Derived Coordination

Philipp Weisser

• Publisher: De Gruyter
• Year: 2015
• ISBN:
  9783110435313
• Edition: 1
• Topic: Filología
• Subtopic: Sintaxis en lingüística

This monograph explores the different types of clausal relations in the world's languages. In the recent literature, there have been claims that the strict dichotomy of subordination and coordination cannot be maintained since some constructions seem to be in between these two categories. This study investigates these constructions in detail.

The first part is concerned with clause chaining constructions, while the second is concerned with different cases of asymmetric coordination in English. In both parts, it is shown that the different tests to distinguish clausal relations indeed yield different results for the specific constructions. This poses a severe challenge for the established theories of clausal relations. However, as it is argued, recent analyses of coordination provide for the possibility to map a subordinate structure onto a coordinate one by means of regular transformational rules. It is shown that a single movement step derives all the peculiar properties of the phenomena in question.

This book thus provides the first comprehensive solution for a long-standing problem in theoretical syntax.

___

Part I: Clause Chains, Medials, and Converbs

1 Introducing Clause Chains

Clause chaining constructions are long sequences of a potentially infinite number of clauses within the same sentence. These constructions are attested in languages all over the world but they are especially widespread in East and Southeast Asia, Papua New Guinea and Australia.

The individual clauses in clause chaining constructions are usually quite small consisting of a verb and its arguments. Since the arguments (especially subjects) are usually dropped if they can be inferred from discourse, many clauses consist of nothing but a verb. Morphologically, the major characteristic of clauses within a clause chain is that all verbs but the final one bear incomplete marking. That is, they are not inflected for categories usually associated with finiteness such as tense or mood. This is illustrated in (1).

‘When theyi (the foreigners) arrived, theyj (the villagers) got up and chased them away. Theyi threw away their stuff and fleed. Then, theyj stole their stuff.’

Kâte (Trans-New-Guinea); Pilhofer (1933) as cited in Bickel (2011)

Example (1) consists of six distinct clauses, but only the verb of the final clause is marked for absolute tense. In the case above, it is marked for REMNANT PAST, indicating that the action described took place a while ago. The first five verbs can be marked for other categories such as relative tense but, crucially, they do not bear the full specification. Interestingly, however, even though these verbs are not marked for REMNANT PAST as well, they are obligatorily interpreted as if they were. The events they describe happened at the same time as the events described in the final clause.

Clauses within a clause chain are juxtaposed asyndetically, without coordinating or subordinating conjunctions. Instead, the verbs within a clause chain are marked for switch-reference, which is often taken to be a characteristic for these constructions.

There are two different types of clauses usually contained in a clause chaining construction: Medial clauses and converb clauses. They can be distinguished on the basis of the function they fulfill. The main function of converb clauses is to mark adverbial subordination.1 This contrasts medial clauses whose main function is to provide additional foreground information. In other words, medial clauses “provide mainline information to move the discourse forward”.2 These kinds of definitions are, of course, very vague, and often a definite decision of individual cases seems problematic. Hence, in this work, the definitions above are not taken to be the final criteria to decide whether a clause is a converb or a medial clause. Rather, these statements serve as a rough point of orientation. The decisive criteria, however, will be given in the following chapters. As the first part of this work will show, medial clauses differ from converb clauses with respect to certain syntactic tests. Accordingly, both medial clause and converb clause can be defined on the basis of their syntactic behavior rather than on the basis of their function. A converb clause is syntactically subordinate to another clause, whereas a medial clause is, according to various syntactic tests, in between subordination and coordination.

These definitions are clear and discrete and thus, the terms are much more concrete. However, one should be careful to transfer these terms to other works. In the literature about clause chains, medial constructions and converbs, these terms are often used in completely different ways. To a certain extent, these terms are also used interchangeably. The same construction often has different names in different languages. The major reasons for this confusion are different terminological traditions in different language families. The term converb was originally used for verb forms in Turkic and Mongolic languages, whereas the term medial comes from Papuan linguistics.

The term clause chaining refers to the whole construction which can include medial clauses and/or converb clauses. Thus, the term is neutral with respect to the question of clausal relations.

In the course of the following chapters, I discuss the phenomenon of clause chaining constructions in detail. I give an overview of the morphosyntactic properties of this construction from a typological, theory-neutral perspective. Then, I discuss the implications of the previous findings for syntactic theories and propose a novel theory couched in the framework of the Minimalist Program.

More concretely, the discussion of clause chaining constructions will proceed as follows: In the next chapter, I will illustrate the syntactic behavior of medial clauses with respect to the standard tests of clausal relations. As was hinted at in the introduction above, the results will be inconclusive. Medial clauses are neither fully subordinate nor fully coordinate. Hence, they pose a challenge for syntactic theories. In Chapter 3, I will discuss how this challenge was tackled by previous analyses and why these analyses are problematic. In Chapter 4, I present my own analysis and show how it can derive the inconclusive syntactic properties of medial clauses. Chapter 5 shows how the analysis can handle long sequences of medial clauses. The empirical facts about the dependency relations in these long sequences will serve as further strong arguments for my theory. In Chapter 6, I discuss the syntactic properties of converb clauses and how the theory of Chapter 4 can account for their behavior as well. Chapter 7 is concerned with two case studies of clause chains in Tsakhur and Korean. In Chapter 8, I show how the present approach can be parametrized to account for the attested empirical variation we find with clause chaining constructions. Finally, Chapter 9 will briefly discuss what the present theory predicts for another topic that is closely related to the topic of clause chaining constructions, namely switch-reference marking.

2 Properties of Medial Constructions

In this chapter, I introduce the general morphosyntactic properties of medial constructions. Then, I will focus on the question whether these constructions are structurally coordinate or subordinate. For this purpose, I will apply the standard tests which distinguish coordinate and subordinate structures to medial constructions in different languages. These tests yield, as we will see, contradictory results and hence support the claim that clause chaining constructions are neither canonically subordinate nor canonically coordinate.

2.1 The Morphological Form of Medial Clauses

We have seen a prototypical example of a medial clause chain from Kâte above. A similar example from Tauya, another clause chaining language, is given below. The following sentence consists of nine distinct clauses and, again, only the final verb is fully specified (for indicative mood). In Tauya, all non-final verbs bear nothing but a marker glossed as medial, which also encodes the category of switch-reference. Again, the first eight verbs cannot be inflected for tense or mood but nevertheless they are all interpreted as if they bore the indicative mood marker located at the final verb.

'She carried the child and came up and stayed; and they killed the pigs and showed them to the children and they showed them the sacred flutes and stayed and they killed the pigs and put them and they danced and cut (the pigs).'

Tauya: MacDonald (1990, p.361)

The previous examples showed that medial clauses are (usually) inflected. Often, they are inflected for relative tense as in the Kâte example above (1). These markers encode whether the event described by the medial clause and the event in the following medial clause (or the final, finite verb) happen simultaneously or successively. In many languages, medial verbs are also inflected for person and number features of their respective subjects. In both examples above, the medial verbs are also marked for switch-reference, another category frequently found on verbs within a clause chain. According to the standard definition of switch-reference, these markers indicate whether the subject of a clause is identical to the subject of the immediately following clause. In case of identical subjects, the verb bears a same subject marker (glossed as SS), whereas in case of non-identical subjects, a different subject marker appears (glossed as DS). As one can see in the example from Tauya (in (1)), the non-final verbs in clause chaining constructions can also bear a so-called medial marker, which only indicates that the marked verb is a medial verb, i.e. that it is dependent on the last verb of the chain.

There are a number of other categories which occasionally appear on medial verbs. Usually, medial verbs can be marked for voice/diathesis such as passives, causatives, etc. Not all (but some) languages allow for medial verbs to ...