Subordination and Coordination Practice Directions: Combine the sentences by using the coordinating or subordinating conjunctions or the adverbial conjunctions in parentheses. Check for punctuation and logic. Sometimes, more than one answer is possible; however, be sure each revision is either a compound or complex sentence. Example: Jim followed the directions carefully. The pie tasted great. (because) Because Jim followed the directions carefully, the pie tasted great. The pie tasted great because Jim followed the directions carefully. (so) Jim followed the directions carefully, so the pie tasted great. (for) The pie tasted great, for Jim followed the directions carefully. (therefore) Jim followed the directions carefully; therefore, the pie tasted great. 1. Jill always studied hard. She earned only average grades. (but) (although) (however) (nevertheless) 2. Barry hoped to be a professional musician. He practiced daily. (since) (consequently) (for) (thus) 3. I read three magazines each week, I read two newspapers every day. (and) (moreover) (furthermore) (in addition) 4. She wants to be an engineer. She wants to be a chemist. (or) (and) M:\9-TLC\TLC Web Design\Handouts Worksheets\Grammar.Punctuation.Writing\Subordination and Coordination Practice.doc ivory 5. His dad has a terrible temper. He scares people. (so) (therefore) (because) (for) (hence) 6. He wants to be a politician. He can make changes in the law. (so that) (then) 7. Josh visits us. He makes us laugh. (when) (whenever) (as soon as) 8. My sister hates to stop at the mall. She cannot tolerate crowds. (so) (for) (therefore) (consequently) (since) 9. I know exercising is important. I don’t allow any time for it. (nevertheless) (however) (although) (even though) (yet) Revista Alicantina de Estudios Ingleses 7 (1994): 83-95 Apposition in English: A Linguistic Study Based on a Literary Corpus María Dolores Gómez Penas Universidad de Santiago de Compostela I Different approaches have been taken towards what apposition is, and as a consequence, divergences of opinión have emerged when deciding what is or is not an apposition. Diverse problems occur when we encounter structures that have a type of relation that, although different, shares the linguistic features that have been assigned to other relations within the linguistic system. In the studies on apposition that have been carried out, the following questions are asked: What kind of grammatical relation is there between the elements in apposition? Is it a relation of coordination or subordination? What kind of structures can be considered appositions? A lot of answers have been given to these questions. In the following we shall examine the criteria that have been used in order to define apposition, from the traditional ones to those used nowadays. Once we have described those criteria, we shall go on to analyse the syntactic, semantic and pragmatic characteristics of the examples in our literary corpus. The following corpus has been used in this research: A GoodMan in África (GMA), by William Boyd; White Mischief (WM), by James Fox; Hotel du Lac (HL), by Anita Brooker; The Child in Time (ChT), by Ian McEwan; TheRemains oftheDay (RD), by Kazuo Ishiguro. II Traditional English grammars do not deal much with apposition. G. O. Curme affirms that there is a relation of subordination in the appositional structure and includes it in that part of his book that he dedicates to the subordínate elements in the clause. He ñames these elements modifiers. He uses the formal criterion of position in the clause to define apposition: "A noun which explains or characterizes another is placed alongside of it and from its position is accordingly called an appositive (i.e. placed alongside of): Smith, the banker" (129). 84 Revista Alicantina de Estudios Ingleses The same formal criterion is found in Otto Jespersen's definition. Unlike Curme, he considers that apposition is a kind of coordination in which the absence of a firm cohesión between its elements is indicated by a pause and by intonation (13). These grammarians do not restrict apposition to noun phrases, they consider that linguistic units, such as the clause and the sentence, can also be elements of an appositional structure. Both, Curme and Jespersen, identify different kinds of apposition. And so, Curme mentions two types of apposition that he ñames loóse apposition: "Where the appositive noun follows the governing noun in a rather loóse connection . . . Mary, the belle of the village" (129), and cióse apposition: "King George . . . the relation here between the appositive and its governing word is so cióse that they are in many cases felt as a compound" (131). Neither Jespersen ñor Curme clearly define what apposition is. The criteria they use to define it are not very precise and do not help us to establish the difference between apposition and other syntactic relations. R. W. Zandvoort does not study apposition thoroughly either. He gives a series of examples and just differentiates one from the other. He asserts that between "His father [or: a] renowed physician, died last week" and "his brother the explorer" there exists a difference "similar to that between continuative and restrictive clauses" (202), and he mentions subordination when comparing the examples "My brother Charles" and "King George." According to Zandvoort, a clear relation of subordination is not observed between the two elements in the structure "My brother Charles." However, he considers that in "King George" there is subordination of the element "King" with respect to the proper noun "George." This same idea will appear in later studies as we shall see in the following. For the American distributionalist Ch. F. Hockett appositives are a subtype of coordínate constructions. Both have in common the linguistic feature of having a double head. He postulates the following criteria for the appositives: "(1) the constitute must be endocentric; (2) the ICs must belong to the same major form class; (3) there must be no more justification for taking the first IC as attribute of the second as head than for the reverse; (4) the ICs must refer to the same entity" (101). The second criterion, that the immediate constituents must belong to the same syntactic class, can be easily refuted. As we shall see in some examples from our corpus, both elements in apposition may belong to different syntactic classes. And so, one of them may be an adverbial phrase and the other a prepositional phrase. The third criterion is the one that makes apposition similar to coordination, because it considers that both elements are on the same level. However, instead of classifying the elements as heads, more in accordance with his definition of coordínate constructions, Hockett mentions the existence of an attributive relation between both elements, and so, he contradicts himself. As a consequence of this, his definition of apposition is not very clear.1 The fourth criterion, which establishes that both constituents may refer to the same entity, that is to say, that they must be coreferent, points out the main difference between apposition and coordination. In a coordinate construction the constituents are not coreferent, they lead us to different extralinguistic entities; while in an appositional construction, both elements refer to the same entity. The combination of semantic and formal criteria is also present in H. Sopher's definition of apposition (401-12). This linguist, unlike Hockett, considers that the appositional elements may belong to different syntactic classes. He also considers that Apposition in English 85 apposition differs from both subordination and coordination. However we can also observe a contradiction in this grammarian's definition, due to the fact that when he refers to the appositional elements, he speaks of head group and appositional group, which, in a way, implies subordination. For Sopher, the elements in apposition constitute a functional unit. Both are on the same grammatical level. If we omit one element and leave only the other element, the utterance in which they are inserted does not change. Both elements are interchangeable and there is a semantic relation of coreference between them. Therefore, when they are functioning as subject, they concord with the verb in singular. Lastly, we can use between both elements a connector such as that is, namely. In this way, the existing differences between apposition and other syntactic relations in the linguistic system are established. Sopher includes in his corpus examples of loóse apposition and cióse apposition. However, like Zandvoort, he makes a difference between structures, such as my brother the explorer and others like King George the Fifth. He takes an important step forward by stating that the latter structure is not an apposition, but instead a kind of modification in which the first element, King, is the head, the second element, George, is the modifier of the head and the Fifth modifies both King and George,2 Cióse apposition has caused much more controversy than loóse apposition when it comes to analysing it. Questions such as which of the two elements is the identifying one?, what kind of relation is there between the two elements?, are there two heads or a modifier and a head? have had different answers. So, E. Haugan in his study "On Resolving the Cióse Apposition" (165-70), unlike Sopher, considers that the first element is the modifier that helps to identify the second element. A different point of view is held by Hockett (102), who asserts that both elements may be heads and attributes, that is to say, the first element can identify the second, just as much as the second element can identify the first. The existing subordínate character between the members of a cióse apposition has led some experts not to consider these structures examples of apposition. So Burton-Roberts draws boundaries on the field of apposition, and considers that it exits only in those cases of loóse apposition. We are in agreement with his opinión when he affirms that cióse apposition presents a structure of modifier followed by head, that is, we think that it has an adjectival purpose and so it is not apposition. R. Quirk et al. put forward syntactic and semantic criteria in their definition of apposition. Their criteria, as we shall see in the following, are basically the same as those used by Sopher: "(1) each of the appositives can be omitted without affecting the acceptability of the sentence; (2) each fulfils the same syntactic function in the resultant sentences; (3) it can be assumed that there is no difference between the original sentence and either of the resultant sentences in extralinguistic reference" (1302). Quirk et al. consider that apposition has similarities with both coordination and subordination. Even though these grammarians affirm that "[a]pposition resembles eoordination in that not only do coordínate constructions also involve the linking of units of the same rank, but the central coordinators and and so may themselves occasionally be used as explicit markers of apposition" (1301-1302), later on they mention that in some cases, in the type of apposition they cali partial (the financial expert Tom Timber), the subordínate character of one of the elements is highlighted due to the fact that it is the only one that can be omitted without altering the utterance (1305). 86 Revista Alicantina de Estudios Ingleses Following Quirk et al.'s study, Ch. F. Meyer centres on apposition, basing his remarles on syntactic, semantic and pragmatic characteristics. This linguist, like Quirk et al., includes more semantic relations between the elements in apposition. Semantically, that relation, according to Meyer, may be co-referential ("my father, John"), hyponymous ("a tree, an oak tree"), synonymous ("a priest, a man of the clergy") and attributive ("my sister, a tax accountant") (103). The criteria established by Meyer to define apposition are the following: "semantic constraint Ul and U2 are coreferential, hyponymous, synonymous, or attributively related; pragmatic constraint. U2 suplies new information about Ul; syntactic constraint: either Ul and U2 are juxtaposed or they must be able to be juxtaposed without the resulting sentence becoming unacceptable" (120). Using these criteria an ampie number of linguistic structures are considered to be appositions. We have used Quirk et al.'s criteria to form our corpus because they were more suitable for it. But, we have restricted the corpus we are going to analyse to loóse apposition only, because, as we have previously mentioned, we believe that a structure in a cióse apposition has an adjectival character and so, it cannot be considered an apposition. III Syntactic characteristics (a) Syntactic class of the elements in an appositive relation It has been repeatedly pointed out that an appositive relation may be found in linguistic units higher than the phrase. However, grammarians are in agreement that it is a type of relation that takes place mainly between phrases. And they make their definition even more specific by stating that it is to be found for the most part in noun phrases. After analysing our corpus we agree with this, noun phrases are much more common in our corpus than other grammatical units. Those noun phrases may have a common noun (example 1), a proper noun (example 2), or a personal pronoun (example 3), demonstrative (example 4) or indefinite (example 5) as heads: 1. Denzil Jones, the accountant, poked his head round it. (GMA 14) 2. ErrolPs daughter, Diana ... had come to England to live with her aunt. (WM 98) 3. The sort of woman she, Mrs. Pursey, should not be asked to admit into her presence. (HL 84) 4. This—Innocence—was the first dead person he had ever encountered. (GMA 73) 5. Eyerything about her seemed exaggerated: her height, the length of her extraordinary fíngérs (HL 70) Table A on the next page shows the different syntactic classes that the elements in apposition present in our corpus. As one can see, the appositive relation is found in 95% of the examples of noun phrases and only 5% in other types of phrases. We can also observe that the elements in an appositive relation do not have to belong neccesarily to the Apposition in English 87 same syntactic class, although in most cases they do. We shall quote some of the examples in our corpus: 6. She behaved well...: quietly, politely, venturing little. (HL 85) 7. He was getting too large: fifteen and a halfsíone at the last weigh-in. (GMA 42) 8. He was still there, in the same oíd fíat, that Julie had gone. (ChT 136) Syntactic Class NP + NP NP + PreP NP + Clause AdjP + AdjP AdjP + NP AdvP + AdvP AdvP + PreP PreP + PreP PreP + AdverbP Clause + NP Total NP = Noun Phrase PrepP = Prepositional Phrase AdjP = Adjective Phrase AdvP = Adverb Phrase GMA 71 2 1 1 2 77 WM 355 1 5 1 1 4 2 368 HL 58 2 1 62 ChT 80 1 1 2 85 RD 83 1 1 1 1 1 86 Total 647 11 1 3 1 2 4 1 7 678 % 95% Table A Syntactically, the structures which the elements in apposition present may be simple or complex. The head of the noun phrase may stand alone, as in examples (2) and (4) or accompanied by other elements, such as determiners and modifiers, as in the following example: 9. His wife, a tiny smiling woman with a creamy caramel sfán and huge dangling earrings, was in a lacy blouse.... (GMA 238) One striking formal characteristic of the second appositive element is the absence of a determiner before the common noun. There are twenty-six examples in our corpus in which the common noun in the second element is not preceded by a determiner. This lack of a determiner is commented on by Quirk et al.: "In a type of partial apposition expressing attribution (particularly a unique role), an article (definíte or indefinite) is absent from the defining appositive" (1313). In all the examples of our corpus, a semantic relation of attribution exists between the two elements. The first element is generally a proper noun and the second is a common noun in singular. In most of the examples (19 out of 26), the 88 Revista Alicantina de Estudios Ingleses second element is a noun which indicates blood ties and postmodification is a common characteristic in all of them, as example (10) illustrates: 10. Michael Lafone, a fierce womanizer with an eyeglass, who was briefly and disastrously married in Kenya to E. Byng, daughter ofthe Earl ofStratford. (WM 33) The complexity ofthe appositive elements is not only because there is an accumulation of determiners, premodifiers and postmodifiers but also because they consist of two or more noun phrases that are juxtaposed (example 11) or coordinated (example 12). In other cases that complexity is due to the fact that there is an accumulation of appositions (example 13): 11. His wife laughed scornfully, knowing him to be burdened with responsabilities—houses, children, professionalstanding that he couldnot shed. (HL 85) 12. And there were two other guests for lunch: Juanita Carberry, Junes fifteen year oíd step daughter, and her governor, Isabel Rutt. (WM 88) 13. He linked the ñame with the person who was Fanshave's wife: Mrs. Chloe Fanshawe, wife to the Deputy. (GMA 29) In (12), the second element is formed by two noun phrases joined by the conjunction and; at the same time, each of them forms an apposition with the noun phrase which immediately follows it, that is, we have an apposition within an apposition. In (13), the third element is in apposition with the second element and both the second and the third afe in apposition with the first element. However, we have to mention that these complex appositions are not very common in our corpus. The author, by using an accumulation of appositions, gives a more complete description of the character in a concise, brief and almost telegraphic way. (b) Syntactic function ofthe elements in an appositive relation The appositive elements in our corpus have the following functions in the clause in which they are inserted: subject, direct object, indirect object, subject complement, object complement, prepositional object and adverbial. Table B shows the percentages of these functions. Syntactic Function non-existential subject existential subject direct object indirect object prepositional object GMA 19 1 14 23 WM 136 7 56 3 134 HL 28 2 11 19 ChT 25 5 17 28 RD 23 21 31 Total 228 15 119 3 235 % 33% 2% 17% 0.45% 35% Apposition in English 89 subject complement object complement adverbial Total 18 1 76 30 2 364 1 1 62 6 2 83 9 1 85 64 1 6 671 9% 0.15% 0.90% Table B Both elements in apposition have always the same function, that is to say, they form a functional unit. As has been suggested before, the two appositive elements are generally noun phrases, and so, as Meyer states (Apposition in Contemporary English 34-35), their functions are, in most cases the same as those of noun phrases. Table B shows that the functions of existential subject together with those of prepositional object, direct object, indirect object, subject complement, object complement and adverbial, which promote end-weight, are more numerous, 66% to be more precise, than the function of nonexistential subject. So the two appositive elements are in most cases placed after the verb phrase of the clause in which they are inserted. Therefore, we agree with Meyer when he says that: "In addition to having functions associated with noun phrases, appositions had functions associated with positions in the sentence, clause or phrase that promoted Quirk et al's principie of end-weight" (Apposition in Contemporary English 35). According to that principie, complex linguistic units which the speaker or the writer want to emphasize are placed towards the end of the utterance. We shall quote some of the examples in our corpus, indicating the function of the apposition: 14. Mrs. Woodhouse had moved to Hastings in the mid- 1960s and met there, in the Queen'sHotel, on a windy day in late July. (WM230). Adverbial. 15. Now her boss was Mr. Middlebrook, a tall, thin man.... (ChT 167). Subject complement. 16. . . . the little man was hopelessly inept, had never got to grips with the English language, and was cordially detested by Moses, the Morgan's cook. (GMA 64). Prepositional object. 17 Morgan had bought theirpriciest wine, a sweetish highly scentedPiesporter.... (GMA 105). Direct object. 18. For there was love there, love between mother and daughter... .(HL 48). Existential subject. 19. Harriet, the life-loving young divorcee, had first met Broughton at Highclere.... (WM 61). Non-existential subject. In our corpus, the number of appositions which function as non-existential subject, a function which does not promote end-weight, is high, 228 examples to be precise. Of those, only 9% are placed after the verb. Table C on the next page shows the order of the apposition and the verbal predícate in the clause in which they are inserted. As we can observe, 82% of the examples are placed after the verb and are juxtaposed (example 20): 20. And my kindhosts, Mr. and Mrs. Taylor, would never, I am certain, have knowingly put me through what I have just endured. (RD 180) 90 Revista Alicantina de Estudios Ingleses Only 9% of the examples are placed after the verb, as we have mentioned before. This inverse order of Verb + Subject is found in those cases where the subject is a long sequence and the predícate is short (example 21), or in a direct language clause (example 22), or in structures which are mainly found in an informal register (example 23): 21. Among them was Jack Soames, an oldEtonian wno was thirty-two when he arrived in 1920. (WM 22) 22. He was the most boring man in the world, said his neighbow on lake Naivasha, a trophy-hunting Austrian calleé Barón Knapitsch. (WM2A6) 23. Here was Stephen now, afoot soldier in this army ofexperts.... {ChT 81) The appositive elements are, in most cases, placed one after the other;however, the contínuity of the elements is interrupted by the verbal predícate in some examples, as can be observed in Table C. In other examples, the discontinuity is due to the fact that an apposition marker, such as, that is, namely, or a parenthetical phrase or clause is placed between both elements. Table D shows the existing proportion between juxtaposed and unjuxtaposed elements in our corpus. Order ofthe appositive elements in subject position lst el. + 2nd el. + VP lst el. + VP + 2nd el. VP + lst el. + 2nd el. Total el. = element VP = verbal predícate Juxtaposed and unjuxtaposed Juxtaposed Unjuxtaposed Total GMA WM HL 13 116 24 3 6 3 3 13 19 135 27 elements in our corpus GMA WM HL 68 355 53 9 13 9 77 368 62 ChT 15 5 5 25 ChT 74 11 85 RD 19 3 22 RD 79 7 86 Total 187 20 21 228 Total 629 49 678 % 82% 8% 9% Table C % 92% 7% Table D As can be seen, in the majority of cases the appositive elements are juxtaposed, in 92% ofthe examples. In other cases, all the predícate or part of the predícate is placed between both elements. In this way, the contínuity of the apposition is broken. This discoritinuity Apposition in English 91 takes place when the apposition, functioning as a subject, is a long sequence and the predícate is very short (example 24): 24. The requirements were simple: an unobstructed view of a changing landscape, however dull, and freedom from the breath of other passengers, their body warmth, sandwiches and limbs. (ChT 50) We agree with Meyer when he says that this order of the constituents is based on the principie of end-weight. As Quirk et al. assert, some elements may be interposed in a noun phrase "to achieve a stylistically well-balanced sentence in accordance with the norms of English structure" (1398). That discontinuity is also due to what Quirk et al. ñame endfocus. Thus, the second element is given more importance because it is placed at the end of the clause. Therefore, in example (24), the second element is emphasized not only because the adverb specially is used, but also because the second element women is placed after the predícate. 25. People love this one, especially women. (HL 27) In some examples there is only one order possible in the structure of the utterance. In example (26), the two elements may not be juxtaposed: 26. The car keys were in his hand, his wallet was smug in his inside pocket—the equipment of audulthood. (ChT 121) However, for the second element to be given more importance than the others, it does not necessarily have to go in final position in the clause. With a simple change in the order of the elements the author emphasizes the second element more, as in example (27): 27. Had she said that he was cross and peevish on the night of the murder, while they were drinking brandy—June and he—in the Muthaiga bar? (WM 90-91) Semantic and pragmatic characteristics The semantic classifications of apposition made by Quirk et al.3 and Meyer4 do not differ greatly. In both of them the specificatión of the elements is mentioned, that is to say, when one of the elements is more, less or as specific as the other. We have found that Meyer 's classification is more suitable for the analysis of our corpus and that is why we employ it in this paper. Table E on the next page shows the different semantic classes of apposition in our corpus. As Table E illustrates, the highest percentage of examples, 50% to be more precise, are those appositions in which the second element is more specific than the first. We have to point out in the more specific group the widespread use of examples belonging to the semantic classes of appellation (example 28) and identification (example 29): 92 Revista Alicantina de Estudios Ingleses 28. My eider brother, Leonard, was kílled during the Southern Afincan War while I was still a boy. (RD 40) 29. Morgan realized, with some alarm, as he approached that this—Innocence was the first dead person he had ever encountered... (GMA 73) In example (28) the two elements refer to the same person. There is a semantic relation of coreference, but the second element, Leonard, is more specific than the first. The proper noun specifies the ñame of my eider brother. As Quirk et al. say, it is a naming relation (1309). In example (29), the second element, Innocence, identifies the first, this, a demonstrative pronoun. Therefore, the second element is more specific than the first. Less numerous in our corpus are the examples belonging to the semantic classes of particularization (examples 30 and 31) and exemplification (example 32). The referents are not the same and in both classes an element is included within the other: 30. It was noticeable, moreover, that the Wakefields—Mrs. Wakefield in particular—were themselves by no means ignorant of the traditions in our country . . . (RD 122-23) 31. It does not seem to have occurred to anybody that Joss Erroll might have been murdered by a woman—a wotnan who used a gun in the Gare du Nord and wounded the man she married later. (WM 219) 32. Morris called only eight witnesses, including Broughton and the loyal Major Pembroke against the prosecution's twenty. (WM 108) In (30) the Wakefields refer to both, the husband and his wife, and Mrs. Wakefield stands out by being used as the second element in the appositive structure. In (31) the two elements are in a hyponymous relation; in this case, unlike the other examples of this class, the head of the second element is the same lexeme as the head of the first element and, in our corpus, it is always accompanied by postmodifiers. In this way, a certain part or a characteristic of something/somebody just mentioned is made to stand oüt by the author. As Quirk et al. observe: "the intention may be rhetorical to provide a climatic effect by repetition and expansión of the first noun phrase" (1312). Like Quirk et al., Joan N. Bitea in her study "An Attempt at Defining Apposition in Modern English" also mentions this effect of apposition: "Rhetorical effects are also obtained by means of developing appositions: appositions which constitute both formally and semantically developments of their correlatives" (472). In example (32), the second element, "including Broughton and the loyal Major Pembroke," specifies some of the witnesses referred to in the first element, and so, it is more specific. Semantic classes of apposition More specific a) Appellation b) Identification c) Particularization GMA 11 21 WM 114 55 5 HL 13 11 10 ChT 10 39 3 RD 10 20 14 Total 158 146 32 % 23% 21% 4.7% Apposition in English 93 d) Exemplification Total a+b+c+d Less specific a) Characterization Equally specific a) Paraphrase b) Reorientation Total a+b 32 44 1 1 6 180 188 34 16 1 11 12 52 33 2 46 39 8 344 320 1 12 13 1% 50% 47% 0.15% 1% 3% Table E In 47% of the examples, those belonging to the semantic class named characterization, the second element is less specific than the first, it describes a particularity of it (example 33): 33. When the letter arrived it was not from Charles Darke, the young sénior editor profiled in the Sunday newspaper.... (ChT 29) In the preceding example, the second element is a noun phrase which attributes a characteristic to the first element. And so, the relation between them is an attributive relation, the first element is a proper noun and is more specific than the second. We have not found a lot of examples in our corpus in which the first element is as specific as the second element. As Table E indicates, we have only found examples belonging to the semantic classes of paraphrase (example 34) and reorientation (example 35): 34. The here and now, the quotidian, was beginning to acquire substance. (HL 37) 35. Mrs. Pusey, thatpinnacle offeminine chic, that arbiter of tosté, that relentless seeker after luxury foods, that charmer of multitudes, is seventy-nine! (HL 103) In (34) the second element paraphrases the first. In (35) the second element describes several characteristics of Mrs. Pusey, the author refers to Mrs. Pusey again by using all those noun phrases, both elements refer to the same person, there is a semantic relation of coreference. By using apposition the author foregrounds a certain aspect of the discourse, an aspect which has great importance in the interchange between the interlocutors. Pragmatically , the second element is generally an explanation of the first. It usually adds information that the speaker or narrator consider necessary to clarify what they have previously expounded. In some cases it avoids possible ambiguity. E. Koktova's words summarize some psychological properties of apposition: "From a psychological viewpoint, apposition (or, more exactly, the apposed elements) should be viewed as the speaker's commentary, as an afterthought, as an implicit predication, or as a secondary information, of a sentence-simply as 94 Revista Alicantina de Estudios Ingleses a message which deflects from the mainstream of communication and which should be kept distinct from the proper assertion (main information) of a sentence" (40-41). In our literary corpus, we have observed that the number of appositions in the dialogues is noticiably less than in the rest of the text, we have found only 17 examples. This is due to the fact that, in a dialogue, the information that the second appositive element gives about the first is no necessary because the interlocutors are generally aware of information implicit in the discourse. As we have previously mentioned, in our corpus apposition is used in most cases to ñame, identify and characterize something or somebody. In this last case, although the author uses a proper noun and identifies the character totally, he still thinks it necessary to make it more palpable. In this way, the reader is aware of exactly who the author is referring to. Among the literary works which make up our corpus, we have to point out in particular J. Fox's Who KilledJohn Erroll? White Mischief, because of the high number of appositions found there, 368 examples out of the total 678 examples that we have analysed. In our opinión, there are two main reasons for the use of apposition in this literary work. On the one hand, the narrator is just a kind of detective who reports a murder that took place in África. As a result, the narration resembles press writing and is, as a consequence, different from the other works in our corpus. In press reportage appositions are much more common, as Meyer states (Apposition in Contemporary English 100), than in other genres, because information in a newspaper has to be explained very accurately as the reader and the journalist do not share much knowledge about what is being reported. On the other hand, J. Fox gives life to a series of people, the majority belonging to the same community who, in one way or another, have been involved in the murder. The number of characters is so high and their ñames are changed so often throughout the book that the reader could have difficulty in following the plot. The fact that the author includes a list with the ñames of the characters at the end of the book indicates that he is conscious of such a difficulty. By adding information through the use of the second appositive element, the author always makes clear who exactly he is referring to. Consequently, in this work we also find a lot of appositions with more than one element (examples 34 and 35): 34. One of the few women who didn't see him as the epitome of sexual attraction was Dushka Repton, a Russian beauty married to a settler farmer, Gruy Reptan, who was insanely jealous of his wife and eventually died of drink. (WM 36) 35.LadyAltrincham (thenLady Grígg, wife ofthe Governor) put Idina on her black list. (WM 31) IV Conclusions In the preceding pages we have carried out a descriptive analysis of apposition based on a literary corpus from a syntactic, semantic and pragmatic point of view. We have analysed the characteristics which differentiate apposition from other relations within the Apposition in English 95 linguistic system, and we have observed that it is a type of relation which is mainly found among juxtaposed noun phrases. These noun phrases have the same function within the clause in which they are inserted. It is mainly used to characterize, ñame and identify characters and everything the author considers necessary. In this way, the author transmits information which helps him not only to give detailed yet concise descriptions, but also, in some cases, to help the reader to follow the story. Notes 1. N. Burton-Roberts (393) mentions this contradiction in his analysis of apposition. 2. Donald W. Lee (268-75) analyses these structures in the same way as Sopher. 3. Quirk et al. (1308) classify apposition in the following semantic classes: most appositive (a) equivalence: (ai) appellation: that is (to say); (aii) identification: namely; (aiii) designation: that is to say; (aiv) reformulation: in other words; (b) attribution (non-restrictive relative clause); (c) inclusión: (ci) exemplification: for example, say; (cii) particularization: especially. 4. Ch. F. Meyer (74) classifies apposition in the following semantic classes: more specific: identification, appellation, particularization, exemplification; less specific: characterization; equally specific: paraphrase, reorientation, self-correction. Works Cited Bitea, Joan N. "An Attempt at Defining Apposition in Modern English." Revue Romaine de Linguistique 13.4 (1977): 453-77. Boyd, William. A GoodMan in África. Hardmondsworth: Penguin, 1981. Brooker, Anita. Hotel du Lac. London: Triad Grafton Books, 1985. Burton-Roberts, N. "Nominal Apposition." Foundations ofLanguage 13.3 (1975): 391-19. Curme, G. O. English Grammar. New York: Barnes and Noble, 1947. Fox, James. White Mischief. Hardmondsworth: Penguin, 1982. Haugan, E. "On Resolving the Cióse Apposition." American Speech 28.3 (1953): 165-70 Hockett, Ch. F. "Attribution and Apposition." American Speech 30.2 (1955): 99-102. Ishiguro, Kazuo. The Remains of the Day. London: Faber and Faber, 1989. Jespersen, Otto. Analytic Syntax. New York: Holt, Rinehart and Winston, 1969. Koktova, Eva. "Apposition as a Pragmatic Phenomenon in a Functional Description." UEA Papers in Linguistics 23 (1985): 39-79. McEwan, Ian. The Child in Time. London: Pan Books, 1988. Meyer, Ch. F. "Apposition in English." Journal of English Linguistics 20.1 (1987): 101-21. .Apposition in Contemporary English. Cambridge: Cambridge UP, 1992. Quirk, Randolph, et al. A Comprehensive Grammar ofthe English Language. London: Longman, 1985. Sopher, H. "Apposition." English Studies 52 (1972): 401-12. Zandvoort, R. W.A Handbook of English Grammar. London: Longman, 1977. coordination Sentence Combining: Coordination and Subordination by Jennifer Sentence variety is one important aspect of writing, and is one of the four areas assessed on the essay test at the end of the semester. Two ways you can successfully combine short sentences is by subordination or coordination. Correctly done, your ideas will flow smoothly. An independent clause is a group of words that can stand alone as a sentence. Join two of these together with a comma and one of the following coordinating conjunctions: for - gives a reason, e.g., I like going to school, for the classes are not too difficult. and - adds a fact or condition, e.g., I like going to school, and I like many of the teachers. nor - means not. Remember to invert the word order following this conjunction, e.g., I am not going out, nor am I going to lie down to rest. but - shows a contrast, e.g., I want to go out and have fun, but I don't want to be out too late. or - gives a choice of alternatives, e.g., We could go to a movie, or we could go for a walk. yet - like "but", this word also shows a contrast, e.g. I want to go out with my friends, yet I noticed there's a really good movie on TV tonight. so - gives a result, similar in use to "therefore", e.g., It's raining out, so I will take an umbrella. The above coordinating conjunctions make an acronym: fanboys. An acronym is a word made up of the first letters of others words. "Fanboys" is an easy way to remember these seven coordinating conjunctions. Subordinating conjunctions also join ideas together. But where fanboys joins two equal clauses together, subordinating conjunctions impose a different relationship, making one clause dependent on the other. There are many different subordinating conjunctions. Some of the most common are as follows: if, even if, provided that - gives a condition, e.g., I will go with you if you need some help. although, even though, though - gives contrast, e.g., Although you need help, I am too busy tomorrow morning to go with you. because, since - show cause, e.g., I will cancel my appointment tomorrow morning and go with you since you really need some help. after, before, when, whenever, while - give a time, e.g. I am going shopping after I leave you downtown tomorrow. where, wherever - show place, e.g., We can go whenever you'd like to. in order that, so that - show purpose, e.g., I am going to class so that I can ask the teacher what will be on the test. Using coordinating and subordinating conjunctions, rewrite the following paragraphs for better sentence variety, and show them to a teacher. Paragraph 1 I am going to the gym. I need to take the bus. My car has broken down. I am going to put my runners, shorts, swimsuit, a towel and a top into my bag. I will jog to the bus stop. It will be part of the my workout. I will get to the gym by 11 am. I will do the bike for 15 minutes. I will lift weights for 30 minutes. I'll stretch for 10 minutes. I will go into the pool. I will go into the hot tub. I will go into the sauna several times. I will get changed. I will go home. Paragraph 2 I went shopping yesterday. I was having friends for dinner. I needed a big roast of beef, two chickens, and twelve dinner rolls. I also bought two heads of romaine lettuce, cherry tomatoes, green pepper, celery, radishes, an orange and fresh garlic. I added baking potatoes, fresh baby carrots, snow peas, corn niblets, and cauliflower. I stopped by the florist. I picked up two bunches of deep red roses, baby's breath, and some salal. I went to the bakery for a blueberry-chocolate cheese cake. I drove to my friend's to pick up a pair of beautiful pink-and-white beeswax candles. I was pleased with my purchases. I went home to cook and decorate for that evening's dinner. "The Principles of Coordination and Subordination" by Johnie H. Scott, Assistant Professor Pan African Studies Department - California State University, Northridge Introduction Coordination: linking together words, groups of words (clauses), or sentences of equal type and importance, to put energy into writing. Coordinating Conjunctions: and, or, nor, for, but, so, yet, either/or, and neither/nor. Two principles to keep in mind: By combining words and groups of words, you avoid repetition that steals energy from what you write; and By combining whole sentences, you reveal the relationships between the thoughts. Example: Over the past decade many African American students have chosen to complete their formal education at Southern colleges and now in the city of Atlanta there is a major educational center built expressly to accomodate this upsurge of interest in the New South. (Two main clauses are given equal emphasis and connected by the coordinating conjunction and ) Subordination: clearly empashizes which words, groups of words (clauses), or sentences are the most important in the writing. Subordinate Conjunctions: Takes into account five (5) factors -- (1) Time: when, after, as soon as, whenever, while, before; (2) Place: where, wherever; (3) Cause: because, since, in order that, so that; (4) Contrast/Concession: although, as if, though, while; and (5) Condition: if, unless, provided, since,as long as. Example: Because CSUN is located in the San Fernando Valley, the university has become very attractive to students living in the inner city who want to stay close to home and yet not face the pressures of city life. (Dependent clause introduced by the subordinating conjunction because; independent or main clause begins with the university) Caveat: One wants to avoid faulty or excessive coordination. Faulty coordination: gives equal emphasis to unequal or unrelated clauses. Example: The African American playwright August Wilson has won two Pulitzer Prizes for drama, and he now lives in Seattle, Washington. The clause he now lives in Seattle, Washington has little or no connection to The African American playwright August Wilson has won two Pulitzer Prizes for drama. Therefore, the clauses should not be coordinated. But you, writer, may want to include this information in the paragraph because it is interesting and perhaps even important, even though it does not pertain directly to the main idea of the paragraph. Placing he now lives in Seattle, Washington might detract from the paragraph's unity. We can revise faulty coordination by putting part of the sentence in a dependent clause, modifying phrase, or appositive phrase (an appositive is a noun or pronoun -- often with modifiers -- placed near another noun or pronoun to explain, describe, or identify it). CSUN's Square, a hangout for its African American student community, has been quiet of late. (A hangout for its African American student community describes CSUN's Square); or My sister Tiyifa lives in Colorado Springs. (Tiyifa identifies sister ) Typically, an appositive follows the word it refers to, but it may also precede the word: A very inspirational tale of courage and honor, Glory is based on actual accounts of the all-black 54th Regiment during the American Civil War. (A very inspirational tale of courage and honor describes Glory) Resuming with means of correcting faulty or excessive coordination, we note the following examples: The African American playwright August Wilson, who now lives in Seattle, Washington, has won two Pulitzer Prizes for drama. (Dependent Clause) The African American playwright August Wilson, now Seattle-based, has won two Pultizer Prizes for drama (modifying phrase) The African American August Wilson, a Seattle playwright, has won two Pulitzer Prizes for drama (An appositive phrase). We can go further by noting that when a single sentence contains more than one clause, the clauses may be given equal or unequal emphasis. Clauses given equal emphasis in one sentence are coordinate and should be connected by a cooordinating word or punctuation. Clauses given less emphasis in a sentence are dependent, or subordinate, and should be introduced by a subordinating word (conjunction). Rules to Remember Concerning Faulty Subordination: There are three (3) rules to keep in mind with respect to faulty or excessive subordination in writing: Faulty subordination occurs when the more important clause is placed in a subordinate position in the sentence or when the expected relation between clauses is reversed. Example: Japanese-made cars are popular with American consumers although their import poses at least a short-term threat to the livelihood of some American workers (In an essay or composition about he problems of the American worker this sentence would take attention away from the worker and incorrectly emphasize Japanese-made cars.) Correct faulty subordination by changing the position of the subordinating word or phrase; Example: Although Japanese-made cars are popular with American consumers, their import poses at least a short-term threat to the livelihood of some American workers. Keep in mind that excessive subordination occurs when a sentence contains a series of cluses, each subordinate to an earlier one. To correct excessive subordination, break the sentence into two or more sentences or change some of the dependent clauses to modifying phrases or appositives. Example: LaTosha Robinson, who was a San Francisco-native who lived in the University Park Apartments, enjoyed those special moments when a group of students who also came from Northern California visited her dorm, which was lonely for most of the school year. This sentence is very confusing for the reader. The writer seems to have added information as it came to mind. To correct excessive subordination, note the following: LaTosha Robinson, a San Francisco-native, lived in the University Park Apartments. Because her dorm was lonely most of the school year, she enjoyed those special moments when a group of students who also were from Northern California would visit. One dependent clause, who was a San Francisco-native, has been changed to an appositive. A second dependent clause, who lived in the University Park Apartments, is now the predicate of the first sentence. These changes make the sentence more direct. The subordinator of the third dependent clause has been changed from which (identification) to because (cause) to show clearly the connection between the loneliness of the dormitory and LaTosha's enjoyment of those special visits. Questions Construct (9) grammatically correct and balanced sentences that use each of the coordinating conjunctions. Underline the coordinating conjunction in each sentence. Construct five (5) sentences that reflect the factors for subordinating conjunctions. Underline each subordinating conjunction and, in parenthesis at the end of the sentence, indicate which factor was used (i.e., Time ) Key Terms Coordination Subordination Faulty Coordination Coordinate Appositive(s) Predicate Dependent clause(s) CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 1 * For questions and comments, thanks to Anna Maria Di Sciullo, Jan Koster, Yordanka Kavalova, Jan- Wouter Zwart, Henk van Riemsdijk, Hana Skrabalova, Hans Broekhuis, Anneke Neijt, Olaf Koeneman, Janneke ter Beek, Herman Heringa, Marlies Kluck, and the anonymous reviewers of the Canadian Journal of Linguistics. Asymmetric Merge and Parataxis MARK DE VRIES University of Groningen Abstract. I argue that syntactic phrase structure encodes three major asymmetries. The first represents the asymmetry between mothers and daughters that is called dominance, i.e. syntactic hierarchy. The second is the selectional asymmetry between sisters, which is translated into precedence in the phonological component. The third, called ‘behindance’, is an alternative for dominance, and represents parataxis. Parenthesis, coordination and apposition are analyzed on the basis of behindance. In our derivational model of grammar it is defined as a special type of inclusion that blocks c-command. It follows that parenthetic material can neither move toward the matrix, nor be bound by a constituent from the matrix. The syntactic asymmetry between first and second conjuncts is established theoretically and empirically in a new way. 1. INTRODUCTION AND OVERVIEW Parataxis is the equipollent ranking of clauses or phrases, of which coordination is a canonical exemplar, (1a). In contrast, hypotaxis is the unequal ranking of clauses or phrases, of which subordination is a canonical exemplar (1b). And while hypotaxis is associated with a syntactic hierarchy, parataxis is not, or at least not in the same way. (1) a. Coordination as parataxis: The woman was standing and the man was sitting. b. Subordination as hypotaxis The woman was standing because the man was sitting. But the notions of parataxis and hypotaxis comprise much more than coordination and subordination (Van Es and Van Caspel 1975; Quirk et al. 1999: section 13.2; Schelfhout et al. 2003a). For instance, the subject-predicate relation can be viewed as an instance of hypotaxis, since it involves the unequal ranking of two phrases. Other examples of parataxis—understood as the equipollent ranking of clauses or phrases—include parenthesis (2) and apposition (3): (2) Parenthesis (comment clause, hedge, appended clause, etc.) as parataxis: a. He was walking, he said, toward the railway station. b. He asserted – and this is how all moralists speak – that the young are spoiled. c. These weapons are meant to wound, to kill, even. d. I told them, mistakenly, it turned out, that she had already left. (3) Apposition and appositive relative clauses as parataxis: a. She gave Joop, our friend, a present. b. She gave Joop, who is our friend, a present. With the exception of common coordination, a theoretical account of parataxis is CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 2 2 lacking in the literature, especially in the Minimalist Program. This gap reflects the fact that syntax is preoccupied with hypo- and hypertactic relations, which are usually modelled with hierarchical structures in the form of tree diagrams or constituent bracketing. How paratactic relations are to be represented remains unclear. In this context, the two questions I address here are: (i) How can the non-subordinative properties of paratactic constructions be represented in syntax? (ii) Can we generalize over coordination and other types of parataxis? Section 2 reviews the differences between hypotaxis and parataxis, and illustrates the link between apposition and coordination. It is concluded that, in addition to the relations of dominance and precedence, the relation of behindance is needed to describe paratactic phenomena. Section 3 discusses theoretical preliminaries such as the independence of dominance and behindance, and the properties of Merge. Section 4 presents the theoretical proposal, and its application to coordination and parenthesis. Section 5 argues that paratactic material is invisible to relations based on c-command. Section 6 concludes. 2. PARATAXIS AND BEHINDANCE After reviewing the properties of parataxis (section 2.1), I argue that apposition is a special case of coordination (section 2.2), and that that parataxis does not involve c-command (section 2.3). I propose that there is a common ground to all types of parataxis (section 2.4): this is what I call behindance, following Grootveld (1994). 2.1. Some properties of parataxis There are simple tests that distinguish coordination from subordination. For example, in Dutch and German, conjoined main clauses display V2, whereas subordinated clauses have V-final order. This is illustrated in (4) for Dutch: the verb las ‘read’ occupies a V2- position when it occurs in a conjoined clause (4a), but a V-final position when it is in a subordinate clause (4b). (4) Dutch: a. Karel keek televisie en Joop las de krant. Karel watched television and Joop read the newspaper ‘Karel watched television and Joop read the newspaper.’ b. Karel keek televisie omdat Joop de krant las. Karel watched television because Joop the newspaper read ‘Karel watched television because Joop read the newspaper.’ CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 3 3 While conjoined DPs generally appear in the same case-form, a subordinate DP has an oblique Case (determined by the preposition).1 This is illustrated in (5) for German: in (5a) the two conjoined DPs both bear nominative case (der Mann, die Frau); in (5b), the subordinate DP is marked for dative case (dem Hut). (5) German:2 a. Der Mann und die Frau sind unzertrennlich. the.NOM man and the.NOM woman are inseparable ‘The man and the woman are inseparable.’ b. Der Mann mit dem Hut ist zurück. the.NOM man with the.DAT hat is back ‘The man with the hat is back.’ Third, coordination is category-neutral in many languages (6). (But see Johannessen 1998:84ff and Haspelmath 2004 for a qualification.) This contrasts with subordinators, which typically select for a complement of a particular category: prepositions canonically select for nominal DP complements (7a); complementizers canonically select for clausal IP complements (7b). (6) XP and XP (where X = A, P, N, V, D, I, C, etc.) (7) a. [P DP] b. [C IP] Fourth, gapping is generally possible in coordinate structures (8a), but not in subordinated clauses (8b). (8) a. Bill bought a CD, and John _ a book. b. * Bill bought a CD because John _ a book. Fifth, it has been claimed that there is a valency difference between coordination and subordination (Van der Heijden 1999): a coordinator conjoins conjuncts, but a subordinator does not subordinate subjuncts, that is, there is no such thing as a first or second subjunct. Sixth, a conjunct or a part of it cannot move (Ross’s 1986 Coordinate Structure Constraint), but a subordinate phrase can: (9) a. * And who did you see Mary _ ? b. * Who did you see Mary and _ ? c. * Who did John kiss Anne and Mary hit _ ? 1 But see Johannessen (1998) for discussion of syntactically unbalanced coordination. 2 The following abbreviations are used: Co=conjunction, CoP = coordinate phrase, DAT = dative, NOM = nominative, ParP = parenthetic phrase, PRON = pronoun. CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 4 4 (10) a. In which garden did you rest _ ? b. Which garden did you rest in _ ? c. Who did you say that Mary hit _ ? In short, there is a clear distinction between coordination and subordination. But see Van der Heijden (1999) and Culicover and Jackendoff (1997) for discussion of non-parallel coordination and insubordination, which fall somewhere between the canonical cases of coordination and subordination discussed here. Let us turn from coordination to other types of parataxis, such as parenthesis and apposition, which have the following properties: (11) a. They are not selected by any part of the matrix clauses. Therefore they cannot have a restrictive meaning, but are interpreted as additional information. b. They are phonologically marked by a low intonation. They interrupt the intonation contour of the matrix, but do not affect it. c. (i) They are linearly integrated with the matrix; (ii) but they are not part of the syntactic hierarchy in terms of c-command; (iii) yet, they can be added on a constituent level; and (iv) depending on the type, they can be inserted in various positions, but not just anywhere. The linear integration of paratactic clauses with a matrix clause (11c-i) implies that they are integrated into their host environment before spell-out, that is, in the overt syntax. Given a model of grammar where syntax feeds PF and LF, and given that a paratactic phrase is pronounced, it cannot be the case that this phrase is added at the discourse level, or some other level beyond LF, contra Fabb (1990) or Safir’s (1986) treatments of appositive relative clauses. Conversely, given the fact that a paratactic phrase has a meaning, it cannot be the case, either, that it is introduced only after PF. This last point is strengthened by (11c-iii)—paratactic clauses and phrases can be added on a constituent level—which holds of not only appositions and appositive relatives, but also of parentheticals. For example, in (12a), the scope of the parenthetical I think is ambiguous between the paraphrases given in (12b-i, ii). The first reading arises if the parenthetical is added onto a constituent of the matrix clause, namely his grandmother, as in (12c-i). The second reading arises when the parenthetical is added on at the level of the matrix clause, as in (12c-ii). (12) a. Tomorrow John will visit his grandmother, I think. b. (i) I think that it is his grandmother that John will visit tomorrow. (ii) I think that John will visit his grandmother tomorrow. c. (i) Tomorrow John will visit [[his grandmother], I think] (ii) [[I think that John will visit his grandmother] tomorrow] Moreover, as stated in (11c-iv), paratactic phrases are sensitive to syntactic structure, and so cannot be freely inserted into any position. This is illustrated in (13): the CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 5 5 parenthetical reporting clause he said can be inserted between two major constituents (13a), but not inside a major constituent (13b). (13) a. He was walking, he said, toward the railway station. b. * He was walking towards the, he said, railway station. The above establishes that paratactic material is sensitive to syntactic information, and so must be represented in syntax, to then be passed on to both the phonological component (PF) and the semantic component (LF). 2.2. The relation between apposition and coordination The relation between apposition and coordination is not immediately obvious. Consider the appositions in (14) from English and in (15) from Dutch. (14) English: a. John, our boss b. a nice present: a book by Golding c. John, a nasty liar d. the White House, or the house with the Oval Office (15) Dutch: a. Fik is een hond, en wel een poedel. Fik is a dog, and indeed a poodle ‘Fik is a dog, namely a poodle.’ b. Jan begaf zich naar beneden, en wel naar de kelder. Jan proceeded SE toward downstairs, and indeed to the basement ‘Jan went downstairs, namely to the basement.’ c. het Witte Huis, ofwel het huis met het ovalen kantoor the white house, or the house with the oval office ‘the White House, or the house with the Oval Office’ Relevant to the present discussion is the presence, in English, of the coordinator or (14d), and in Dutch of the coordinative expressions en wel ‘and indeed’ (15a-b) and ofwel ‘or’ (15c). The fact that coordinators sometimes occur in appositive constructions suggests that the latter may be a kind of coordination, as discussed by Quirk et al.: Apposition resembles coordination in that not only do coordinate constructions also involve the linking of units of the same rank, but the central coordinators and and or may themselves occasionally be used as explicit markers of apposition. (1999:1301–1302) As regards the semantics of apposition, observe that in all of the examples given in (14) and (15), the second part further specifies the first part. For English, in (14a) our boss further specifies John; in (14b) a book by Golding specifies what constitutes a nice present; in (14c) a nasty liar further specifies John; and in (14d) the house with the Oval CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 6 6 Office provides a further specification of the White House. Similarly, for Dutch, in (15a) een poedel ‘a poodle’ further specifies dog; in (15b) naar de kelder ‘to the basement’ further specifies downstairs; and in (15c) het huis met het ovalen kantoor ‘the house with the Oval Office’ specifies the White House. If apposition is indeed a type of coordination, then it can be claimed that, next to conjunction and disjunction, there is a third type of coordination, which designates specification (Kraak and Klooster 1968: chapter 11). Specifying coordination can be explicitly marked by a specifying phrase such as or rather, namely, that is, depending on the exact semantic subtype, but often the connection is asyndetic, that is, phonologically empty. Prosodically, it always triggers a comma and a low intonation of the second conjunct. Many authors have stressed the similarity between appositions and appositive relative clauses: Delorme and Dougherty (1972), Halitsky (1974), Klein (1977), Doron (1994), and Canac-Marquis and Tremblay (1998). Furthermore, Sturm (1986), Koster (2000a), and M. de Vries (2006a) explicitly advance a coordination analysis of appositive relatives. 2.3. Parataxis and c-command In her overview article on coordination, Progovac (1998) concludes that there is no clear evidence for a potential c-command relation between conjuncts; in fact, there is some counterevidence. For instance, in (16), the antecedent in the first conjunct cannot bind the anaphor in the second conjunct; given that binding requires c-command, this indicates that the first conjunct does not c-command the second one. (16) a. * Either Johni or a picture of himselfi will suffice. b. Serbo-Croatian: * Jovani i svojai zena su stigli. Jovan and self’s wife are arrived Intended: ‘Jovan and self’s wife have arrived.’ (Progovac 1998:3) The absence of a c-command relation between the first and second conjunct is confirmed in Dutch. In (17), the antecedent in the first conjunct (Joop) cannot bind the local anaphor zichzelf contained in the second conjunct.3 (17) Dutch: a. * Ik luisterde naar een gesprek tussen Joopi en zichzelfi. I listened to a conversation between Joop and SE-SELF. Intended: ‘I listened to a conversation between Joop and himself.’ 3 In (17a) hemzelf ‘him-SELF’, which is discourse-licensed, would be felicitous; in (17b) hem ‘him’ would be felicitous. CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 7 7 b. * Toon me Joopi of een foto van zichzelfi! Show me Joop or a picture of SE-SELF! Intended: ‘Show me Joop or a picture of him!’ Moreover, quantifier-variable binding between conjuncts is unacceptable: in (18a) the quantifier elke ‘every’ in the first conjunct cannot bind the pronoun zijn ‘his’ in the second conjunct; in (18b) the quantifier alle ‘all’ in the first conjunct cannot bind the pronoun hun ‘their’ in the second conjunct. (18) Dutch: a. * Elkei man en zijni vrouw gingen naar de film. every man and his wife went to the movies Intended: ‘Every man and his wife went to the movies.’ b. * Willen allei honderd kinderen en huni moeder naar voren komen? want all hundred children and their mother to the.front come Intended: ‘All hundred children and their mother, please advance.’ Progovac (1998) accounts for English examples like Everyi man and hisi dog left by quantifier raising; see also Sauerland (2001). If this is so, this leaves (18) unexplained, unless the application of quantifier raising is constrained by language-specific parameters. In certain contexts quantifiers can be discourse-related to a variable; for example Everyi rice-grower in Korea owns a wooden cart. Hei uses it when he harvests the crop (Sells 1985:3). If anything, these points strengthen the argument: quantifiervariable binding normally requires c-command, but under certain conditions the licensing of the variable can be rescued by another mechanism. By contrast, the hypotactically construed equivalents to (18)—using met ‘with’ instead of en ‘and’—are fine: (19) Dutch: a. Elkei man ging met zijni vrouw naar de film. every man went with his wife to the movies ‘Every man went to the movies with his wife.’ b. Willen allei honderd kinderen met huni moeder naar voren komen? want all hundred children with their mother to the.front come ‘All hundred children with their mother, please advance.’ The above establishes that material contained in the first conjunct cannot bind material contained in the second conjunct. It is also the case that movement from the second conjunct into the first conjunct is unacceptable (20). This is expected, given that movement to a non-c-commanding position is impossible. (20) * [Which mani and a friend of ti] are both handsome? Another context that requires c-command is the licensing of negative polarity items. As expected, the first conjunct cannot license a negative polarity item in the second CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 8 8 conjunct, as shown in (21). (See Hoeksema 2000 for discussion.) (21) He chased nobody and no/*any dogs. (Progovac 1998:3) We have seen that the second conjunct cannot be bound by the first conjunct and it cannot move into the first conjunct. Nor can a negative polarity item in the second conjunct be licensed by material in the first conjunct. A simple generalization captures this: there is no c-command relation between conjuncts (M. de Vries 2007). At first glance, Principle C effects such as (22) constitute a counterexample to this claim (Munn 1993). However, the same effect is observed across sentences (Progovac 1998) (22b). (22) a. *Hei and Johni’s dog went for a walk. b. *Hei finally arrived. Johni’s dog went for a walk. I conclude that whatever pragmatic principle rules out (22b) also accounts for the impossibility of (22a). This means that we can retain the generalization there is no ccommand relation between conjuncts. If so, then the relation between conjuncts is not hypotactic. 2.4. Behindance There remains the question of how to characterize the relation between conjuncts. One possibility explored by Progovac (1998) is to embed both conjuncts in a coordination phrase of their own: [[CoP ø XP] [CoP and YP]]. The position of the first conjunction is reserved for an initial coordinator (as in both…and…, for instance). However, this cannot be correct. First, initial coordinators are different from regular conjunctions (Johannessen 1998; Bredschneijder 1999; Hendriks and Zwart 2001; Skrabalova 2003; Hendriks 2004; M. de Vries 2005a; Johannessen 2005). Second, there is an asymmetry between conjuncts that can only be explained if there is a paratactic relationship; I return to this below. Another possibility, which is the one I pursue here, is an analysis in terms of parallel structures or three-dimensional structures. In such analyses, conjuncts, rather than being hierarchically organized, are viewed as situated behind each other. For instance, in (23), Mary is behind Bill, and Sue behind Mary. (23) Bill, Mary, went to the movies. and Sue This idea has been expressed in different ways by a number of authors, including Williams (1978), Goodall (1987), G. de Vries (1987, 1992), Mu’adz (1991), Moltmann (1992), Grootveld (1992, 1994), Te Velde (1997), and Van Riemsdijk (1998). The approach by Goodall (1987) and G. de Vries (1987, 1992) is based on set union of reduced phrase markers (in the sense of Lasnik and Kupin 1977), an idea which can be attributed to Rini Huybregts. In the resulting set, there can be elements (monostrings) that neither dominate nor precede each other; such an object is not representable by a conventional tree-diagram. Coordination is assumed to be sentential only, and the CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 9 9 combined sentences occupy different planes. Somewhat differently, Mu’adz (1991) allows for phrasal coordination, and defines planes explicitly. Moltmann (1992) further formalizes this approach: making use of McCawley’s (1968, 1982) graph theory, she distinguishes between m-planes (meaning-planes) and f-planes (formal-planes) which relate to formal syntax. (See Rogers 2003 for a discussion of multidimensional graphs.) In all of these works, an explicit behindance relation is lacking: it is inferred from the absence of dominance and precedence. Grootveld (1992:70) states that this means that Goodall “does not take the third dimension seriously”. Van Oirsouw (1987) criticizes the parallel structure approach to coordination, but Haegeman (1988:287) stresses the merits of Goodall’s work, noting that the difficulties are often matters of execution. Grootveld (1992) replies to Van Oirsouw that the problems are not inherent to a three-dimensional approach: rather, the major issues relate to (i) Goodall’s commitment to a sentential analysis of coordination, (ii) the lack of a structural position for the conjunction markers, and (iii) the absence of an explicit behindance relation. I agree with Grootveld that these problems can be avoided by combining the concept of behindance invoked in parallel structure analyses with an approach that recognizes the existence of a Coordinate Phrase (CoP). It may at first seem unattractive to complicate syntax in this way simply to accommodate common coordination. However, if behindance is the basis for all types of parataxis—as claimed here—then the idea may have some merit. 3. THEORETICAL PRELIMINARIES After showing that dominance and precedence relations are defined independently of each other (section 3.1), I discuss in some detail the logical outputs of the structurebuilding operation Merge (section 3.2). 3.1. The independence of dominance and precedence Kayne (1994) claims that precedence derives from asymmetric c-command—hence indirectly from dominance. If this is correct, precedence is not an independent degree of freedom in syntax. Consequently, behindance, if it exists, does not constitute the third dimension in syntax—but the second. Kayne’s theory is formulated as the Linear Correspondence Axiom. Consider (24), where YP is the specifier of X and ZP the complement of X. (In Kayne’s notation, a specifier is an adjunct; X-bar labels do not exist.) (24) According to Kayne (1994), the asymmetry between the sister nodes X and its complement ZP need not be stipulated: X asymmetrically c-commands the components of ZP, so by the Linear Correspondence Axiom X precedes Z in the output string. The c-command relation between X and ZP is mutual, and so is irrelevant in determining word order. In this way, asymmetric c-command leads to the conclusion that the YP X ZP XP XP CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 10 10 dominance relation derives the precedence relation. However, in many cases, information about precedence cannot be derived from dominance. Consider the specifier YP and its sister node, the lower segment of XP. YP and XP mutually c-command each other, and so no linear order between their terminals can be established. This is not the desired outcome; Kayne (1994:16ff) solves this by adding proviso (i) to the definition of c-command (25). (25) X c-commands Y iff: (i) X and Y are categories, and (ii) X excludes Y, and (iii) every category that dominates X dominates Y. The lower XP in (24) is a segment and not a category. Therefore, it cannot enter into a c-command relation with anything. So YP c-commands all the components of XP, but XP does not c-command the components of YP. (As for the higher XP, it does not ccommand YP because (25-ii) is not satisfied, since XP does not exclude YP.) Consequently, YP asymmetrically c-commands the components of XP, and so linear order can be established. Kayne’s attempt to derive precedence from hierarchy is ultimately unsuccessful because—to create an asymmetry between the sister nodes YP and XP—he must stipulate that a segment cannot enter into a c-command relation (25-i). Now consider how Chomsky’s (1995) Minimalist Program fares: precedence is not part of the core syntax, and word order is relegated to the phonological component. This departs from Kayne, for whom conditions on word order are at the heart of the grammar. Still, Chomsky (1995:334ff) accepts Kayne’s idea that syntax is antisymmetric, and that word order can be derived from dominance. But Chomsky faces the same problem as Kayne: just as Kayne excludes XP segments from c-command, Chomsky excludes intermediate X-bar nodes from c-command. Such a move is equivalent to the proviso in (25-i). Moreover, if c-command is derived in the course of the syntactic derivation (Epstein 1999), and if X and Y Merge, then c-command is the total relation between X on the one hand, and Y and all its constituents on the other hand. (Epstein uses the term category rather constituent, but the former plays no substantive role in his discussion.) But if c-command derives from Merge (Epstein 1999), then a proviso such as (25-i) is not only unnecessary but impossible. This leads to the conclusion that precedence does not follow from dominance, but is an independent relation. This finding is in keeping with Koster (1999, 2000b), who identifies precedence as one of the basic properties of the Configurational Matrix, which says that, universally, syntactic structures are formed as [b a d]. Here, d is an element that depends on a, the antecedent (in the broadest possible sense). This configuration applies not only to anaphora, but also to structure-building operations such as Merge. Zwart (1999) comments that the Configurational Matrix should have two, and preferably only two, properties: bi-uniqueness (i.e., one a relates to one d) and asymmetry. In terms of Merge we may say that a is added to the target d. So the asymmetry between sisters concerns the question of what depends on what (Zwart 2006). This asymmetry is the second degree of freedom in syntax; following traditional terminology, I call it precedence. Precedence in syntax is the abstract asymmetry between a and d; linearization translates abstract CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 11 11 syntactic precedence into precedence between elements in a string. From the discussion above, it follows that Merge is not “set Merge”, contra Chomsky (1995:243). If A and B Merge, they combine as an ordered pair < A, B >. This is also claimed by Koster (1999, 2000b), Zwart (1999, 2004, 2006), Di Sciullo (2000), and Langendoen (2003). In a tree, A appears to the left of B; this is equivalent to the ordered set notation. 3.2. Properties of Merge A syntactic derivation makes use of the structure building operation Merge (Chomsky 1995). If we have binary branching, Merge combines two syntactic objects into a larger object. Merge is often distinguished from Move, but, strictly speaking, this is inaccurate, as Move involves Merge; only the input is different. (An aside on terminology: note that Chomsky 2001 uses External Merge [for normal Merge] and Internal Merge [for Move].) Even if we presuppose strict cyclicity, there are in fact seven logical possibilities to consider regarding the output of Merge. This reflects the status of the relevant input object, which can be: (i) selected from the lexicon (or numeration); (ii) a partial derivation from the syntactic work space (that is, the result of a previous instance of Merge); and (iii) a constituent of a partial derivation in the syntactic workspace. Of the seven logical possibilities, the first three are illustrated in (26). These are instances of simple Merge, and consist of selecting input objects selected from the lexicon (i) and input objects that are partial derivations (ii). For expository purposes, it is assumed that the lexicon contains heads (X, Y), and that partial derivations are phrases (XP, YP). Note that possibility 2—the Merge of an input selected from the lexicon with an input that is a partial derivation —is associated with two symmetrical subcases, 2a and 2b. (26) Simple Merge: Merge(a ,b), input from lexicon or partial derivation a b output 1. lexicon lexicon [X Y] 2a. lexicon partial derivation [X YP] 2b. partial derivation lexicon [XP Y] 3. partial derivation partial derivation [XP YP] Which of the two input objects projects is important, but this is another issue, as is the discussion on the necessity of a label (Chomsky 1995; Collins 2002); I put these matters aside. If re-merging a constituent (iii) is taken into account, we arrive at a fourth logical possibility, shown in (27); this is usually called Move. Note that there are two symmetrical subcases, 4a and 4b, which further subclassify according to whether the remerged constituent is a head or a phrase. For ease of exposition, the movement site is indicated by a trace (with no commitment regarding the theoretical status of the latter). CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 12 12 (27) Internal Remerge: Merge(a ,b), input from (constituent of a) partial derivation a b output____________ 4a-i constituent of YP partial derivation [XPi [YP … ti …]] 4a-ii [Xi [YP … ti …]] 4b-i partial derivation constituent of XP [[XP … ti …] YPi] 4b-ii [[XP … ti …] Yi] Example (27) involves Internal Remerge, where the constituent to be remerged is combined with the root that originally contained it. Another possibility is External Remerge, where the constituent to be remerged (which as before may be a head or a phrase) is combined with a root that did not originally contain it. In terms of outputs, this gives rise to the last three of the seven logically possible outputs of Merge; I adopt the convention of marking with an asterisk the externally merged constituent. (28) External Remerge: Merge(a ,b), input from constituent external to derivation a b output_______ 5a-i. constituent of ZP lexicon [XP* Y] 5a-ii. [X* Y] 5b-i. lexicon constituent of ZP [X YP*] 5b-ii. [X Y*] 6a-i. constituent of ZP partial derivation [XP* YP] 6a-ii. [X* YP] 6b-i. partial derivation constituent of ZP [XP YP*] 6b-ii. [XP Y*] 7-i. constituent of WP constituent of ZP [XP* YP*] 7-ii. [X* YP*] 7-iii. [XP* Y*] 7-iv. [X* Y*] The options in (28) are usually ignored, and of course they could be explicitly excluded. There are two ways to interpret (28): as multidominance or as interarborial movement (M. de Vries 2005b, 2005c).4 The latter is explored by Bobaljik and Brown (1997) and Nunes (2001); the former by Van Riemsdijk (2004) and Citko (2005). Interarborial movement is also called sideward movement (Nunes 2001). Multidominance is sometimes called sharing, grafting (Van Riemsdijk 1998, 2004), or Parallel Merge (Citko 2005). Multidominance, as such, was proposed earlier by Sampson (1975), McCawley (1982), and by now, numerous others, especially in the context of Right Node 4 One could also interpret regular movement (Internal Remerge) in terms of multidominance; see Sampson (1975), Blevins (1990) [ref], Gärtner (2002), Frampton (2004), and M. de Vries (2005b, 2005c). CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 13 13 Raising. Here, I refrain from discussing constructions that involve the application of External Remerge. This completes the survey of all of the logical outputs of the structure-building operation Merge. I conclude this section with some comments on what I take to be the core properties of Merge, the boundary conditions that it is subject to, and the relation of Merge to c-command. Merge has two core properties: it is a structure-building operation that combines syntactic objects (29a), and in so doing induces hierarchical relations amongst those objects (29b). (29) Core properties of Merge: a. Structure building: Merge combines syntactic objects into one new, larger object. b. Hierarchical: The input objects are included in the newly created output object. Merge is subject to a number of boundary conditions: because it takes two input objects, it necessarily generates binary branching structures (30a); it freely selects its input objects (30b), it is strictly cyclic (30c), and it is asymmetric (30d). (30) Boundary conditions on Merge: a. Binary branching: Merge takes two input objects. b. Free selection: An input object is: (i) selected from the lexicon (or numeration), or (ii) a partial derivation selected from the syntactic work space, or (iii) a constituent of a partial derivation. c. Strictly cyclic: The output of Merge is not included in a larger syntactic object. d. Asymmetric: The output of Merge is an ordered pair. Thus, Merge builds hierarchical structures whose inputs are binary and freely selected, and whose outputs are strictly cyclic and asymmetric. I now turn to the question of whether Merge can derive c-command. Extending Epstein’s (1999) proposal, the ccommand dependency between words and phrases is generated during the course of the syntactic derivation, as in (31). The equivalent representational definition is given in (32). (31) C-command: derivational definition (preliminary version)5 If Merge(A,B) then A c-commands B and all the constituents included in B. (32) C-command: representational definition A c-commands B iff there is an X, X is B or X includes B, for which holds: A is the preceding sister of X. (The preliminary version in (31) will be modified below.) The c-command relation is total because of the cyclic extension condition (30c) in combination with the fact that inclusion is a transitive relation. Furthermore, since Merge is asymmetric, c-command is 5 I do not use term, which means “constituent included in”. CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 14 14 asymmetric.6 That is, for all A, B: if A c-commands B, it cannot be the case that B c-commands A. The next section shows how the idea of behindance can be treated in terms of Merge. 4. PROPOSAL: PARATAXIS AS BEHINDANCE-INCLUSION After introducing the distinction between dominance-Merge versus behindance-Merge (section 4.1), I show how it applies to coordination (section 4.2), apposition (section 4.3), and parenthesis (section 4.4). 4.1. Dominance-Merge versus behindance-Merge In the three-dimensional accounts cited above, behindance is viewed as an alternative to precedence. Accordingly, there would be two ways of ordering a pair, with additional assumptions needed for linearization. For instance, the algorithm that scans a syntactic structure and produces a (linear) string would need to be able to recognize behindance. For tree structures, one could go to a daughter node, to the +preceding one or the -behind one first; add terminals to the string, etc. The tree is scanned top-down, left-right, and front-back. With binary branching, sister nodes are organized left-right or front-back. Without additional assumptions, such three-dimensional graphs are equivalent to twodimensional ones. Crucially, in such approaches, conjuncts are treated symmetrically, with all conjuncts dominated by nodes higher up in the matrix clause. But it remains unclear how such analyses account for cases of parataxis other than common coordination. There is a more interesting way of representing behindance, namely as an alternative for dominance. Since dominance is itself based on the notion of inclusion, this means that we have to reconsider inclusion. If A and B are included in C, C dominates A and B; I call this dominance-inclusion. My proposal is simple: there is a second type of inclusion, which I call behindance-inclusion. (Alternatively, there is inclusion [dominance], and there is subscripted inclusion [behindance], that is, inclusion with an additional property [Jan-Wouter Zwart, p.c.].) Like dominance and precedence, behindance is determined locally. (Strictly speaking, precedence and dominance parallel minus behindance [“beforeness”].) Inclusion is still a transitive relation: if A x-includes B and B x-includes C, then A x-includes C, where x may be d-included (dominance-included) or b-included (behindance-included). C-command is now reformulated as (33): (33) C-command (definitive version): If Merge(A,B) then A c-commands B and all the constituents dominance-included (d-included) in B. As we shall see, one consequence of defining c-command in terms of dominanceinclusion is that behindance—which is the basis of paratactic structures—will be blind to 6 This is correct if specifiers and heads universally precede intermediate categories (predicates) and complements, respectively (Kayne 1994). If claims about a universal X’-schema turn out to be wrong, the definition of c-command in (31) can be adjusted to: “if Merge(A,B) then A c-commands B […] and B c-commands A […]”. CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 15 15 c-command relations. Recall that one of the core properties of Merge is that it derives syntactic objects that have a hierarchical structure; this reflects the fact that the output of Merge includes its input. But if there are two types of inclusion—dominance-inclusion (d-inclusion) and behindance-inclusion (b-inclusion)—then there are two types of Merge, namely dominance-Merge (d-Merge) and behindance-Merge (b-Merge). (34) a. Dominance-Merge (d-Merge): The input objects are d-included in the output object. b. Behindance-Merge (b-Marge: The input objects are b-included in the output object. While d-Merge is the basis for syntactic hierarchy, b-Merge produces a paratactic hierarchy. On this view, parataxis (and hence b-Merge) is independent of multidominance. Therefore, grafting (Van Riemsdijk 2004) or parallel Merge (Citko 2005) is a different subject: although it often arises in the context of coordination (e.g., across-the-board wh-movement or backward conjunction reduction) it is not restricted to coordination (e.g., it also occurs with transparent free relatives). The latter predictably fail to show the invisibility effects discussed below (see also Espinal 1991). Let us turn to the representation of syntactic objects. We can draw a tree structure, or use sets, or compile a list of local relations: these notations are equivalent. This is illustrated in (35) for d-Merge, and in (36) for b-Merge, with the representations in (a) using tree structures, those in (b) using sets, and those in (c) listing the local relations. To represent b-Merge with tree-structures and set notations, additional conventions are necessary. For example, in (36a), dotted lines are used to suggest a three-dimensional space, so that A and B are behind C; in (36a-ii) and (36b), the paratactic hierarchy is indicated by an asterisk next to C. (35) d-Merge (A,B) ® C a. b. [C A, B ] c. A precedes B C dominates A C dominates B (36) b-Merge (A,B) ® C a.i a.ii. b. [C* A, B ] c. A precedes B A is behind C B is behind C Consider a structure where both dominance-Merge and behindance-Merge have applied the derivation is in (37a), the representation is in (37b). A B C C A B A B C * CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 16 16 (37) a. d-Merge(A,B)®C b-Merge(D,C)®E d-Merge(F,E)®G b. = According to the definition of c-command in (33), A c-commands B (B is d-included in A); D c-commands C, A, and B (A, B, and C are d-included in D); and F c-commands only E (E is d-included in F). Crucially, F does not c-command D and C, as they are not d-included in E. This is because the syntactic hierarchy is interrupted at E by the paratactic hierarchy; hence the constituents of E are in a paratactic relation to the higher nodes. The linearization of syntactic structure must take place at or beyond the syntaxphonology interface, and is complicated by the use of three-dimensional graphs. However, this is not the case for the present theory. A recursive tree-scanning algorithm uses information relating to constituency (inclusion), and to the asymmetry between sisters (precedence). The processing time of such an algorithm grows linearly with the size of the input object. This is an advantage over an exponentially growing procedure like Kayne’s (1994) Linear Correspondence Axiom, which can be implemented as a filter; see M. de Vries (2002:7ff). A tree is scanned top-down from most inclusive to least inclusive node, and the terminals (words) can then be arranged in a string. Since inclusion generalizes over behindance-inclusion and dominance-inclusion, it is irrelevant whether daughter nodes are below or behind their mother node. And binary branching prevents nodes from being simultaneously below and behind some mother node. In (37), D and C are daughters of E; scanning the structure generates the terminal string [F D A B]. The next sections show how the theory outlined above applies to coordination and parenthesis. 4.2. Behindance-Merge and Coordination Simple coordination consists of two conjuncts and a conjunction. The latter, for example and, has been argued to be a functional head (Munn 1987; Johannessen 1998; Van der Heijden 1999), which I here label Co. Co combines with the second conjunct to form Co’; then the first conjunct Merges with Co’ to form CoP. The second conjunct is not hierarchically subordinated, as it would be if the structure were derived by dominance- Merge (d-Merge): [CoPXP1 [Co’ Co XP2]]. Rather, the second conjunct is paratactically construed; in the present analysis, this indicates that the second conjunct combines via behindance-Merge (b-Merge). This is illustrated in (38): the derivation appears in (38a), and the corresponding structure in (38b). F D A B C E * G D A B C F E G CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 17 17 (38) a. b-Merge(Co,XP2)®Co’, d-Merge(XP1,Co’)®CoP b. The structure in (38b) is binary branching. Co and XP2 are behind Co’; therefore, they are not c-commanded by XP1. Furthermore, [Co’ Co XP2] is a constituent (and [CoP XP1 Co XP2] is too), but crucially [XP1 Co] is not a constituent. In other words, as noted in Ross (1986), the coordinator forms a unit with the second conjunct, but not with the first conjunct. This is illustrated in (39)–(41): the well-formed (a) examples are instances of [Co’ Co XP2]; the ill-formed (b) examples are attempts to treat the [XP1 Co] sequence as a unit. (39) a. Bill went to the movies. And Anna stayed at home. b. * Bill went to the movies and. Anna stayed at home. (40) a. Bill bought two books _ yesterday, and one magazine. b. * Bill bought _ one magazine yesterday, two books and. (41) a. Bill, and Anna. b. * Bill and, Anna. A more complicated case, such as (42a), where the first conjunct is itself a coordinate structure, has the derivation in (42b), and is represented as in (42c). Only local relations are computed: Co1 (and) and DP2 (Joop) are behind Co1’; DP2 is paratactically related to DP1 (Jaap); similarly, DP3 (Joep) is paratactically related to the CoP1 (Joop and Jaap). (See M. de Vries 2005a for discussion of initial coordinators and distributivity effects.) (42) a. ((Jaap and Joop) or Joep) b. b-Merge(Co1,DP2)®Co’1 d-Merge(DP1,Co’1)®CoP1 b-Merge(Co2,DP3)®Co’2 d-Merge(CoP1,Co’2)®CoP2 c. = DP 1 Co 1 DP 2 Co' * CoP 1 Co 2 DP 3 Co' * CoP 2 Co XP2 XP Co' CoP 1 DP1 Co' CoP1 Co' CoP2 Co1 Co2 DP3 DP2 CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 18 18 The present proposal combines the Coordinate Phrase (CoP) hypothesis with a binary branching three-dimensional structure, formalized in terms of behindanceinclusion (b-inclusion). It differs from analyses such as Williams (1978) and Goodall (1987) in important respects: (i) the structure is asymmetric; (ii) it involves a Coordination Phrase (which provides a structural position for the conjunction as the Coordinative head); (iii) the paratactic dimension is explicitly defined; and (iv) no multidominance is needed for coordination. Apart from the issue of b-inclusion (to which I return below), the Coordinate Phrase proposed here is essentially that of Johannessen (1998). I have nothing new to offer concerning Case and agreement: the latter requires that the structure be asymmetric, so that phenomena such as unbalanced Case and first conjunct agreement can be handled. (See Munn 1993, Johannessen 1998, Camacho 1997, Progovac 1998, Aoun et al. 1999, and Citko 2004 for discussion.) 4.3. Behindance-Merge and Apposition As already discussed above, appositions and appositive relative clauses can be analyzed as instance specifying coordination; some examples are repeated in (43). (43) a. Joop, our boss b. Joop, (i.e., he) who is our boss c. the White House, or the house with the Oval Office The derivation of such strings typically involves a derivation such as (44a), with the structure in (44b). Here, DP1 is Joop or the White House; Co is the specifying coordinative head (ø, or); and DP2 is our boss, (he) who is our boss or the house with the Oval Office. I assume that the paratactic intonation is triggered by the specifying Co head. (44) a. b-Merge(Co,DP2)®Co’ d-Merge(DP1,Co’)®CoP b. 4.4. Behindance-Merge and Parenthesis Now consider how a parenthetic clause (CPpar) attaches to a matrix clause. Its position within the host sentence is relatively free (Stoltenburg 2003; Schelfhout et al. 2003b) and it is somehow indifferent to the syntactic hierarchy of the matrix clause. Suppose (incorrectly, as it will turn out) that CPpar is adjoined via b-Merge with the matrix VP. This gives [VP* CP VP]. But then not only CPpar but also the existing part of the matrix clause (anything included in the VP) will be behind the node created (VP*). This cannot be correct. The solution is straightforward: a parenthetic clause is embedded in, say, a parenthetic phrase (ParP). The combination of ParP with CPpar via behindance-Merge (45a) yields the structure in (45b). Co DP2 DP Co' CoP 1 CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 19 19 (45) a. b-Merge(Par,CPparenthesis)®ParP b. = ParP can then adjoin (by dominance-Merge) to some projection XP in the matrix. This would correspond to a derivation such as (46a) with the structure in (46b). (Here, only the star notation is used.) (46) a. b-Merge(Par,CPpar)®ParP d-Merge(ParP,XP)®XP+ d-Merge(Y,XP)®YP b. Observe that Par and CPpar are behind ParP. The c-command relations are as follows: Y c-commands XP+, ParP, and XP; ParP c-commands XP; Par c-commands CPpar. Crucially, neither Y nor XP c-commands the constituents of ParP, namely Par and CPpar, because these are not d-included in ParP. Furthermore, CPpar does not c-command XP or a constituent of XP, for the simple reason that it is embedded. What is ParP? It seems to be a monovalent coordination phrase. This is reminiscent of Munn (1993), who analyzes common coordination as (right)-adjunction of a monovalent Boolean Phrase BP)—which contains the conjunction and the second conjunct—to the first conjunct. However, there are crucial differences: (i) the contents of BP, but not ParP, are included in the syntactic dominance hierarchy; and (ii) Par, but not B, is a specifying conjunction. In the analysis advanced here, like any coordinative head, Par triggers behindance. Furthermore, Par contains the same intonational trigger as a specifying coordinator. In this regard, it is telling that hedges can start with a coordinator, as in (47a). That an overt coordinator can appear before the parenthetic CP indicates that Par is spelled out as and in some cases. However, since specifying coordination is often asyndetically construed (i.e., it has no phonological content), it is not surprising that this also holds of parenthetic clauses (47b). Consistent with the present analysis, with some modifications a coordinator can be made visible (47c).7 7 So in (47c) may be related to quotative operator, discussed in Collins and Branigan (1997), Schelfhout (2000), Corver and Thiersch (2002), and M. de Vries (2006b). Y Par CPpar ParP * XP XP YP Par CPpar ParP * ParP Par CPpar CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 20 20 (47) a. Hank—and I hate to tell you this—stole my bike. b. Hank, I think, stole my bike. c. Hank, or at least so I think, stole my bike. Since there are many types of parenthetic phrases, the complement of Par (here, CP) can have many different shapes; moreover, there can be ellipsis, etc. What is relevant here is that they all have a common basis, namely a phrase structure that involves behindance. 5. THE INVISIBILITY OF PARATACTIC MATERIAL A parenthetic clause is behindance-Merged, and is outside the c-command domain of the matrix. Consequently, a constituent of CPpar cannot be syntactically bound from outside. This is illustrated in (48) for quantifier binding: (48) * Everyi boy—(and) hei just ran away—had stolen an apple. Similarly, a variable in an appositive relative clause cannot be bound by a quantifier in the matrix (49a). This contrasts with restrictive relatives, which are clearly subordinated, and which participate in quantifier-variable binding, (49b).8 (49) a. * Everybodyi was talking about the Louvre, which hei visited yesterday. b. Everybodyi was talking about the museum that hei visited yesterday. Since movement is always to a c-commanding position, we predict that movement out of a parenthetic clause is impossible. This prediction is confirmed: (50) a. Lisa grumbled—who stole her bike, you know—all day long. b. * Whoi did Lisa grumble—ti stole her bike, you know—all day long? (51) a. Hank saw Lisa, who carried a torch. b. * Whati did Hank see Lisa, who carried ti ? While the anti-movement effects in (50) and (51) could be derived from island constraints, these examples are worse than standard island violations. I take this to indicate that something else is going on. It seems that there are no c-command-based relations between paratactic material and constituents in the matrix sentence; I call this invisibility (M. de Vries 2007):9 8 See Demirdache (1991) for similar data. The antecedent of an appositive relative is related to the relative pronoun via E-type anaphora, which is a discourse relation that does not involve syntactic binding. See Sells (1985) and Del Gobbo (2003). 9 Invisibility is but one expression of the syntactic independence of paratactic material. It has also been claimed that parentheses have an independent illocutionary force, focus-background structure, and tense. See Espinal (1991), Pittner (1995), and Burton-Roberts (1999). CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 21 21 (52) Invisibility of paratactic material: If B is paratactically construed with A (i.e., B is behind A), B is invisible to c-command relations with any object that is Merged with A or a projection that includes A. The reverse is also true; this could be called blindness of paratactic material. No paratactically construed constituent B (or a constituent included in B) can c-command into the matrix clause. For instance, there is no movement into a parenthetic clause, nor is there any binding from within a parenthetic clause. This is trivial: because B is embedded—whether in a coordinate or a parenthetic phrase—the intended (movement or anaphoric) relation would be countercyclic. How does behindance-Merge affect coordinate structures? Consider the derivation in (53a), and the corresponding structure in (53b). (53) a. b-Merge(Co,YP)®Co’ d-Merge(XP,Co’)®CoP d-Merge(Z,CoP)®Z’ d-Merge(RP,Z’)®ZP b. According to the definition of c-command in (33), YP (the second conjunct) is not ccommanded by XP because it is not d-included in Co’, the sister of XP. For the same reason, YP is also not c-commanded by RP and Z. In other words, the line of dominance from ZP to YP is broken at Co’. XP (the first conjunct), however, is d-included in CoP, and therefore c-commanded by Z and RP and any phrase higher up in the matrix. Thus, we predict that the two conjuncts will behave differently relative to c-command relations: (54) Asymmetry Between Conjuncts: In a coordinate structure, the second conjunct but not the first is invisible for the syntactic context, in terms of c-command. What is relevant here is the relation between the two separate conjuncts and the host context. Consider movement. Usually, movement out of a conjunct is impossible (55); this is the Coordinate Structure Constraint (Ross 1986).10 10 An exception is Across-The-Board (ATB) movement, for example, What did Peter buy _ and Bill sell _ ? ATB is a more general phenomenon; for instance, ATB quantifier binding is also possible: Everyi man loves hisi wife and hisi children; and so is ATB Case distribution: I saw him and her. It seems that the first conjunct can pass on properties to the second; how this can be explained is outside the scope of this article. RP Z XP Co' CoP Z' ZP Co YP CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 22 22 (55) a. * What did you buy _ and sell a book? b. * What did you buy a book and sell _ ? However, the Coordinate Structure Constraint does not apply to semantically asymmetrical coordination (Goldsmith 1985; Culicover and Jackendoff 1997; Van der Heijden 1999). Examples where a constituent is raised from the first conjunct are given in (56) for English and in (57) for Dutch.11 (56) How much can you drink _ and still stay sober? (57) Dutch: a. Hoeveel chocola denk je dat je kunt _ eten en toch niet misselijk worden? how.much chocolate think you that you can _ eat and still not sick get ‘How much chocolate do you think you can eat _ and still not get sick?’ b. Hoe lang kun je op een dag _ studeren en daarbij toch vrolijk blijven? how long can you on one day _ study and thereby still cheerful stay ‘How long are you able to study _ on one day and still stay cheerful?’ c. Wie zei je dat er _ nog niet vertrokken was of Joop kocht een duur cadeau? who said you that there _ not yet left because Joop bought an expensive gift ‘Who did you say _ had barely left before Joop bought an expensive gift?’ d. Wat had Joopje nog niet _ gekregen of hij begon ermee te gooien? what had Joopje not yet _ got or he started therewith to throw ‘What did Joopje just receive _ and he already started demolishing?’ Movement from the second conjunct in similar sentences is impossible; this is shown in (58) for English, and in (59) for Dutch.12 (58) * What did Joop finally overcome his inhibitions and ask Jaap _ ? (59) Dutch: a. * Wat kun je een pond chocola eten en toch niet _ worden? what can you a pound chocolate eat and still not _ become Intended: ‘What can you eat a pound of chocolate and still not become _ ?’ 11 Examples (56) and (57) cannot be analyzed as matrix CP coordination of a question with a proposition, with forward deletion into the second conjunct. There is no correspondence between what would be elided in the second conjunct and its antecedent in the first conjunct. For instance, in (57a) the missing part would have to be je denkt dat je kunt ‘you think that you can’, but the first conjunct contains denk je dat je kunt. Moreover, this string is not a constituent. Therefore, the construction at hand provides evidence for the existence of “small conjuncts” (contra Wilder 1997). 12 Colloquial English and Afrikaans have a quasi-serial verb construction of the type go and get, as in What did John go and get _? If this is coordination of verbal heads (De Vos 2005), then it is not movement from a second conjunct. More problematic is the transitive variant, What did John go to town and buy _? I have no explanation for the latter, except to note that it is limited to a small number of predicates. See also Lakoff (1986) and Postal (1998) for discussion. CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 23 23 b. * Wat kun je op een dag zes uur studeren en toch _ blijven? what can you on one day six hours study and still _ stay Intended: ‘What can you study for six hours on one day and still stay _ ?’ c. * Wie was Joop nog niet vertrokken of _ kocht een duur cadeau? who was Joop still not left or _ bought an expensive gift Intended: ‘Who had Joop barely left before _ bought an expensive gift?’ d. * Wat was Joop nog niet vertrokken of Jaap heeft _ gekocht? what was Joop still not left or Jaap has bought Intended: ‘What had Joop still not left before Jaap bought _ ?’ That extraction is possible from a first conjunct (56)–(57) but not from a second conjunct (58)–(59) is predicted by Asymmetry Between Conjuncts (54). Asymmetry Between Conjuncts also holds of anaphora. I illustrate this with the complex pronoun hemzelf in Dutch, with is an identifying emphatic expression consisting of a pronominal part hem ‘him’ subject to Condition B, and an emphatic part zelf. As shown in (60), while zichzelf is a local anaphor, hemzelf is not.13 (60) Dutch: Joopi beloonde zichzelfi / *hemzelfi . Joop rewarded SE-self / PRON-self ‘Joop rewarded himself.’ In a coordinated DP, when hemzelf is the first conjunct, it cannot be bound (61a); when hemzelf is the second conjunct, it can be bound (61b). (See M. de Vries 1999 for a discussion of the discourse conditions that favour the use of an identifying emphatic expression over a simple pronoun in contexts such as (61b).) (61) Dutch: a. * Joopi beloonde hemzelfi en Anna rijkelijk. Joop awarded PRON-self and Anna richly Intended: ‘Joop richly awarded himself and Anna.’ b. Joopi beloonde Anna en hemzelfi rijkelijk. Joop awarded Anna and PRON-self richly ‘Joop richly awarded Anna and himself.’ The contrast in (61) is consistent with Asymmetry Between Conjuncts. The first conjunct is visible to a c-commanding phrase, and so cannot be bound by the subject Joop; a Condition B effect. The second conjunct is not visible to a c-commanding phrase, and so can be bound by the subject. We might expect the local anaphor zichzelf to give the opposite pattern, namely to be felicitous in the first conjunct, but not in the second conjunct. Instead, zichzelf may occur in either the first or the second conjunct (although I have a preference for (62a)). 13 In some dialects of Dutch the reduced form ’mzelf is used as an anaphor; this is not the relevant target here. CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 24 24 (62) Dutch: a. Joopi beloonde zichzelfi en Anna rijkelijk. Joop awarded SE-self and Anna richly ‘Joop richly awarded himself and Anna.’ b. Joopi beloonde Anna en zichzelfi rijkelijk. Joop awarded Anna and SE-self richly ‘Joop richly awarded Anna and himself.’ Why is (62b) acceptable? Although (62b) is excluded if we use DP coordination, it has a possible analysis in terms of CP coordination, with forward ellipsis, as in (63) where zichzelf is locally bound within the second conjunct. Thus, we explain the complementary distribution between anaphors and pronouns in a first conjunct, and the overlapping distribution in a second conjunct. (63) [Joop beloonde Anna] en [Joopi beloonde zichzelfi] Joop awarded Anna and Joop awarded SE-self Note that a CP analysis cannot be invoked for (61a), as in (64), because the first conjunct would in any case contain a violation of Condition B: (64) * [Joopi beloonde hemzelfi] en [Joop beloonde Anna] Joop awarded PRON-self and Joop awarded Anna Support for the clausal analysis of (62b) comes from the fact that the appearance of zichself in the second conjunct is prohibited if a clausal analysis is impossible, as with with the Exceptional Case Marking constructions in (65) and (66). (65) Dutch: a. Op TV zag Agassii zichzelfi en Sampras een tenniswedstrijd tegen elkaar spelen. on TV saw Agassi SE-self and Sampras a tennis.game against each.other play ‘On TV, Agassi saw himself and Sampras play a tennis game against each other.’ b. ?* Op TV zag Agassii Sampras en zichzelfi een tenniswedstrijd tegen elkaar spelen. on TV saw Agassi Sampras and SE-self a tennis.game against each.other play Intended: ‘On TV, Agassi saw himself and Sampras play a tennis game against each other.’ (66) Dutch: a. Na de eerste zangles hoorde Joopi in gedachten zichzelfi en Pavarotti al een duet zingen. after the first singing.lesson heard Joopin his.mind SE-self and Pavarotti already a duet sing ‘After his first singing lesson, Joop already heard himself and Pavarotti sing a duet in his mind.’ b. ?* Na de eerste zangles hoorde Joopi in gedachten Pavarotti en zichzelfi al een duet zingen. after the first singing.lesson heard Joop in his.mind Pavarotti and SE-self already a duet sing Intended: ‘After his first singing lesson, Joop already heard himself and Pavarotti sing a duet in his mind.’ In the (b) examples, the use of hemzelf instead of zichzelf would make the sentence CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 25 25 acceptable. Hemzelf also makes a second reading available in (66b), in which Pavarotti sings a duet with himself, for example, as in a special effects film. Thus, Asymmetry Between Conjuncts in (54) is corroborated by the anaphora data from Dutch. More generally, this supports the claim that that coordination and parenthesis are instances of the more general relation of parataxis. 6. CONCLUSION Coordination differs from subordination. More generally, parataxis (of which coordination is a special case) differs from hypotaxis (of which subordination is a special case). Paratactic material has the following properties: (i) it provides additional information; (ii) it is not selected by any part of the matrix; (iii) it is phonologically set off from the matrix by a low intonation; (iv) it is linearly integrated with the matrix clause; and (v) it is part of a complete syntactic object. Models of syntax are largely designed to produce the hierarchical structures needed for hypotaxis. But this leaves open the question of how to produce the structures needed for parataxis. I have argued that behindance is the basic concept underlying parenthesis and coordination (with the latter subsuming common coordination as well as the specifying coordination relevant for apposition). It constitutes the third degree of freedom in syntactic phrase structure, the first two being dominance and precedence. Accordingly, phrase structure encodes three major asymmetrical relations. The first is dominance, and produces a syntactic hierarchy. The second is the selectional asymmetry between sisters, which is translated into precedence in the phonological component. The third is behindance, and leads to a paratactic hierarchy. In a derivational grammar based on Merge, behindance can be implemented as a type of inclusion, called behindance-inclusion (b-inclusion). It is an alternative to dominance, which is equated with dominance-inclusion (d-inclusion). Moreover, c-command is restricted to instances involving d-inclusion. Consequently, paratactic construal is predictably invisible to c-command relations, as evidenced by the impossibility of movement out of a parenthetic clause or binding into a parenthetic clause. In the present analysis, coordination is a syntactic configuration whose meanings reflect the properties of different coordinative heads. Some of the most important ones are listed in (67), which classifies the paratactic conjunctions discussed here. I have argued that that, in addition to conjunction and discussion, one must also recognize another type of coordination, namely specifying coordination, which includes apposition and parentheticals. (67) conjunction (Ù) disjunction (Ú) Coordinative heads specification bivalent (&: Apposition) monovalent (Parenthetic) other (opposition, etc.) CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 26 26 All paratactic conjunctions project into a coordination phrase. On the phonological side, specifying coordination is associated with a paratactic intonation break. The heads &: and Par are often asyndetic (i.e., they have no phonological content), but are sometimes spelled out as regular coordinators such as and and or. The syntactic status of additional connecting phrases like that is to say is not clear to me at this point. Bivalent specification is used for apposition, monovalent specification for parenthesis. The latter has no anchor; parentheses are embedded in a parenthetic phrase, whose head is a monovalent specifying coordinator, with the parenthetic phrase as a whole adjoined to some projection of the matrix clause. Because parenthesis also involves adjunction, this points to a parallel with adverbial material. There remains the question of the formal status of behindance. In the present analysis, the association between paratactic conjunctions and behindance-inclusion is not theoretically forced. I submit the following heuristic: (68) Coordinative heads, and only coordinative heads, trigger the application of behindance-Merge. Here, the and only clause prevents overgeneration of structures involving behindance- Merge. Thus, what all types of parataxis have in common is that the paratactic constituent is construed behind the projection of the coordinator that selects it, that is, Co— drawn for the set {Ù, Ú, &:, Par}—and XP are all behindance-Merged. REFERENCES Aoun, Joseph, Elabbas Benmamoun, and Dominique Sportiche. 1999. Further remarks on first conjunct agreement. Linguistic Inquiry 30:669–681. Blevins, James. 1990. Syntactic complexity: evidence for discontinuity and multidomination. Doctoral dissertation, University of Massachusetts. Bobaljik, Jonathan, and Samuel Brown. 1997. Interarboreal operations: Head movement and the extension requirement. Linguistic Inquiry 28:345–356. Bredschneijder, M. 1999. Reeksvorming: initiële coördinatie in het Nederlands. TABU 29:1–20. Burton-Roberts, Noel. 1999. Language, linear precedence and parentheticals. In The clause in English: In honour of Rodney Huddleston, ed. Peter Collins and David Lee, 33–52. Amsterdam: John Benjamins. Camacho, J. 1997. The syntax of NP coordination. Doctoral dissertation, University of Southern California, Los Angeles. Canac-Marquis, Réjean, and Mireille Tremblay. 1998. The wh-feature and the syntax of restrictive and non-restrictive relatives in French and English. In Theoretical analyses on Romance languages, ed. José Lema and Esthela Treviño, 127–142. Amsterdam: John Benjamins. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Chomsky, Noam. 2001. Beyond explanatory adequacy. MIT Occasional Papers in Linguistics 20. CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 27 27 Citko, Barbara. 2004. Agreement asymmetries in coordinate structures. In Formal Approaches to Slavic Linguistics 12: The Ottawa Meeting 2003, ed. Olga Arnaudova, Wayles Browne, María Luisa Rivero, and Danijela Stojanovic. Ann Arbor, MI: Michigan Slavic Publications. Citko, Barbara. 2005. On the nature of Merge: External Merge, internal Merge, and parallel Merge. Linguistic Inquiry 36:475–496. Collins, Chris. 2002. Eliminating labels. In Derivation and explanation in the Minimalist Program, ed. Samuel Epstein and T. Daniel Seely, 42–64. Oxford: Blackwell. Collins, Chris, and Phil Branigan. 1997. Quotative inversion. Natural Language and Linguistic Theory 15:1–41. Corver, Norbert, and Craig Thiersch. 2002. Remarks on parentheticals. In Progress in grammar: Articles at the 20th anniversary of the Comparison of Grammatical Models group in Tilburg, ed. Marc van Oostendorp and Elena Anagnostopoulou. Amsterdam: Meertens Institute Electronic Publications in Linguistics. http://www.meertens.knaw.nl/books Culicover, Peter, and Ray Jackendoff. 1997. Semantic subordination despite syntactic coordination. Linguistic Inquiry 28:195–217. Del Gobbo, Francesca 2003. Appositives at the interface. Doctoral dissertation, University of California, Irvine. Delorme, Evelyne, and Ray Dougherty. 1972. Appositive NP constructions. Foundations of Language 8:2–29. Demirdache, Hamida. 1991. Resumptive chains in restrictive relatives, appositives and dislocation structures. Doctoral dissertation, Massachusetts Institute of Technology. De Vos, Mark. 2005. The syntax of pseudo-coordination in English and Afrikaans. Doctoral dissertation, University of Leiden. Di Sciullo, Anna-Maria. 2000. Asymmetries: Consequences for morphological configurations and paradigms. Acta Linguistica Hungarica 47:81–101. Doron, Edit. 1994. The discourse function of appositives. In Proceedings of the 9th annual conference of the Israel Association for Theoretical Linguistics and the Workshop on Discourse, ed. Rhonna Buchall and Anita Mittwoch, 53–62. Jerusalem: Hebrew University of Jerusalem. Epstein, Samuel. 1999. Un-principled syntax and the derivation of syntactic relations. In Working minimalism, ed. Samuel Epstein and Norbert Hornstein, 317–345. Cambridge, MA: MIT Press. Es, G. van, and P. van Caspel. 1975. Samengestelde zin: parataxis. Archief voor de Nederlandse syntaxis, University of Groningen. Espinal, M.T. 1991. The representation of disjunct constituents. Language 67:726–762. Fabb, Nigel. 1990. The difference between English restrictive and nonrestrictive relative clauses. Journal of Linguistics 26:57–78. Frampton, J. 2004. Copies, traces, occurrences, and all that: Evidence from Bulgarian multiple wh-phenomena. Ms., Northeastern University, Boston, MA. Gärtner, Hans-Martin. 2002. Generalized transformations and beyond: Reflections on Minimalist syntax. Berlin: Akademie Verlag. Goldsmith, J. 1985. A principled exception to the coordinate structure constraint. CLS 21, Part 1, 133–143. Chicago Linguistic Society, University of Chicago, Ill. CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 28 28 Goodall, Grant. 1987. Parallel structures in syntax: Coordination, causatives and restructuring. Cambridge: Cambridge University Press. Grootveld, Marjan. 1992. On the representation of coordination. Linguistics in the Netherlands 9, 61–73. Grootveld, Marjan. 1994. Parsing coordination generatively. Doctoral dissertation, Leiden University. Haegeman, Liliane. 1988. Review of Grant Goodall, Parallel structures in syntax: Coordination, causatives and restructuring. Lingua 75:273–287. Halitsky, D. 1974. Deep structure appositive and complement NPs. Language 50:446– 455. Haspelmath, Martin. 2004. Coordinating constructions: An overview. In Coordinating constructions, ed. Martin Haspelmath, 3–39. Amsterdam: John Benjamins. Heijden, E. van der. 1999. Tussen nevenschikking en onderschikking. Doctoral dissertation, Catholic University of Nijmegen. Hendriks, Petra. 2004. Either, both and neither in coordinate structures. In The composition of meaning: From lexeme to discourse, ed. Alice ter Meulen and Werner Abraham, 115–138. Amsterdam: John Benjamins. Hendriks, Petra, and Jan-Wouter Zwart. 2001. Initiële coördinatie en de identificatie van woordgroepen. TABU 31:105–118. Hoeksema, Jack. 2000. Negative polarity items: Triggering, scope, and c-command. In Negation and polarity: Syntactic and semantic perspectives, ed. by Laurence R. Horn and Yasuhiko Kato, 115–146. Oxford: Oxford University Press. Johannessen, Janne Bondi. 1998. Coordination. Oxford: Oxford University Press. Johannessen, Janne Bondi. 2005. The syntax of correlative adverbs. Lingua 115:419–443. Kayne, Richard. 1994. The antisymmetry of syntax. Cambridge, MA: MIT Press. Klein, M. 1977. Appositionele constructies in het Nederlands. Doctoral disssertation, Catholic University of Nijmegen. Koster, Jan. 1999. De primaire structuur. TABU 29:131–140. Koster, Jan. 2000a. Extraposition as parallel construal. Ms., University of Groningen. Koster, Jan. 2000b. Variable-free grammar. Ms., University of Groningen. Kraak, A., and W.G. Klooster. 1968. Syntaxis. Culemborg, Netherlands: Stam- Kemperman. Lakoff, George P. 1986. Frame semantic control of the Coordinate Structure Constraint. In CLS 22, part 2: Papers from the parasession on pragmatics and grammatical theory at the 22nd regional meeting, ed. Anne M. Farley, Peter T. Farley, and Karl-Erik McCullough, 152–167. Chicago: Chicago Linguistic Society. Langendoen, D. Terence. 2003. Merge. In Formal approaches to function in grammar: In honor of Eloise Jelinek, ed. Andrew Carnie, Heidi Harley, and MaryAnn Willie, 307– 318. Amsterdam: John Benjamins. Lasnik, Howard, and Joseph Kupin. 1977. A restrictive theory of transformational grammar. Theoretical Linguistics 4:173–196. McCawley, James. 1968. Concerning the base component of a transformational grammar. Foundations of Language 4:243–269. McCawley, James. 1982. Parentheticals and discontinuous constituent structure. Linguistic Inquiry 13:91–106. CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 29 29 Moltmann, Friederike. 1992. Coordination and comparatives. Doctoral dissertation, Massachusetts Institute of Technology. Mu’adz, Husni. 1991. Coordinate structures: A planar representation. Doctoral dissertation, University of Arizona. Munn, Alan. 1987. Coordinate structure and X-bar theory. McGill Working Papers in Linguistics 4:121–140. Munn, Alan. 1993. Topics in the syntax and semantics of coordinate structures. Doctoral dissertation, University of Maryland. Nunes, Jairo. 2001. Sideward movement. Linguistic Inquiry 32:303–344. Oirsouw, Robert R. van. 1987. Three dimensionality. In Formal parameters of generative grammar, Yearbook III, ed. Ger de Haan and Wim Zonneveld, 31–46. Dordrecht: Instituut A.W. Groot voor Algemene Taalwetenschap van de Rijksuniversiteit te Utrecht. Pittner, Karin. 1995. Zur Syntax von Parenthesen. Linguistische Berichte 156:85–108. Postal, Paul M. 1998. Three investigations of extraction. Cambridge, MA: MIT Press. Progovac, Ljiljana. 1998. Structure for coordination. Part I and II. Glot International 3.7:3–6, and 3.8:3–9. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1999. A comprehensive grammar of the English language. 15th ed. London: Longman. [1985.] Riemsdijk, Henk van. 1998. Trees and scions––science and trees. Fest-web-page for Noam Chomsky’s 70th birthday. Riemsdijk, Henk van. 2004. Graft is the logically missing case of Merge. Visnyk of the Kiev National Linguistic University 7. Rogers, James. 2003. Syntactic structures as multi-dimensional trees. Research on Language and Computation 1:265–305. Ross, John Robert. 1986. Infinite syntax! Cambridge, MA: MIT Press. Safir, Ken. 1986. Relative clauses in a theory of binding and levels. Linguistic Inquiry 17:663–689. Sampson, Geoffrey. 1975. The single mother condition. Journal of Linguistics 11:1–11. Sauerland, Uli. 2001. On quantifier raising in German. Ms., University of Tübingen. Schelfhout, Carla. 2000. Corpus-based analysis of parenthetical reporting clauses. In Computational linguistics in the Netherlands 1998: Selected papers from the ninth CLIN meeting, ed. Frank van Eynde. Ineke Schuurman, and Ness Schelkens, 93– 107. Amsterdam: Rodopi. Schelfhout, Carla, Peter-Arno Coppen, and Nelleke Oostdijk. 2003a. Intercalaties? Dat zijn geloof ik van die tussendingen… Gramma/TTT. Schelfhout, Carla, Peter-Arno Coppen, and Nelleke Oostdijk. 2003b. Positions of parentheticals and interjections: A corpus-based approach. Linguistics in the Netherlands 20, ed. Leonie Cornips and Paula Fikkert, 155–166. Sells, Peter. 1985. Restrictive and non-restrictive modification. CSLI Report No. 85– 28:1–33. Skrabalova, Hana. 2003. La syntaxe de la coordination [Conj DP Conj DP] : comparaison entre le français, le tchèque et l’anglais. Paper read at the Workshop on coordination, Université Paris 7. Stoltenburg, B. 2003. Parenthesen im gesprochenen Deutsch. InLiSt – Interaction and CJL/RCL Interfaces #2005-064, Copyed RMD/KP, 16 Sept 2008 30 30 Linguistic Structures 34. Sturm, A.N. 1986. Primaire syntactische structuren in het Nederlands. Leiden: Martinus Nijhoff. Velde, J. te. 1997. Deriving conjoined XPs: A minimal deletion approach. In German: Syntactic problems – problematic syntax, ed. Werner Abraham and Elly van Gelderen, 231–259. Tübingen: Max Niemeyer. Vries, G. de. 1987. De representatie van coördinatie. Tilburg Studies in Language and Literature 7: Grammaticaliteiten, ed. Norbert Corver and Jan Koster, 331–377. Vries, G. de. 1992. On coordination and ellipsis. Docotoral dissertation, Catholic University of Tilburg. Vries, Mark de. 1999. Het schemergebied tussen pronomina en anaforen. Nederlandse Taalkunde 4:125–160. Vries, Mark de. 2002. The syntax of relativization. Doctoral issertation, University of Amsterdam. Vries, Mark de. 2005a. Coordination and syntactic hierarchy. Studia Linguistica 59:83–105. Vries, Mark de. 2005b. Merge: Properties and boundary conditions. Linguistics in the Netherlands 22, ed. Jenny Doetjes and Jeroen van de Weijer, 219–230. Vries Mark de. 2005c. Internal and external remerge: On movement, multidominance, and the linearization of syntactic objects. Ms., University of Groningen. (Revised version 2007.) Vries, Mark de. 2006a. The syntax of appositive relativization: On specifying coordination, false free relatives and promotion. Linguistic Inquiry 37:229–270. Vries, Mark de. 2006b. Reported direct speech in Dutch. Linguistics in the Netherlands 23, ed. Jeroen van de Weijer and Bettelou Los, 212–223. Vries, Mark de. 2007. Invisible constituents? Parentheses as b-merged adverbial phrases. In Parentheticals, ed. Nicole Dehé and Yordanka Kavalova, 203-234. Amsterdam: John Benjamins. Wilder, Chris. 1997. Some properties of ellipsis in coordination. In Studies on universal grammar and typological variation, ed. Artemis Alexiadou and T. Alan Hall, 59–107. Amsterdam: John Benjamins. Williams, E. 1978. Across-the-board rule application. Linguistic Inquiry 9:31–43. Zwart, Jan-Wouter. 1999. C-commanderen en de configurationele matrix. TABU 29:185– 190. Zwart, Jan-Wouter. 2004. Een dynamische structuur van de Nederlandse zin. Part 1 and 2. TABU 33:55–71 and 151–172. Zwart, Jan-Wouter. 2006. Local Agreement. In Agreement Systems, ed. Cedric Boeckx, 317–339. Amsterdam: John Benjamins. ' Herman Heringa*,i University of Groningen 1. Overview This paper discusses appositional constructions, such as the one in (1). (1) Pieter©s girl friend, a modest person, laughs about his ambitions. These constructions consist of two elements: the anchor and the apposition, where the apposition is the non-restrictive postmodifier of the anchor (Heringa & De Vries to appear). Though these elements are symply connected by comma intonation, a parenthetical dip in the intonation pattern (e.g. Schelfhout et al. to appear), this connection takes care of several things at once. First, it links the apposition to the anchor, which results in a combination that expresses something like "[the anchor] is [the apposition]". Second, it links the proposition that is created in this way to the proposition of the main sentence. In section 2, I argue that this complex connection is realized in the form of coordination on two levels at the same time. Sentence coordination links the two propositions and * Email: H.Heringa@rug.nl i I thank Mark de Vries for his useful comments and suggestions. This research was financially supported by the Netherlands Organisation for Scientific Research (NWO). ' 68 constituent coordination relates the anchor to the apposition. The combination of the anchor and the apposition results in a relation of predication. In section 3, I will argue that the type of predication involved depends on the semantic class of the appositional construction. These different classes also correspond to different overt apposition markers. 2. Apposition involves coordination At first sight, it is not clear what kind of connection there is between the anchor and the apposition. On the one hand the apposition and the anchor seem to be connected so closely that they together have one function in the sentence. On the other hand, the apposition is connected so loosely to the matrix sentence that the information it provides does not really belong to the main proposition, but constitutes just an aside. In the most common (or at least most studied) cases of apposition, the comma intonation is the only overt realization of this connection. This has led Potts (2007) to the proposal of a so-called comma operator, which takes care of both the relation between the anchor and the apposition and the relation between the anchor and the rest of the sentence, resulting in a more-dimensional semantics. However, if we take into account other cases as well, we find evidence for an alternative view, namely that the connection involves coordination. In this section, I will first give general evidence for a relation of coordination between the anchor and the apposition. After that, I will show that this connection operates on two levels at the same time: it connects both sentences and constituents. The most important argument for the idea that the anchor and the apposition are coordinated, comes from overt apposition markers.1 These 1 Apposition markers show that there is more in the structure between the anchor and the apposition than juxtposition. Also, they may be used to reveal something more about the interpretation of the relation there is between the anchor and the apposition. After all, they express this relation explicitly. Quirk et al. (1985: 1307) give the (non-exhaustive) overview of possible markers in (i), in groups that mark similar relationships. Which of the markers can be used, depends on the class of apposition. See also Heringa & De Vries (to appear) for a classification of apposition markers in Dutch. (i) that is to say, that is, i.e. namely, viz to wit in other words or, or rather, or better and = = 69 markers can precede the apposition in order to make the relation with the anchor more explicit. For now, it is important to note that the coordinators and, or and but can all be used as apposition markers. This implies that the three main types of coordination, conjunction, disjunction and adversative coordination, are all represented. Thus, the presence of coordinators suggests a parallel between appositional and coordinate constructions, as noted by Kraak & Klooster (1968), Quirk et al. (1985) and De Vries (2006, 2007). The following examples illustrate the use of coordinators in appositional constructions. The examples a and b are from Quirk et al. (1985:1311/12). (2) a. The United States of America, or America for short... b. You could cut the atmosphere with a knife, and a blunt knife at that. c. John is interested in science, but especially linguistics. The use of coordinators as apposition markers is not restricted to English or the Germanic languages. It is also possible in a Slavic language like Czech (Radek !im#k, p.c.), for example2 (3) a. Spojen$ St%tu Americk$, neboli Amerika ... United States Amer.-ADJ or America b. Vid l jsem n co kr%sn$ho, a to zlat! saw AUX-1SG-PAST something beautiful and it golden d m. house `I saw something beautiful, namely a golden house.' c. V+echny opice, ale/a p edev"#m orangutani, jsou ohro4en;m druhem. All apes but/and mainly orang-outans are threatened species A second general argument for a relation of coordination in appositional constructions is the fact that more than two elements can be combined in as follows for example, for instance, eg, say, including, included, such as especially, particularly, in particular, notably, chiefly, mainly, mostly 2 Something has to be added to the coordinators in Czech in order to let them function as apposition markers. The usual word for or in Czech is nebo. The words a and to form a unit in the case of apposition. ' 70 these cases. It is clear that coordination is able to connect multiple elements as well. (<) a. John, Mary©s boyfriend, a doctor, is a linguistic celebrity. b. John, Mary and Pete went to the store. Furthermore, coordination can connect sentences as well as constituents. This is an essential property for the connection in appositional constructions, because it operates on both levels at once. Below, I elaborate on this. To argue that appositional constructions involve coordination on the sentence level, it is of course necessary to show that they contain more than one sentence. Several people (e.g. Berckmans 199<= Dever 2001= Corazza 2005= Potts 2007) have argued in favour of this idea by claiming that utterances including an apposition consist of two separate propositions, each with their own, independent truth value. The reason is that the content of the appositional construction seems to be semantically independent of the content of the host clause (Del Gobbo 2003). The two propositions in (1), for example, can be described as follows: (5) a. Pieter©s girl friend laughs about his ambitions. b. Pieter©s girl friend is a modest person. The proposition in the b sentence is de-emphasized. Following Nouwen (2007), I use the terms primary and secondary content for main clause and apposition respectively. The independency of the truth values for primary and secondary content is intuitively clear in the sense that it is impossible to say whether an utterance is true in a situation where either the primary content is true and the secondary is not, or the other way round (Dever 2001). Also, if one denies the utterance, he denies either one of the propositions. Consider the possible reactions to (1) in (6) below: (6) a. No, she does not. b. Well yes, but she is not a modest person. The a-sentence does not imply that my brother©s girl friend is not a modest person, just as the b-sentence does not imply that my brother©s girlfriend does not laugh about his ambitions. The examples in (6) suggest that the whole utterance is false if the main proposition is false, whereas the utterance is neither true nor false if the appositional proposition is false. Note therefore that my claim about coordination on the sentence level does not imply that the truth values of the propositions are combined as in = = 71 normal sentence conjunction or disjunction. Rather, for the whole utterance to be true, the truth of the secondary content seems to be required. In that sense, the appositional proposition behaves like a presupposition (but see Potts 2007 for a discussion of the differences between presuppositions and the appositional proposition, which he calls a conventional implicature). A further argument that appositions are like sentences comes from the intuitive parallel with parentheticals, which are often overt sentences. The apposition in (7a) can easily be rephrased by the parenthetical sentence in (7b). (7) a. John, a clever guy, won the quiz. b. John - he is a clever guy - won the quiz. In addition, Quirk et al. (1985:131<) show that appositions can be preceded by sentential adverbs, another indication that appositions are sentence-like. (8) a. Norman Jones, then a student, wrote several bestsellers. b. They elected as chairman Martin Jones, also a Cambridge graduate. An important finding supporting the idea that appositional constructions not only involve multiple sentences, but also a relation of coordination between them, is the possibility to express a separate illocutionary force. The example in (9a) is from Corazza (2005:13)= those in (10) are from Verstraete (2005:61<). (9) a. Is Jane, the best doctor in town, already married? b. ?Jane, perhaps the best doctor in town?, is already married. (10) a. John was imprisoned, but did he really rob the bank? b. John was imprisoned, but don©t forget that he robbed the bank> Clearly, if a matrix sentence is interrogative, the apposition is not included in the question. The secondary content in that case is usually assertive. It seems to me that it is even possible to some extent to have an interrogative secondary content if the primary content is assertive, by using question intonation for the apposition (9b). Verstraete (2005) claims that a distinct illocutionary force is the most important difference between coordination ' 72 and subordination of sentences, which supports the idea of including appositions in the former category. The word order in Dutch apposition markers provides a further argument for sentence coordination. In contrast with subordinate clauses, coordinated sentences have the finite verb in the second position. Complex Dutch apposition markers including a verb, like dat is (`that is'), and dat wil zeggen (lit. `that wants to say') show verb-second order, too. Finally, the semantic interaction between the apposition and operators in the host sentence is cross-clausal (cf. Dever 2001). For example, elements in the apposition generally cannot be in the scope of a quantifier in the anchor (11a). Also, the secondary content stays out of the scope of negation in the matrix sentence (11b). (11) a. *Every child, a communicative being, starts making noise. b. John did not kiss Mary, his girl friend. The appositions behave as if they were separate sentences with a discourse anaphor referring to the anchor. (12) a. Every child starts making noise. *It is a communicative being. b. John did not kiss Mary. She is his girl friend. Del Gobbo (2003) and Nouwen (2007) indeed analyse appositional constructions as involving a discourse anaphor in the apposition, using an e-type strategy and dynamic semantics respectively. It seems reasonable, then, to conclude that coordination in appositional constructions connects sentences. There are empirical findings which suggest that the connection of coordination in appositional constructions not only operates between the matrix sentence and the appositional proposition, but also between the anchor and the apposition itself, on the constituent level. The first indication 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 (13). = = 73 (13) 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. Furthermore, both coordination and apposition can combine elements of all possible syntactic categories (see for example Burton-Roberts 1975). The restrictions on these combinations are on the constituent level. (1<) a. John reads and writes books. (verb) b. John writes books, but Mary only reads them. (sentence) (15) a. John reads, or rather devours, books. (verb) b. John reads books, that is to say he devours them. (sentence) A well-known restriction is the so-called law of coordination of likes, which states that the coordinated elements must belong to the same category. This was first formulated in terms of syntactic categories by Williams (1981). Later on however, several people (a.o. Munn 1993), have shown that this law has a semantic nature. Compare the examples in (16) from Munn (1993:(3.22)) to their appositional counterparts in (17). The a-sentences show that elements with different syntactic categories can be combined. The b-sentences are infelicitous, because the combined elements belong to different semantic categories. (16) a. John is sick and in a foul mood. b. *John is sick and in the park. (17) a. John is sick, in a foul mood. b. *John is sick, in the hospital. A final observation supporting the idea of coordination on the constituent level in appositional constructions concerns case marking. Just like in normal constituent coordination, the apposition either gets the same case as the anchor, or the default case of the language (cf. De Vries 2007= SchZtze 2001). The German examples in (18ab) are from Durrell (1996:<2), the Czech example in (18c) is from Radek Sim#k (p.c.) and the English and Norwegian examples in (19) are from SchZtze (2001:210, 21<, 226, 227). For English and Norwegian, accusative is the default case. ' 7< (18) a. Es spricht Herbert Werner, der Vorsitzende des Vereins. it speaks Herbert Werner-NOM, the chairman-NOM the society-GEN `The speaker is Herbert Werner, the chairman of the society.' b. in Michelstadt, einem kleinen St\dtchen im Odenwald ¼ in Michelstadt-DAT, a little town-DAT in the Odenwald-DAT ¼ c. Pan Nov%k nakreslil tuto m#stnost tu4kou, sv;m obl#ben;m n%strojem. Mr Novak drew this room-ACC pencil-INSTR his favourite instrument-INSTR. `Mr Novak drew this room with a pencil, his favourite instrument.' (19) a. The best athlete, her/*she, should win. b. Did your parents or him/*he pick up Mary? c. 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.' (Johannesen 1998, citing Berntsen & Larsen 1925) d. Laereren sa at den smarteste studenten, alts^ meg/jeg, the-teacher said that the smartest student, thus me- ACC/I-NOM skulle gi en tale. should give a speech. `The teacher said that the smartest student, namely me, should give a speech.' In summary, I have argued that appositional constructions involve a connection realized in the form of coordination. This connection coordinates elements on two levels. On the sentence level the main proposition is combined with an appositional proposition. The truth values of the two sentences are not combined as in normal sentence conjunction. Rather, the construction can be compared to a sequence of main sentences, where the second refers to the first with an anaphor. On the constituent level the anchor and the apposition are combined, resulting in = = 75 the appositional proposition. This level of coordination is visible in the syntactic features of the construction, such as case marking. 3. Apposition involves predication In the previous section, I concluded that appositional constructions use coordination to combine propositions expressing primary and secondary content. This leaves us with the question what the secondary content is. Considering the description "[the anchor] is [the apposition]" given before, this seems to be predication in the form of a copular sentence. Various proposals indeed assume that the apposition has to be analysed as a predicate of either the anchor (Potts 2007) or a co-indexed discourse anaphor (Del Gobbo 2003= Nouwen 2007), resembling the subject of a copular construction. The most exentensive argumentation for a relation of predication between the anchor and the apposition comes from Doron (199<). She shows that appositions are like predicates and unlike arguments. Appositions, for example, allow i-within-i constructions (20), make it possible to use profession NPs without an article (21), and are negated like predicates (22). The examples are from Doron (199<:55/6). (20) 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. (21) a. George Washington, President of the Union, planted a cherry tree. b. We elected him President of the Union. (22) Orville Wright, not Wilbur, made the first flight at Kitty Hawk. The arguments are strong enough to conclude that appositional constructions can be analysed as copular sentences. It is well known, however, that there are different types of copular sentences (e.g. Higgins, 1979). Therefore, I investigated whether these types are all possible in appositional constructions. It turned out that this is indeed the case and that the type of predication correlates with the semantic class of apposition. In order to show this, I will shortly introduce the semantic classification of appositional constructions (Quirk et al 1985= Heringa & De Vries to appear) and the typology of copular sentences. Semantically, appositional constructions can be divided into three main classes: identification, inclusion and attribution. These classes express different relations between the anchor and the apposition. In an identificational construction, the apposition is another way of describing ' 76 the anchor, in an inclusive construction, the apposition describes a part of the anchor, and in an attributional construction the apposition expresses a class to which the anchor belongs, thus adding the property of belonging to that class to the anchor. Consider the following examples: (23) a. My only brother, Pieter, is a member of the student council. (identification, _) b. He likes to talk to important people, for example the dean. (inclusion, {) c. His girl friend, a modest person, laughs about that. (attribution, |) These classes differ in the relative specificity between the anchor and the apposition (Heringa & De Vries to appear). In general, the anchor and the apposition are equally specific in a relation of identification, in a relation of inclusion the apposition is more specific than the anchor and in attribution the apposition is always generic, independently of the specificity of the anchor. The classic typology of copular clauses is that of Higgins (1979). He describes four types of copular clauses: predicational, specificational, identificational and equative. To keep things simple, I will take the last three groups together under the term specificational. To get an idea of the distinction between predicational and specificational copular clauses, consider the examples in (2<) and (25), from Partee (1998:363): Predicational copular sentences: (2<) a. Helen is a teacher. b. My best friend is tall. c. What I©m giving to Sean is in the car. Specificational copular sentences: (25) a. The only thing he eats is junk food. b. The number of planets is nine. (Higgins 1979) c. What I don©t like about John is his tie. (Higgins 1979) This distinction can be described as follows. Predicational sentences add a property to the subject, whereas specificational sentences give a value to a variable introduced by the subject. The predicate in specificational sentences corresponds to a conventional answer to the subject if one has to fill in a form. = = 77 If we compare the classes of appositional constructions and the types of copular sentences, it is immediately clear that constructions in the class of attribution show a parallel with predicational copular sentences. The apposition in these cases denotes a property that is applied to the referential noun phrase in the anchor. The second proposition in (23c) can be described as in (26), a predicational sentence. This can be done for all attributive appositions. (26) His girl friend is a modest person. Note that all attributive appositional constructions can be paraphrased with an appositional relative clause using a form of the copula to be. For (22c), the paraphrase is given in (27). (27) His girl friend, who is a modest person, laughs about that. Next, consider the examples of identificational appositions in (28) and the description of their secondary content in (29): (28) a. The lion, the panthera leo, is threatened with extinction. b. The youngest animal in the zoo, this lovely ape, jumped on John©s shoulder. c. John is going to paint a car, Bill©s red BMW. d. John wanted to buy a car, a red BMW, but the garage didn©t have one. (29) a. The lion is the panthera leo. b. The youngest animal in the zoo is this lovely ape. c. The car John is going to paint is Bill©s red BMW. d. The (type of) car John wants to buy is a red BMW. Again, there is a clear correlation between this class of apposition and certain copular clauses, namely the specificational type.3 In these constructions, paraphrasis with an appositional relative clause is generally not possible. Compare (28cd) and 30(ab). (30) a. *John is going to paint a car, which is Bill©s red BMW. 3 If Den Dikken (2006) is correct in saying that some specificational copular clauses (and especially equatives) involve predicate inversion in their structure, a part of the class of identification may not only differ semantically from attribution, but also syntactically. ' 78 b. *John wanted to buy a car, which is a red BMW, but the garage didn©t have one. Also, the interpretation of the secondary content is more complex in this class. If the anchor is indefinite, as in (28c/d), the subject of the copular clause does not only contain the anchor, but also includes the anchor©s scope (cf. Doron 199<). For the c-example this can be explained if a discourse anaphor is used (as in the analyses of Del Gobbo 2003 and Nouwen 2007), because such an anaphor also includes the scope of a quantifier in its reference= see (31a). However for the d-example this does not work, see (31b). (31) a. John is going to paint a car. It is Bill©s red BMW. b. John wanted to buy a car. }It is a red BMW. Though the second sentence in (31b) is not impossible, it does not give the reading in (28d), where a car is non-specific. This problem is related to the fact that the subject in a sentence like (29d) has to be interpreted as an individual concept (cf. Janssen 198<). Unlike a referential nominal phrase, an individual concept does not refer to a specific element, but its reference can vary over a group of entities, depending on time or location. In that way, it refers to a class, like generics. The car John wants to buy does not refer to one specific car, but to a class of cars, one of which John wants to buy. An anaphoric pronoun, however, is specific and can not refer to an individual concept. The question then remains, whether analyses of appositional constructions involving a discourse anaphor are on the right track. These analyses explain the semantic interaction between the anchor and the apposition, as shown in (11/2), but cannot account for the existence of sentences like (28d). A possible solution might be that an abstract anaphor is used. This would explain the crossclausal behaviour that does not have the reference restrictions of normal pronouns. The third class of appositional constructions is inclusion. Below I give some examples and a description of their appositional proposition again: = = 79 (32) a. The ape, especially the orang-outang, is threatened with extinction. b. John wants an exotic pet, for example an ape, for his birthday. c. Bill met a group of his class-mates, including Mary. (33) a. The ape especially threatened with extinction is the orang-outang. b. The (kind of) exotic pet John wants for example is an ape. c. The group of his class-mates Bill met is including Mary. The examples show that the class of inclusion is close to the class of identification. The type of copular clauses involved in these cases is again specificational. In this case, however, also if a definite anchor is used the ©scope© has to be taken into account (32a). What is also special in this class of appositional constructions is that the apposition marker is obligatory. Unexpectedly, in some cases the apposition marker appears on the subject-side of the copular clause. This challenges the idea that it forms a constituent with the apposition. Future research is needed to find out what this means for the structure of appositional constructions. Despite these remaining issues, the general conclusion of this section is clear. Appositional constructions indeed involve predication and the type of predication correlates with the class of appositional construction. Attributive appositional constructions are interpreted as predicational copular clauses, and both identificational and inclusive appositional constructions are interpreted as specificational copular clauses. The subject of the corresponding copular clauses does not include the anchor alone, but also its scope. 4. Conclusion and future research This paper argues for an account of appositional constructions in terms of coordination. It provides data suggesting that coordination in this type of constructions operates at two levels at once. On the sentence level, it combines two propositions, one expressed by the matrix sentence and one that describes the relation between the anchor and the apposition. On the constituent level, it combines the apposition and the anchor. This combination results in the second proposition, which expresses a relation of predication. It is shown that the type of predication corresponds to the semantic class of apposition. These observations lead to several interesting questions. How is it possible that coordination operates on two levels at the same time and how these two operations are divided over ' 80 syntax and semantics? How can it be explained that coordination of the two propositions is not interpreted as in normal sentence conjunction? Furthermore, the correspondence between predication and apposition has to be worked out in more detail. It is especially interesting what exactly functions as the subject. It is clear that not only the anchor, but also its scope has to be taken into account. Assuming a discourse anaphor in the structure could explain this, but this idea faces for instance the problem that pronouns cannot refer to non-specific elements. Finally, the role of apposition markers needs further study. These words, or sequences of words need a position in the structure. Some of them are so complex that they may even have an internal structure, possibly resembling the complete structure of appositional constructions. Also, it has to be investigated how the different meanings these words make explicit can be interpreted from the construction in general. References Berckmans, P. (199<). `Demonstration, apposition and direct reference'. Communication and Cognition 26, <99-512. Berntsen, M. & A.B. Larsen (1925). Stavanger bym$l. Oslo: Utgitt av Bym^lslaget, I kommisjon hos H. Aschehoug & Co. Burton-Roberts, N. (1975). `Nominal apposition'. Foundations of language 13, 391-<19. Corazza, E. (2005). `On epithets qua attributive anaphors'. Journal of Linguistics <1, 1-32. Del Gobbo, F. (2003). Appositives at the interface. Dissertation, University of California, Irvine. Dever, J. (2001). `Complex demonstratives'. Linguistics and Philosophy 2<: 271-330. Dikken, M. den (2006). Relators and linkers. The Syntax of Predication, Predicate Inversion, and Copulas. Cambridge, Massachusetts: MIT Press. Doron, E. (199<). `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. Durrell, M. (1996). Hammer's German grammar and usage. London: Arnold, 3d edition. Geist, L. (to appear). `Predication and equation in copular sentences: Russian vs. English'. In: I. Comorovski & K. von Heusinger Existence: Semantics and Syntax. Springer. = = 81 Haspelmath, M. (to appear). `Coordination'. In: T. Shopen (ed.) Language typology and syntactic description. 2nd ed. Cambridge: Cambridge University Press. Heringa, H. and M. de Vries (to appear). `Een semantische classificatie van apposities'. Nederlandse Taalkunde. Higgins, F.R. (1979). The pseudo-cleft construction in English. New York: Garland Publishing. Janssen, T.M.V. (198<). `Individual concepts are useful'. In: F. Lantman & F. Veltman (red.), Variaties of formal semantics. Dordrecht: Foris. Johannesen, J.B. (1998). Coordination, New York: Oxford University Press. Kraak, A. & W. Klooster (1968). Syntaxis. Culemborg: Stam- Kemperman. Munn, A. (1993). Topics in the syntax and semantics of coordinate structures. Dissertation. University of Maryland, College Park. Nouwen, R. (2007). `On appositives and dynamic binding'. Research on Language and Computation 5(1), 87-102. Potts, C. (2007). `Conventional implicatures, a distinguished class of meanings'. In: G. Ramchand and C. Reiss (eds.) The Oxford Handbook of Linguistic Interfaces. Oxford: Oxford University Press, <75-501. Partee, B. (1986). `Ambiguous pseudo-clefts with unambiguous be'. In: S. Berman, J. Choe and J. McDonough (eds.) Proceedings of NELS 16. Amherst: GLSA, 35<-366. Partee, B. (1998). `Copula Inversion Puzzles in English and Russian'. In: K. Dziwirek, H. Coats and C. Vakareliyska. (eds.) Annual Workshop on Formal Approaches to Slavic Linguistics: The Seattle Meeting. Ann Arbor: Michigan Slavic Publications, 361- 395. Quirk, R., S. Greenbaum, G. Leech and J. Svarvik (1985). A comprehensive grammar of the English language. London: Longman. Schelfhout, C.R.M., P.A.J.M. Coppen and N.H.J. Oostdijk (to appear). `A parenthetical approach to conjunction reduction'. Leuvensche bijdragen. SchZtze, C.T. (2001). `On the nature of default case'. Syntax <(3): 205- 238. Verstraete, J.-C. (2005). `Two types of coordination in clause combining'. Lingua 115(<), 611-626.