from the main proposition by showing that the appositive content is always speaker-oriented. He notes a difference between the two sentences in (9) (Potts 2007: 3/4). (9) a. Sheila says that Chuck is fit to watch the kids and that Chuck is a confirmed

psychopath. b. Sheila says that Chuck, a confirmed psychopath, is fit to watch the kids. Whereas in (9a) the message that Chuck is a confirmed psychopath is included in the report of what Sheila said, this is not the case for the appositional construction in (9b). In the latter construction, there is a secondary message that comes from the speaker, not from the person quoted in the indirect quotation. This is confirmed by the fact that the appositive content cannot be denied. It is impossible to continue (9b) by adding but Chuck is not a confirmed psychopath. Potts argues that a secondary message is a conventional implicature that can be construed semantically. I will argue, however, that a secondary message has to be present in syntax as well, having the structure of a complete copular clause. A first suggestion in that direction comes from the parallel between the semantic classes of apposition (cf. Quirk et al. 1985; Heringa & De Vries 2008) and the types of predication in copular clauses (see for example Partee 1998). The main classes of appositions are identification and attribution. An identificational apposition gives an alternative description for the anchor (10a), whereas an attributional apposition expresses a class to which the anchor belongs (10b). (10) a. My only brother, Pieter, is a member of the student council. b. His girl friend, a modest person, laughs about that. These two types clearly correspond to the main types of copular clauses, specificational and predicational. Predicational sentences add a property to the subject, whereas specificational sentences give a value to a variable introduced by the subject. If the subject of a copular clause would be used as a prompt on a form, a specificational predicate would correspond to a conventional answer (Name: John Johnson), whereas a predicational predicate would correspond to a funny answer (Name: nice). Examples (11a/b) and (12a/b) are from Partee (1998: 363). The c-examples give the implicit messages in (10) and show that the identificational apposition in (10a) relates to a specificational clause, whereas the attributional apposition in (10b) relates to a predicational clause.

Page 5: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

5

Specificational copular clauses: (11) a. The number of planets is nine. b. What I don't like about John is his tie. c. My only brother is Pieter. Predicational copular clauses: (12) a. Helen is a teacher. b. My best friend is tall. c. His girl friend is a modest person. If the second message is really present in syntax as a copular clause, the apposition itself has to be a nominal predicate. Doron (1994) shows that appositions indeed have properties that distinguish them from arguments and put them in the class of predicates. Some of these properties are the possibility of i-within-i constructions (13), which is restricted to predicate positions, and the way appositions are negated (14), which is impossible in argument positions. Examples (13) and (14a) are Doron’s (1994: (5) and (10)). Sentence (14c) comes from Potts (2003: 176). (13) a. Johni, [hisi own worst enemy]i, lost the elections again. b. Johni is [hisi own worst enemy]i. c. * [Hisi own worst enemy]i lost the elections again. (14) a. Orville Wright, not Wilbur, made the first flight at Kitty Hawk. b. * Not Wilbur made the first flight at Kitty Hawk. c. We spoke with Hillary, no amateur climber, about the dangers. d. * We spoke with no amateur climber about the dangers. The arguments seem strong enough to conclude that appositional constructions indeed have to be analysed as a predication relation between the anchor and the apposition. This leads to the question which projections are present in the structure of this relation. Is it just a juxtaposition of a subject and a nominal predicate, as in a small clause, or does the structure include a TP and maybe even a CP? Since TP is related to tense, this question can be answered if we find out whether it is possible to have tense in the secondary message, independent of the tense of the matrix clause. It turns out that this is possible. (15) a. Keith, once a drug addict, now leads a rehabilitation centre. b. I never realised that appositions, currently my subject, could be so interesting. In the examples in (15) the appositions contain modifiers resulting in a distinct tense for the appositive content. In (15a) the matrix is in the present, whereas the appositive message is in the past. In (15b) it is exactly the other way around. Another argument to show that the appositive structure is bigger than a small clause can be made by using (non-temporal) high adverbs, such as obviously and also in (16a/b), from Quirk et al. (1985: 1314) and unfortunately in (16c) from O'Connor (2008: 97). These adverbs modify TP. They can be used in appositions, but definitely not in argument positions.

Page 6: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

6

(16) a. Your brother, obviously an expert on English grammar, is praised in this book. a'. * Obviously an expert on English grammar is praised in this book. b. They elected as chairman Edna Jones, also a Cambridge graduate. b'. * They elected as chairman also a Cambridge graduate. c. Racial profiling, unfortunately a frequent occurrence in American society,

must be stopped. c'. * Unfortunately a frequent occurrence in American society must be stopped. One could object that it is possible to use both temporal and other high adverbs with DPs anyway. Consider, for instance, the examples in (17). The examples in (17c/d) are from Klein (1977); the glosses are mine. (17) a. Charlotte was the eldest daughter of the then king of Naples. b. Last year, the now president elect tried to shoot holes in Clinton's claims. c. De vermoedelijk blonde daders zijn nog niet gevonden. [Dutch] the supposedly blond offenders are yet not found d. De helaas nog te jonge vader mag voorlopig niet trouwen. the unfortunately yet too young father can temporarily not marry Though these examples look very similar to the appositional constructions given in (15) and (16) above, there is an important difference. In the appositional constructions in (15) and (16), the modifiers precede the determiner and are thus outside of the DP. In argument positions, such as in (17), however, the modifiers in question cannot precede the determiner. They have to be positioned within the DP. (18) a. * Charlotte was the eldest daughter of then the king of Naples. b. * Last year, now the president elect tried to shoot holes in Clinton's claims. It is not clear to me why adverbs that usually occur high in the clause can occur within DPs in argument positions and where they are in the structure in this case. An analysis like that from Kayne (1994), where adjectives originate from (reduced) relatives, comes to mind, but this issue is beyond the scope of this paper. In predication, however, it is clear that these adverbs are in a position higher than the tense head. Since I argued that DPs in appositions are predicate DPs, I conclude that the adverbs in appositions are in such a high position in the structure as well. Therefore, these findings provide evidence that the structure of the secondary message projects at least up to TP. O'Connor (2008: 154/156) shows that subordinators can be present in appositions as well, implying that even CP is projected. (19) a. The road, though no longer an officially designated route, has been celebrated

in books ('The Grapes of Wrath'), song ('Get Your Kicks on Route 66') and a TV series ('Route 66'). (Los Angeles Times, 26/12/02, p. B2, col. 1)

Page 7: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

7

b. In allowing that 11th-hour add-on to a $397 billion spending bill, Republican Congressional leaders handed the nation's real estate brokers an important symbolic victory - if perhaps a temporary one - in their concerted push to keep banks at bay. (New York Times online, 'Business ; Brokers to bankers : "No trespassing"', 02/03/03)

Another way of testing the presence of CP in appositions is to check its possibility to code illocutionary force. If the appositive message can have a separate illocutionary force, the structure must include a C head. It turns out that this is indeed possible. Example (20a) is from Corazza (2005: 13). (20) a. Is Jane, the best doctor in town, already married? b. Please, say hello to Jane’s friend, Mike, when you visit her. Clearly, if a matrix sentence is interrogative or imperative, the apposition is not included in the question or the request. The secondary message in that case is usually assertive. Potts (2003) notes that appositions and appositive relatives have even more freedom than coordinated clauses in this respect. He gives the following examples (p. 265), comparing an appositive relative and coordinated clauses to illustrate this point: (21) a. Did the officer arrest Clyde, who was the subject of a long manhunt, before he

could strike again? b. * Did the officer arrest Clyde before he could strike again, and {Clyde/he} was

the subject of a long manhunt. c. * Clyde was the subject of a long manhunt, and did the officer arrest

{Clyde/him} before he could strike again? d. * Clyde was the subject of a long manhunt. Did the officer arrest {Clyde/him}

before he could strike again? In coordination, different illocutionary forces can only be combined if the content of the first clause is known or uncontroversial (cf. Ross, 1967). The examples are again from Potts (2003: 265), citing Huddleston and Pullum (2002:1332). (22) a. It'll be very hot, so take plenty to drink. b. They've finished the job, but why did they take so long? I conclude that the structure of the secondary message expressed by appositional constructions consists of a full copular clause with its own illocutionary force and tense and a silent copula with the meaning of be.

Page 8: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

8

(23)

It is not surprising that the copula be is silent here. An empty verb is always interpreted as a form of be and many languages have an empty copula in normal copular sentences as well (cf. Stassen, 2008). For more arguments in favor of the idea that appositional constructions involve predication, I refer to Heringa (2007) and the references there. In the next sections, I will explain how it is possible that the apposition and the anchor are related by coordination and by predication at the same time, that is, how we can integrate the copular clause in (22) in the coordination phrase in (8). 4. Parenthesis The first step to take in combining predication and coordination in appositive clauses has to do with the non-restrictive nature of appositional constructions, the fact that the appositive content is secondary, an aside. We need a way to connect secondary messages like this to a matrix in general. Several people (eg. Dever 2001, Nouwen 2007, De Vries 2009) have shown that these messages really behave as if they are independent sentences following the matrix sentence, instead of being a part of that matrix. It is well known, for instance, that they are out of the scope of semantic operators in the host, such as quantifiers, negation, modals and adverbs. (24) a. * Every woman, a talkative person, participated in the discussion. b. A woman, a talkative person, participated in the discussion. c. Every woman participated in the discussion. # She was a talkative person. d. A woman participated in the discussion. She was a talkative person. (25) a. John did not kiss Mary, his girl friend. b. (i) = ... Mary is his girl friend (ii) � ... Mary is not his girl friend c. John did not kiss Mary. She is his girl friend.

VP T

CP

TP

SC

apposition

V Copula

ti

C [ill. Force]

SAdv TP

T'

anchori

Page 9: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

9

Also in other respects, secondary messages seem really invisible for elements in the matrix. For instance, they are not part of the interpretation of VP-ellipsis. Paraphrasing the apposition with an independent sentence results in exactly the same interpretation. (26) a. Links ziet u een oude kerk, een prachtig gebouw, [Dutch] left see you an old church a wonderful building

en rechts een nieuwe [e]. and right a new ‘ On the left, you see an old church, a wonderful building, and on the right, you see a new one.’

b. (i) = ... and on the right a new one (ii) � ... and on the right a new one, a wonderful building c. Links ziet u een oude kerk. Dat is een prachtig gebouw. left see you an old church that is a wonderful building Rechts ziet u een nieuwe [e]. right see you a new Being internal to the matrix, but still behaving independently, appositions are in the class of parenthesis. This relates them to constructions like appositive relative clauses, parenthetical clauses, parenthetical adverbials and comment clauses, as illustrated below. (27) a. Maaike, a nice girl, lives in Groningen. (apposition) b. Maaike, who is a nice girl, lives in Groningen. (appositive relative clause) c. Maaike - she is a nice girl - lives in Groningen. (parenthetical clause) d. Maaike lives in Groningen, if you want to know. (parenthetical adverbial) e. Maaike still lives, I suppose, in Groningen. (comment clause) For all these constructions the same independency and invisibility has been attested by various researchers (eg. Espinal 1991, Haegeman 1991, De Vries 2007). What does this mean for the structure? Syntactically, an element is visible via dependency relations. In particular, an element is visible for some other element if it is c-commanded by that element. This means that parenthetical construal somehow blocks a c-command relation. De Vries (2007) has shown extensively that parenthetical constructions are indeed not c-commanded by any element in the host. C-command, in turn, can be related directly to the basic syntactic operation merge. After all, if a derived structure is merged to a new element, this element c-commands exactly the piece of structure it merges with, that is its sister and everything included in that sister. Therefore, if c-command is blocked in some part of the structure, as in parenthetical constructions, one might argue that they are connected to the matrix by an operation other than merge, or at least a different version of merge. In this paper, I follow De Vries (2007), who proposes to distinguish two types of merge. On the one hand there is the usual merge from Chomsky (1995), which results in the dependency hierarchy with dominance and therefore c-command. On the other hand, there is a parenthetical merge, which results in a different type of dependency hierarchy, without c-command. An element merged in that way is invisible to its host. In the visual representation, elements related by this parenthetical construal can be combined by dotted

Page 10: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

10

instead of straight lines, as if they go behind the structure of the host, suggesting that they are no longer visible by looking down in the host tree. Since the position of parenthetical elements is usually free, I will assume that a parenthetical clause adjoins to a projection in the host. The position of the appositive clause is not free, though. It is a description of the apposition and is therefore always adjoined to that element. The apposition itself has to be adjacent to the anchor. The coordination relation discussed earlier takes care of that. The question then remains how parenthetical construal comes about. I take it that this is caused by a special head, Par, which merges parenthetically with the parenthetical structure and projects a parenthetical phrase, ParP. This ParP is then adjoined to some element in the matrix. The Par-head marks its complement as a secondary message, which results in the special 'comma intonation' pattern that characterises parenthetical material and in a non-restrictive semantics. This head triggers the parenthetical merge. Taking everything above into account, we now arrive at the following structure for appositional constructions. Note that in this structure Par and CP are daughters of ParP. (28) This section is about the secondary message, an interesting element.

It is clear that the structure given above is problematic, because both the anchor and the apposition are represented twice, whereas they are only spelled out once. The next two sections deal with this problem. 5. Pronouns In section 4, we saw that appositional constructions are in the class of parenthesis. In particular, they look very similar to a restricted set of appositive relative clauses and parenthetical clauses, namely the set of those constructions that contain a copula. Consider the examples from (27 a-c), repeated here as (29), again.

[preliminary version]

This section is about :P

DP the secondary

message

:

VP T

DP an interesting element

DP

ParP

CP

Par

TP

DPi the secondary

message

SC

CP

DP an interesting element

V is

ti

Page 11: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

11

(29) a. Maaike, a nice girl, lives in Groningen. (apposition) b. Maaike, who is a nice girl, lives in Groningen. (appositive relative clause) c. Maaike - she is a nice girl - lives in Groningen. (parenthetical clause) Especially if we assume the idea that (29a) implicitly expresses the secondary message Maaike is a nice girl, the resemblance between these sentences is very strong. This might lead us to the idea that appositions have exactly the same structure as either appositive relatives or parenthetical clauses. This is indeed proposed by several people. For instance, Smith (1964) and, in different ways, Cardoso & De Vries (2008) and O’Connor (2008) propose to analyse appositions as appositive relative clauses. The visible differences between appositions on the one hand, and the corresponding appositive relatives and parentheticals on the other hand, are the presence of a pronoun and a copula. In section 3, I argued that a (silent) copula is also present in appositions. That raises the question whether we also want to assume a pronoun in the structure, referring to the anchor. Even if we do not follow the above-mentioned analyses in terms of appositive relatives, we could use the idea of having a pronoun in the structure. This would help us to get rid of the first double representation in our analysis. Instead of doubly representing the anchor, we could use a (silent) pronoun. As we saw in section 4, appositional constructions behave as if the secondary message is an independent sentence, following the matrix. Typically, these sentences contain pronouns referring to 'the anchor' as well. These are so-called E-type pronouns, characterised by the fact that they do not just refer to their antecedent, but also take over the scope of that antecedent in the sentence. The same phenomenon shows up in appositional constructions (cf. Doron 1994, Del Gobbo 2007, Nouwen 2007). (30) a. Maaike is watching a movie, a thriller. b. => the movie Maaike is watching is a thriller. c. Maaike is watching a movie. It is a thriller. This behaviour is typical for E-type pronouns. Just repeating the apposition as a copy would not include the scope. This means that the secondary clause in appositional constructions indeed has to contain a pronoun. The question remains, however, whether appositions can be straightforwardly analysed as appositive relatives, which implies the presence of a relative pronoun. There are some problems with an analysis in terms of relative clauses. The main problem is that they can only paraphrase the attributive type of appositions. Whereas the appositive relative (29b) is a perfect equivalent of the attributive construction (29a), it is impossible to use an appositive relative to describe an identificational apposition like (31a). Parenthetical clauses do not have this restriction (31c). (31) a. Edinburgh’s main attraction, the castle, has over one million visitors a year. b. * Edinburgh’s main attraction, which is the castle, has over one million visitors

a year. c. Edinburgh’s main attraction – that is the castle – has over one million visitors

a year.

Page 12: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

12

The reason that appositive relatives cannot be used to paraphrase identificational apposition is probably related to the movement of the relative operator. Remember from section 3 that identificational appositions are related to specificational predication. Heycock and Kroch (1999: 377) have shown that wh-extraction of the subject from this type of copular clauses is impossible. An explanation for this restriction on wh-extraction would explain the blocking of the movement of the relative operator in this type of appositive relative constructions as well. (32) a. * [Which of the themes]I do you think ti is that phrase of music? b. * [Whose opinion of Edinburgh]I do you think ti is your opinion of Philadelphia? Another difference between appositions and appositive relatives is the possibility to use subordinators such as though. As shown in section 3, appositions have this possibility. Appositive relatives, on the other hand, do not. Again, parentheticals are on a par with appositions. (33) a. Bill, though a big sports fan, did not see the match. b. * Bill, though who is a big sports fan, did not see the match. c. Bill – though he is a big sports fan – did not see the match. Since paraphrasing appositions with parenthetical clauses does not have the problems that paraphrasing with appositive relatives runs into, I conclude that appositions can be analysed as parenthetical clauses with a silent pronoun, little pro. This silent pronoun behaves as an E-type pronoun in the sense that it takes over the scope of the element it refers to. This is due to the parenthetical construal of the appositional message, which causes it to be interpreted as a quasi-independent sentence. A potential problem for this approach can be inferred from Wang et al. (2004). They show that appositions and their paraphrases by relative clauses with be sometimes differ in interpretation. The anchor of an apposition can have a non-specific reading, but the anchor of an appositive relative with be cannot. The examples in (34a,b) are Wang et al.’s (33a,b). (34) a. John wants a car, a red one. b. John wants a car, which is red. c. John wants a car - it is a red one. d. John wants a car. It is a red one. In (34a), a car can be interpreted both non-specifically and specifically. In (34b), on the other hand, it can only be interpreted specifically. In this respect, parenthetical clauses like the one in (34c) show the same behaviour as appositive relatives. One could argue that this is due to the fact that the antecedent is in the scope of an intensional verb, whereas the pronoun is out of it. This can also be illustrated by means of subsequent sentences in a discourse. If an indefinite has narrow scope with respect to an intensional verb, it cannot function as an antecedent for a pronoun in the next sentence. Example (34d) is infelicitous if the indefinite in the the first sentence has a de dicto reading.

Page 13: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

13

It is well known, however, that this restriction can be overcome by using another intensional or modal verb in the sentence that contains the anaphoric pronoun. This phenomenon is called modal or intensional subordination (see Roberts 1996 and the references there). Roberts (1996: 217) gives the following examples, where the intensional verbs are in bold and the relevant pronouns are underlined: (35) a. Jan expected to get a new puppy soon. She intended to keep it in her back yard. b. John wants to catch a fish. He plans to eat it for supper. c. Alice fears there’s a squirrel in her kitchen cabinets. She hopes to trap it alive and turn it loose outside. Roberts argues that modal subordination is a pragmatic phenomenon involving accommodation. The speaker accommodates to the world in which the intensions of the subject are true. The same effect could be obtained in the paraphrases of the apposition in (34a), repeated here as (36a), if we use a modal like should. If we do this, an appositive relative (36b) and a parenthetical clause (36c) get exactly the same interpretation as their appositional counterpart (36a). (36) a. John wants a car, a red one, for his birthday. b. John wants a car, which should be red, for his birthday. c. John wants a car – it should be red – for his birthday. I therefore assume that appositions can have an implicit form of modal subordination, accommodating to the desirable world of the subject of an intensional verb. To resume, I argued to analyse appositions as parenthetical clauses with a silent pronoun, little pro, and a silent verb, be. This pronoun behaves as an E-type pronoun in taking over the scope of its antecedent. Using a silent pronoun solves the problem of the double representation of the anchor. The next section deals with the double representation of the apposition. 6. Multidominance If we use a pronoun to solve the problem of the double representation of the anchor in the structure of appositional constructions, one might wonder whether the same solution can be used for the apposition. However, it is reasonable to look for a different solution in this case. Remember that one of the reasons to assume a pronoun as the subject of the secondary message was the E-type reference of this subject to the anchor, including the anchor's scope. The apposition, on the other hand, does not show such behaviour. (37) a. Maaike is watching a movie, a thriller. b. => the movie Maaike is watching is a thriller c. ≠> the movie Maaike is watching is the thriller Maaike is watching d. ≠> the movie is the thriller Maaike is watching

Page 14: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

14

The apposition and the element referring to it in the secondary message are exactly the same. Thus, we may assume that there is just one element, which functions in two structures at the same time. The apposition is shared by the coordinative phrase and the secondary CP. In terms of merge, this means that the apposition is merged twice, with two different sisters, resulting in two different dominance relations. In the literature, this type of construal has been called multi-dominance, sharing, or external remerge. With the idea of a silent pronoun representing the anchor in the secondary message and a multiply dominated apposition, we now arrive at the following structure for appositional constructions: (38) Maaike is watching a movie, a thriller.

The idea of a multiply dominated apposition immediately explains its ambivalent relationship with the anchor. An apposition is both coordinated and subordinated to its anchor. This makes the appositional construction a short and simple spell-out of a fairly complex structure. It gives a speaker the opportunity to add a complete message about an element in the sentence he pronounces, without having to use a new sentence and therefore without spelling out a subject or a verb.

[preliminary version]

Maaike is watching :P

DPi a movie

:

VP T

DP

ParP

CP

Par

TP

DPi pro

SC

CP

DP2 a thriller

V is

ti

Page 15: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

15

Note that the subject and the verb of the secondary message are covertly present in the form of a silent pronoun and a copula. The combination of these two can be pronounced as 'that is' or the Latin translation 'id est', abbreviated as 'i.e.'. This suggests that these complex apposition markers do not occupy the head of the colon phrase, but form the overt realisations of the subject and the verb of the appositional message. 7. Layered DPs Though the structure in (38) looks quite complicated, all the properties of appositions we have seen so far follow from it straightforwardly. The idea that the apposition is simultaneously part of a coordinative structure and of a predicational structure explains the intuitive difference in interpretation between the matrix (39c) and the secondary clause (39d) for attributitive appositions (39a). In the matrix, the apposition a talkative person has a referential, or strong, interpretation. In the secondary clause, on the other hand it is interpreted non-referential, or weak. (39) a. John had a meeting with his colleague, a talkative person. b. John had a meeting with his colleague. (colleague = strong) c. John had a meeting with a talkative person. (person = strong) d. His colleague is a talkative person (colleague = strong; person = weak) The coordinative structure represents the strong reading of the apposition, a talkative person in (39). After all, coordination combines two elements with a similar semantic function. The anchor in (39), his colleague, has a referential function and therefore the apposition, being the second conjunct, should have this function as well. This phenomenon is known as the Law of Coordination of Likes (LCL), first stated in terms of syntactic categories by Chomsky (1957) and Williams (1981), but later adjusted by Sag et al. (1985) in terms of syntactic feature bundles, and by Munn (1993) in terms of semantic types. The following example shows that DPs with a referential reading, of type <e>, indeed cannot be coordinated with predicative DPs, of type <e,t>: (40) a. My favourite animal is an herbivore and a good runner. (<e,t> and <e,t>) b. My favourite animals are the giraffe and the rhinoceros. (<e> and <e>) c. * My favourite animal is the giraffe and an herbivore. (<e> and <e,t>) The predicative structure represents the predicative meaning of the apposition, a talkative person in (38), as shown in the paraphrase of the appositive content in (38d). The apposition has this interpretation because it is attributive. In section 3 we saw that this means that the appositional message is a predicational copula clause; not a specificational one. Therefore, the apposition denotes a property in this case, and therefore has a weak interpretation. This implies, however, that the very same DP, shared by two different structures, could be of two semantic types simultaneously. How can this be explained? I propose that what is actually shared between the two structures is not a DP, but the predicative core of a DP. The coordinative structure then has an extra layer on top of that predicative core, resulting in a referential reading for the DP as a whole. In this, I follow Zamparelli

Page 16: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

16

(2000), who argues that DPs indeed have several functional layers. These layers are used for a mapping between the syntax and the semantics of the DP. For our purposes, the strong determiner phrase, SDP, and the predicative determiner phrase, PDP, are relevant. The SDP represents referential entities, of type <e>. These SDPs always contain a PDP, a property of type <e,t>. A DP in Zamparelli's view can only be referential if the SDP layer is present. I want to propose that SDP is always present, and contains a feature which indicates whether the DP gets a strong interpretation or not. A feature with value [+strong] type-shifts the DP from type <e,t> to <e>. The SDP is the extra layer we need and will be present both in the coordinative and in the predicative structure. The PDP will be shared. This results in the following structure for appositional constructions: (41) Mary kissed her friend, a linguist.

Note that in (41) the features on the heads of the two SDPs have different values. In the coordinative structure the SDP gets a strong interpretation, whereas the SDP in the predicative structure gets a weak interpretation. Remember that this is due to the fact that

[final version]

&:P

SDPi [+strong] her friend

SDP

ParP

Mary kissed

CP

SD [-strong]

SDP

SDPi [+strong] pro

T

VP

V

SC

SDP [+strong] ti

SDP

CP

TP

PDP a linguist

SD [+strong]

Par :

Page 17: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

17

the appositional construction in (41) is attributive. The apposition gets a weak interpretation in the secondary clause, because it denotes a property in this case. The coordinative structure, on the other hand, obeys the LCL, which says that the two conjuncts in one construction have to be of the same type. The two SDPs that share the apposition do not need to have different values for their interpretation features. In identificational appositions, such as John, my neighbour, went to the store, also the predicate SD of the parenthetical clause has a strong interpretation, and therefore a [+strong] feature. This means that the subject and the predicate in the secondary message are of the same semantic type, <e>, resulting in a specificational, or equative, copular clause (cf. Heycock & Kroch 1999). Sharing just part of a layered structure is only necessary in the case of DPs, since just this category typically can have different semantic types. Appositions of other categories, like those in (42), are always of the identificational type and have no difference in interpretation between the matrix and the appositive content. (42) a. I met my neighbour over there, in the garden. (PP) b. My wife is nice, even wonderful. (AP) c. That rabbit kicked the bucket, that is, it died. (CP) A problem arises, however, if it would be possible that the anchor and the apposition in one construction have different syntactic categories, each of an inherently semantic type, violating the LCL. At first sight, this type of construction, the unbalanced apposition, seems to exist, for example in the case of predicative postmodifiers: (43) a. The monkey, angry, showed its teeth. b. The children, in a bad mood, ruined everything. The position of these modifiers is more flexible, though. They do not need to be adjacent to the anchor. Compare (44) to the regular appositions in (45): (44) a. John, drunk, knocked at the door. b. John knocked, drunk, at the door. c. John knocked at the door, drunk. (45) a. John, a nice guy, knocked at the door. b. * John knocked, a nice guy, at the door. c. * John knocked at the door, a nice guy. This suggests that the structure of these constructions does not involve coordination. Therefore, I leave these constructions for future research. To sum up, I state that the structure of appositional constructions involves both coordination and predication. The apposition consists of several layers, where one lower layer is shared by two higher layers. One of these higher layers is part of the coordinative structure, while the other one is in the predicative structure. This means that the apposition simultaneously functions in two distinct structures.

Page 18: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

18

8. Case marking This section provides extra evidence for a multidominance analysis of appositions by looking at interaction effects between the two substructures that share the apposition in this analysis. Being a second conjunct in coordination and a predicate at the same time has a potentially conflicting effect on the morphology of the apposition. This becomes clear if we look at the case marking patterns for appositions in several languages Coordination can influence the case marking of the second conjunct in two ways. First, the second conjunct can get the same case as the first conjunct. This happens for example in German and Russian. Second, the second conjunct can get a default or emphatic case. Examples of languages that show this are English and (dialects of) Norwegian, which both have the accusative case as default. The examples in (47) are from Schütze (2001: 214/226): (46) a. Er sah den blauen Himmel und den Ozean. [German] he saw the:ACC blue:ACC sky and the:ACC ocean b. Ja znaju Dimu i Mishu. [Russian] I know Dima:ACC and Misha:ACC (47) a. Did your parents or him pick up Mary? [English] b. Ha nog meg var sammen om det. [Stavanger dialect] he:NOM and me:ACC were together about it 'He and I were in it together.' Languages also have several possibilities to mark the non-verbal predicate. First, the predicate can get zero case, as in Japanese. Note that the nominative case in Japanese does have a marking: -ga. Thus, predicative case is different from nominative case. Second, the predicate can agree in case with the subject. In secondary predication, such as ECM-constructions, this could be just any case, like in coordination, but in the situation that is relevant for us, primary predication, the subject, and therefore the predicate, usually gets nominative case. A good example is Icelandic. Third, a language can have a special predicative case. In Polish, for instance, the predicate always gets marked as instrumental. The examples in (48) are from Citko (2008: 283/284). (48) a. John-wa sugureta suugakusha da [Japanese] John:TOP fine mathematician COP ‘John is a fine mathematician.’ b. Hún er kennari. [Icelandic] she:NOM is teacher:NOM c. Jan jest studentem. [Polish] Jan:NOM is student:INSTR If a language assigns the same case to second conjuncts and to predicates, there is no problem. English and Norwegian are examples of such languages. Remember that these languages get accusative as default case on second conjuncts. Predicates in these languages get accusative case as well. Example (50) is from Schütze (2001: 236).

Page 19: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

19

(49) a. The best athlete is her. b. The smartest student is me. (50) Det er ham / meg. [Norwegian] it is him / me 'It is him / me.' Since coordinative case and predicative case are both accusative, PF does not have to choose between the two and spells out accusative on appositions too. The examples are from Schütze (2001: 210/227). (51) a. The best athlete, her, should win. b. The teacher said that the smartest student, namely me, should give a speech. (52) Laereren sa at den smarteste studenten, altså [Norwegian] the-teacher said that the smartest student, thus meg / jeg, skulle gi en tale. me / I, should give a speech. The question now arises which case the apposition gets if coordination and predication result in a different case. Suppose that an element, in this case the apposition, can get more than one case feature. This would mean that both the coordinative structure and the predicative structure could assign a feature to the apposition. As a consequence, PF has to choose which of these features has to be spelled out. Rules with respect to this choice can be language specific and thus result in variation (see Matushansky 2008 for a similar mechanism). In theory, a mechanism like this gives us three possibilities. First, the apposition gets the coordinative case. Second, the apposition gets the predicative case. Third, the derivation crashes because of the difference in case and therefore apposition is impossible (in certain contexts). It turns out that these three possibilities all occur in natural languages. The first possibility, coordinative case, appears in Russian and Czech. In these languages, the apposition gets the same case as the anchor, independent of the type of apposition or its syntactic function, just like the second conjunct always gets the same case as the first conjunct. (53) a. Džon, lingvist, vystupil s dokladom ob Aranae. [Russian] John:NOM linguist:NOM talked with report about Aranae 'John, a linguist, gave a talk on Aranae.' b. V 1973, Skylab vzjal v kosmos dvux životnyx, paukov In 1973 Skylab took in space two:ACC animals:ACC spiders:ACC Arabellu i Anitu. Arabella:ACC and Anita:ACC 'In 1973, Skylab took two animals, the spiders Arabella and Anita, into space.'

Page 20: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

20

(54) a. Astronauti dali Arabelle, zahradnímu pavoukovi, vodu [Czech] astronauts gave Arabella:DAT garden spider:DAT water:ACC a maso, dobrou pavou�í pochoutku. and meat:ACC good spider delicacies:ACC 'The astronauts gave Arabella, a garden spider, water and meat, a nice spider

meal.' b. Deinopida, pavouk, zahlédl ko�ist svýma o�ima, dv�ma Deinopida:NOM spider:NOM spotted prey his eyes:INSTR, two obrovským složenými o�kami. enormous compound lenses:INSTR 'Deinopida, a spider, spotted a prey with his eyes, two enormous compound

lenses.' We therefore conclude that Russian and Czech have a rule which says that an element with two case features, one of them being predicative, is spelled out with morphology for the non-predicative case. The second language type shows predicative case in appositions. Romanian is such a language. Romanian shows a distinction between the two types of appositions. Identificational appositions, as in (55), get the same case as the anchor, just like in Russian and Czech. Attributive appositions, like (56), on the other hand, always get nominative case. Since Romanian non-verbal predicates get nominative in primary predication, we conclude that this is the predicative case. (55) a. În 1973, Skylab a luat dou� animale, pe [Romanian] in 1973 Skylab PAST:3SG took two:ACC animals the p�ianjeni Arabella �i Anita, în spa�iu. spiders:ACC Arabella and Anita in space 'In 1973, Skylab took two animals, the spiders Arabella and Anita, into space.' b. Au dat-o �i Anitei, celuilalt p�ianjen. PAST:3PL gave-it too Anita:DAT other:DAT spider 'They gave it Anita, the other spider, as well. (56) a. Astronautii au dat Arabelei, un p�ianjen [Romanian] astronauts:DEF PAST:3PL gave Arabella:DAT a:NOM spider de gr�din�, ap� �i carne, o mas� frumoas� pentru un p�ianjen. of garden water and meat, a meal nice for a spider 'The astronauts gave Arabella, a garden spider, water and meat, a nice spider meal. b. Aspectul Deinopidei se aseam�na cu cel al lui Shreck, appearance Deinopida:GEN self look.like with that of he:GEN Shrek c�pc�unul modern. ogre:DEF:NOM modern 'The appearance of Deinopida is similar to that of Shrek, the modern ogre.' How can it be detected whether the apposition is attributive? This is related to the interpretation of the DP in that case. As I argued in section 7, the SDP layer of an

Page 21: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

21

attributive apposition has a [-strong] feature. Since this is the normal interpretation of a predicate, a language can demand a DP with such a feature to get the predicative case. This means that the case marking of appositions in Romanian is the result of two rules, one similar to the rules we gave for Russian and Czech, and one that requires nominative case for [-strong] predicates. Finally, the third possibility is that the difference between the coordinative case and the predicative case results in a crash of the derivation. This means that the apposition is ungrammatical, both with the coordinative and the predicative case. This happens in Hungarian and Icelandic. Like Romanian, these languages show a distinction between identification (57/58) and attribution (59/60). Identificational appositions show the same behaviour as in all the other languages: anchor and apposition get the same case. (57) a. Hann sagði að áttfætlur, það er köngulær, hefðu [Icelandic] He said that araneae:NOM that is spiders:NOM had tvískiptan búk two.divided body 'He said that Aranae, that is spiders, have two body segments.' b. Deinopida kastaði vefnum yfir bráð sína, lirfuna Deinopida threw web:DEF over prey:ACC her:ACC larva:DEF:ACC hans Jóns. his John:GEN 'Deinopida cast the net over her prey, John's caterpillar.' (58) a. Azt mondta, hogy az Araneaknak, vagyis a [Hungarian] that:ACC said:3SG that the Aranea:PL:DAT or the pókoknak, két testtájuk van. spiders:DAT two body.area is 'He told that Aranae, ie spiders, have two body segments.' b. 1973-ban a Skylab az ürbe vitt két állatot, 1973-in the Skylab the space.in brought two animal:ACC, Arabella és Anita pókokat. Arabella and Anita spiders:ACC 'In 1973, Skylab took two animals, the spiders Arabella and Anita, into space.' Attributive appositions, on the other hand, are impossible in Hungarian and Icelandic if the corresponding anchor is not in the nominative. Instead, an appositive relative has to be used: the relative pronoun cannot be left out. In that case, the predicate gets nominative case. (59) Arabella bjó til fyrsta vefinn, *(sem var) [Icelandic] Arabella made to first:ACC web:DEF:ACC that was Ófullkomin / * ófullkomna smíð. incomplete:NOM incomplete:ACC construction 'Arabella made the first web, which was an incomplete construction.'

Page 22: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

22

(60) Az �rhajósok Arabellának, a kerti póknak vizet [Hungarian] the astronauts Arabella:DAT the garden spider:DAT water:ACC és húst adtak, *(ami) kitünö pókétel / * pókételt. and meat:ACC gave, which lovely spiderfood:NOM spiderfood:ACC 'The astronauts gave Arabella, the garden spider, water and meat, which is a nice spider meal.' Again, we can use the interpretation feature in the SDP to explain this behaviour. Instead of forcing to choose the nominative if a DP is [-strong], these languages require that no other case feature than nominative is assigned to this DP. If any other feature is present as well, the derivation crashes. Note that attributive appositions with a nominative anchor are not excluded by this rule, since no other case than nominative is assigned to the apposition in that case. This is illustrated in (61). (61) János, egy nyelvész, a pókokról tartott eladást. [Hungarian] John:NOM a linguist:NOM the spiders:DEL held lecture 'John, a linguist, gave a talk on spiders.' To resume, we predicted that there are three types of languages where predicative case and coordinative case are different. All types show coordinative case in identificational appositions. They differ with respect to case marking in attributive appositions, however. One type of language shows coordinative case in these constructions as well. This is true for Russian and Czech. One type shows predicative case in these constructions. This is shown for Romanian. Finally, one type has no attributive appositions with a non-nominative anchor. This is found in Icelandic and Hungarian. 9. Conclusion In this paper, I argued that the two parts of appositional constructions, the anchor and the apposition, are related in two ways. First, they are coordinated by specifying coordination. Second, the apposition functions as the predicate of the anchor. This predication relation is shown to have the structure of a full CP with an empty copula with the meaning of be. This CP is related to the matrix clause by use of parenthetical construal, resulting in an independent, secondary message. In this secondary clause, the anchor is represented by a silent pronoun. The apposition functions simultaneously in the coordinative and in the predicative structure. This is reflected in the structure by multidominance. The apposition is shared between two clauses: the matrix and the secondary, appositive, clause. The part that is shared is argued to be a Predicative DP, whereas both of the dominating structures involve a higher functional layer of this DP, the Strong DP. This SDP has a feature coding the interpretation of the DP, strong or weak. In one type of appositions, identification, this feature has the same value in the coordinative structure and in the predicative structure. In the other type, attribution, the DP is interpreted strong in the anchor and weak in the apposition. This system, where appositions function both in a coordinative and in a predicative structure, combined with the coded difference in interpretation between two types of apposition, is used to explain the patterns of case

Page 23: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

23

marking in these constructions in natural languages. I assume that one element, in this case the apposition, can get more than one case feature, in this case a coordinative and a predicative one. Languages have rules to choose which of these features has to be spelled out if they are different. For identification, all languages choose to spell out the coordinative case. For attribution, on the other hand, there are three possibilities. Languages like Russian and Czech spell out the coordinative case in these constructions as well. Languages like Romanian choose to spell out the predicative case in attribution, due to the weak interpretation of the appositive DP. In Hungarian and Icelandic, the derivation crashes for attributive apposition if the anchor is not in the nominative. This is again related to the weak interpretation of the appositive DP. To solve this problem, these languages can use an appositive relative clause to express the meaning of an attributive apposition. To resume, the patterns of case marking in appositional constructions can be explained if these constructions involve both coordination and predication. In this paper, I argued that an apposition functions simultaneously as a second conjunct and as a predicate, because it is shared by a coordination phrase and a secondary copular clause. References Berckmans, P. (1994). 'Demonstration, apposition and direct reference.' Communication and Cognition 26, 499-512. Cardoso, A. & M. de Vries (2008), 'Internal and external heads in appositive constructions', Ms., University of Lisbon & University of Groningen. <http://www.let.rug.nl/dvries/pdf/2008-appositive-constructions.pdf> Citko, B. (2008), 'Small clauses reconsidered: Not so small and not all alike', Lingua 118 (3), 261-295. Chomsky, N. (1995), The minimalist program. Cambridge, Mass.: MIT Press. Chomsky, N. (1957), Syntactic Structures. Den Haag: Mouton. Corazza, E. (2005). 'On epithets qua attributive anaphors.' Journal of Linguistics 41, 1-32. Del Gobbo, F. (2007). 'On the syntax and semantics of appositive relative clauses.' In N.

Dehé & Y. Kavalova (eds.) Parentheticals, 173-201. Amsterdam: John Benjamins.

Dever, J. (2001).'Complex demonstratives.' Linguistics and Philosophy 24: 271-330. Doron, E. (1994). 'The discourse function of appositives.' In R. Buchalla & A. Mitwoch (eds.) Proceedings of the ninth annual conference of the Israel Association for Theoretical Linguistics and of the workshop on discourse. Jerusalem: Hebrew University, 53-65. Espinal, M. (1991), 'The representation of disjunct constituents', Language 67: 726-762. Haegeman, L. (1991). 'Parenthetical adverbials: The radical orphanage approach.' In S. Chiba, A. Ogawa, Y. Fuiwara, N. Yamada, O. Koma & T. Yagi (eds.) Aspects of modern English linguistics: Papers presented to Masatome Ukaji on his 60th birthday, 232–254. Tokyo: Kaitakusha. Heringa, H. (2007), Appositional constructions: coordination and predication. In: M.E. Kluck & E.J. Smits (eds.), Proceedings of the Fifth Semantics in The Netherlands Day. Den Haag, RS Staten Kopie, 67-82.

Page 24: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

24

Heringa, H. & M. de Vries (2008). 'Een semantische classificatie van apposities.' Nederlandse Taalkunde 13 (1): 60-87. Heycock, C. & A. Kroch (1999), 'Pseudocleft Connectedness: Implications for the LF Interface Level', Linguistic Inquiry 30 (3), 365-397. Hockett, C.F. (1955), 'Attribution and Apposition', American Speech 30, 99-102. Huddleston, R. & G.K. Pullum (2002). The Cambridge Grammar of the English

Language. Cambridge, Mass: MIT Press. Johannesen, J.B. (1998), Coordination, New York: Oxford University Press. Kayne, R. (1994). The antisymmetry of Syntax. Boston: MIT Press. Klein, M. (1977). Appositionele constructies in het Nederlands. PhD Dissertation, University of Nijmegen. Koster, J. (2000), 'Extraposition as parallel construal', Ms. Groningen: University of Groningen. <http://www.let.rug.nl/koster/papers/parallel.pdf> Kraak, A. & W. Klooster (1968). Syntaxis. Culemborg: Stam-Kemperman. Matushansky, O. (2008). 'A case study of predication.' In F. Maruši� & R. Žaucer (eds.) Studies in formal Slavic linguistics. Contributions from Formal Description of Slavic Languages 6.5. Frankfurt am Main: Peter Lang, 213-239. Munn, A. (1987), 'Coordinate structure and X-bar theory' McGill Working Papers in Linguistics 4(1), 121-140. Munn, A. (1993) Topics in the syntax and semantics of coordinate structures. PhD Dissertation. University of Maryland, College Park. Nouwen, R. (2007). On appositives and dynamic binding. Research on Language and Computation 5 (1), 87-102. O'Connor, K.M. (2008). Aspects de la syntaxe et de l'interprétation de l'apposition à antécédent nominal. PhD Dissertation. Université Charles de Gaulle, Lille 3. Partee, B. (1998). 'Copula Inversion Puzzles in English and Russian.' In K. Dziwirek, H. Coats & C. Vakareliyska. (eds.) Annual Workshop on Formal Approaches to Slavic Linguistics: The Seattle Meeting. Ann Arbor: Michigan Slavic Publications, 361-395. Potts, C. (2003). The logic of conventional implicatures. PhD Dissertation, University of California, Santa Cruz. Potts, C. (2007), 'Conventional implicatures, a distinguished class of meanings', in: G. Ramchand & C. Reiss (eds.) The Oxford Handbook of Linguistic Interfaces. Oxford: Oxford University Press, 475-501. Quirk, R., S. Greenbaum, G. Leech & J. Svarvik (1985). A comprehensive grammar of the English language. Londen: Longman. Roberts, C. (1996). ‘Anaphora in intensional contexts’, in: S. Lappin (ed.) The Handbook

of Contemporary Semantic Theory. Oxford: Blackwell Publishing, 215-246. Ross, J.R. (1967). Constraints on Variables in Syntax. Ph.D. thesis, MIT, Cambridge,

MA. Published as Infinite Syntax, Norwood, NJ: Ablex (1986). Sag, I.A., G. Gazdar, Th. Wasow & S. Weisler (1985). 'Coordination and how to

distinguish categories', Natural Language and Linguistic Theory 3: 117-171. Schütze, C.T. (2001). 'On the nature of default case.' Syntax 4 (3), 205-238. Smith, C. (1964). ‘Determiners and Relative Clauses in a Generative Grammar of

English’, Languge 40, 37-52. Reprinted in: S. Reibel and S. Schane (eds.) Modern Studies in English, New-York: Prentice-Hall, 247-263.

Page 25: my neighbour, my neighbour - Faculteit der Letteren | …heringa/multi_apps.pdf · clauses (see for example Partee 1998). The main classes of appositions are identification The main

25

Stassen, L. (2008). 'Zero copula for predicate nominals.' In M. Haspelmath, M. Dryer, D. Gil & B. Comrie (eds.) The World Atlas of Language Structures Online. Munich: Max Planck Digital Library, chapter 120. <http://wals.info/feature/description/120>. Vries, M. de (2009), Specifying Coordination: An Investigation into the Syntax of Dislocation, Extraposition and Parenthesis. In C. R. Dryer (ed.) Language and Linguistics: Emerging Trends, 37-98. New York: Nova. Vries, M. de (2007). ‘Invisible Constituents? Parentheses as B-Merged Adverbial Phrases.’ In N. Dehé & Y. Kavalova (eds.) Parentheticals, 203-234. Amsterdam: John Benjamins. Vries, M. de (2006), 'The syntax of appositive relativization. On specifying coordination, false free relatives and promotion', Linguistic Inquiry 37, 229-270. Wang, L., E. McCready & B. Reese (2004), 'Nominal appositives in context', in: Proceedings of the Western Conference on Linguistics (WECOL 2004), University of Southern California, Los Angeles, CA. Williams, E. (1981) 'Transformationless Grammar.' Linguistic Inquiry 12, 645–654. Zamparelli, R. (2000). 'Layers in the Determiner Phrase.' Revised PhD Dissertation, University of Bergamo.