Syntaktische konstrukties in gesproken taal

(1981)–Frank Jansen– rechtenstatus

Summary

In this (socio)linguistic study of the syntax of spoken Dutch, I have confined the scope of the inquiry to the following constructions:

Fronting: the occupation of the sentence initial position by a certain element, as for example the adverb of time in (1):

(1)	Tomorrow-comes-Peter ‘Tomorrow Peter comes’

Stranding: the occupation of sentence initial position by the complement of a preposition, which is stranded in a postverbal position:

(2)	That-sort-of-things-must-I-often-of-think ‘That sort of things I have to think of often’

Sentence initial element deletion. If the sentence initial element is a deictic proform, it may be omitted (the omitted element has been put in parentheses):

(3)	(dat) heb ik al gedaan (that)-have-I-already-done ‘I have done that already’

Left dislocation. A nominal element in sentence initial position may be ‘copied’ by a deictic proform directly before the tensed verb:

(4)	dat probleem dat smap ik niet that-problem-that-understand-I-not ‘That problem, I do not understandit’

Repetition of the tensed verb. The tensed verb may be repeated (together with the cluster of enclitic arguments) after a non-pronominal phrase:

(5)	Ik heb dat probleem heb ik nooit gesnapt I-have-that-problem-have-I-never-understood ‘I have never understood that problem’

These constructions have the following characteristics in common:

1.	they are all confined to declarative main clauses
2.	they are not so infrequent in spoken Dutch
3.	they have something to do with the sentence initial position
4.	with the exception of construction a. (which was incorporated in the investigation for other reasons) all the constructions are fully acceptable only in spoken Dutch and not in the written language.

After a short introduction (Ch.1), I have given the goals of the research in Ch. 2. Since the previous discussions to these constructions were for the most part either anecdotal or based on doubtful intuitive opinions, I was interested in the answers to the following questions:

-	What are the properties of the constructions in spoken Dutch?

-	How frequent are they?
-	What is the best way to describe the constructions?
-	Which characteristics condition the production of them?
-	Is it legitimate to compare the constructions with non-standard phonological variants, are they more like slips of the tongue, or are they constructions in their own right?

The first thing I had to do in order to get an answer was to design a corpus of spoken Dutch, because there was none available when we started our investigation. I interviewed and recorded 40 older (age ca. 50-70) residents of Leiden, a town with ca. 100.000 inhabitants in the western part of the Netherlands (20 km from The Hague). All were born in Leiden, and most of them had lived there their entire life. In Leiden the non-standard variety (the dialect) is not so very different from the standard language. There is, in other words, no language gap between the speakers of standard Dutch and the dialect speakers, and the same holds for the differences between formal and informal speech. Our corpus was stratified in the following way: 10 middle Class men, 10 Middle Class women, 10 Lower Class men and 10 Lower Class women. All were asked the same two series of questions: first a series of formal questions, and after that a series of informal questions. The average length of the interviews was 30 minutes.

With this corpus I was able to investigate the correlational properties of the constructions with the speech situation (formal vs informal), the sex of the speaker (man vs woman) and the socioeconomic class of the speaker (middle vs lower). I used this stratification for my answer to the last question above. Sociolinguistic research on non-standard morphological and phonological variants has revealed the following pattern: those variants are more frequent in informal conversations, men use them more than women, and Lower Class speakers more than Middle Class speakers. If (some of) the constructions have the same pattern, we may consider them to be non-standard constructions.

The remainder of Ch. 3 is devoted to the problem of determining the frequency of the syntactic constructions. I have adopted the following method from Labov (1969, 1972c). First I determine the transformational derivation of a construction C. Then I consider the input of the optional transformation(s) which could give rise to C, as a potential construction. The frequency of C is the ratio of the number of occurrences of C to the total number of occurrences of C and the number of constructions with the potential of giving rise to C.

Ch. 4 is about fronting. I have begun by trying to find the neutral order of elements after the finite verb in the declarative main clause, which I assume to be also the order in the underlying structure. Then I have discussed the arguments for assuming that one specific element, viz. the subject occupies the position before the finite verb in the underlying structure. My conclusion is that the alternative explanation that there is no specific sentence initial element, but a position which has to be filled, is more attractive. The result is the following underlying structure:

Fronting is the preposing of one of the postverbal elements to the empty preverbal position. The first thing we may ask is: which properties of a postverbal element condition fronting (i.e. make it a good candidate for being moved to the sentence initial position)? I have answered this question with the help of a method which I will describe in an example.

One of the possible conditioning factors for fronting is the syntactic function of an element. My hypothesis is that subjects are more frequently fronted than, for example, objects. I tested this hypothesis by counting in a given sub-corpus:

a.	the number of subjects in the sentence initial position
b.	the number of objects in the sentence initial position
c.	the number of subjects in the postverbal position
d.	the number of objects in the postverbal position.

The frequency of the fronting of subjects is computed by dividing a by a+c; likewise the frequency of the fronting of objects is b/b+d. If there is a difference between those two fractions, we used the X²-test in order to find out whether or not this difference is statistically significant.

The results of this section are: fronting is positively conditioned by the activatedness of the element, the definiteness, the syntactic function of the element and the type of constituent. With regard to this last conditioning factor is there a remarkable resemblance of the frequency of fronting to the Language Independent Preferred Order of Constituents (Dik 1978). I also investigated the interrelations between these conditioning factors.

The next section deals with the syntactic description of fronting. My conclusion is, that the word order variants can best be described by assuming two fronting rules and one backing rule. Section 4 is devoted to the sociolinguistic conditions of fronting. I have not found any correlation between the different types of fronting and the speaker's style and background, however.

The last section of this chapter is a study of the stranding construction. First, I have described the types of fronted complements of the preposition which I found in my corpus. Their most salient feature is that the majority are pronominal, while the greater part of the nominal complements are generic. Then I have given the characteristics of the verbs and prepositions which allow stranding. Most of them are prepositional objects, but there are examples of stranding of a preposition in an NP, or several types of Adverbial Phrases. I have also examined the position of the stranded proposition which immediately precedes the (nonfinite) verb(s).

In the second subsection I propose a derivation of Dutch stranding, along the lines of the proposals of Van Riemsdijk (1978) for English stranding. The last subsection is devoted to the correlational aspects of stranding. There are good reasons for believing that stranding is a nonstandard construction: most generative grammarians star the Dutch cases of stranding, and the construction is not allowed in writing. However, I have not found a normal correla-

tional pattern for a nonstandard construction. For example, stranding is more frequent in formal style than it is in informal style.

Chapter 5 is about sentence initial element deletion (SIED). After an introduction in which I distinguished SIED from other constructions, I have traced the conditioning factors for this construction: the form of the deleted element (there are examples of the deletion of personal pronouns, but they are different from the deletions of demonstrative proforms; furthermore I make a distinction between demonstrative pronouns, demonstrative adverbs and the so-called pronominal adverbs, which are more frequently deleted in this order), the syntactic function of the deleted element (objects are deleted more frequently than subjects), the type of speech in which the SIED-sentences is (more frequently in dialogues than in monologues, and they are of a different type). A more or less speculative answer is given to the question why the conditioning factors are as they are. I claim that a SIED-sentence is used when a speaker wants to link his utterance in the most radical way to the previous utterances. Objects have to be more activated than subjects in order to be fronted towards the sentence initial position. Therefore they are more plausible candidates for deletion. This occurs especially in conversations, because there a speaker makes a good impression on his audience by linking his utterance to the previous ones.

The next section is concerned with several possible derivations for the SIED-construction; a derivation by deletion of the demonstrative pronoun in the position directly before the finite verb is preferred. The last section is about the sociolinguistic properties of the SIED-construction. I have found very different patterns for the different types of SIED-sentences. I can not say that such constructions are like nonstandard morphemes in their sociolinguistic pattern.

Ch. 6 is about the Left Dislocation structure (LD). After dealing with some definitional problems, I have traced the conditioning factors for this construction: a pause between the LD-element and the remainder of the sentence conditions LD, but not in an absolute way; the frequency of LD is strongly dependent on the formal properties of the LD-element. The frequency rises from ADVLD via PPLD and NPLD to Dependent Clause LD. The condition that the LD-element must be non-pronominal is nearly absolute. Complex NPs are more frequently dislocated than simplex NPs, and definite NPs more frequently than indefinite ones. Two other possible conditioning factors, the function of the LD-element and its activatedness are probably not at work in this construction. LD seems to be a construction that is conditioned by purely formal properties.

Section 6.3. contains a discussion of the different proposals for the description of LD. There have been numerous proposals in the literature for the description of LDs in several languages. Most of these proposals claim that LDs must be generated by the base-rules. I have tried to show however that the arguments for this claim are not valid for the type of LD that I found in our corpus of spoken Dutch. I have proposed a derivation of our LD by assuming a transformation which inserts a demonstrative proform between the nominal element and the tensed verb.

The last section is on the correlational properties of LDs. The frequencies of the different types of LDs indicate that those types do not display the same sociolinguistic pattern. Independent Clause LD seems to be the standard construction, while PPLD and complex NPLD have more or less the properties of a nonstandard construction. Perhaps, this has something to do with the fact, that LD is the only construction treated in this study which may be called lexico-syntactic: the insertion of a word of a very limited class is crucially involved. All other cases of syntactic variation in which a sociolinguistic pattern could be demonstrated (Labov 1969, 1972c) are lexico-syntactic.

The subject of Ch. 7 is the repeating construction (RC). In the first section I have surveyed the examples in the corpus with two or more finite verbs in the independent clause, where the second verb is a variation of the first one instead of a repetition. The next section is devoted to the conditioning factors: the phrase directly before the second finite verb may not be an unaccented proform; the frequency of RC's increases from 6% for NP's via Adverbs and PP's to 64% for dependent clauses as the element before the second finite verb. Another positive factor is the complexity of the phrase in between. One might think that the longer the elements between the two finite verbs the better, but that is not the case: RC's with more than one part of speech in between have a low frequency. Also the function of the phrase in between is a conditioning factor. For example, the finite verb is more frequently repeated after a subject than an object. Other positive conditioning factors are the presence of other elements after the second finite verb, and the difficulties the speakers had in planning the last part of the sentence. In other words, the RC may be used to gain time when a speaker is not sure how to end his sentence.

The syntactic description of the RC is not obvious. I chose a derivation like gapping, a special case of coordination. In section four I have looked for the correlational properties of the RC, I have not found a clear pattern which indicates that RC is a nonstandard construction.

Ch. 8 is a rather preliminary study of some problems concerning the acceptability of SIED, LD and RC. The goal of this chapter was to find an answer to the following question: are the constructions part of the linguistic competence of the Dutch speakers, or are they comparable with slips of the tongue because they are caused by performance difficulties? It will be clear that I could not find the answer using a test with a direct comparison of our constructions with the ‘normal’ construction, because all subjects would have chosen the normal ones. I tried to get around this difficulty by designing tests in which the subjects had to judge a construction with or without one or more of the conditioning factors. If the sentences with the conditioning factor scored higher than the ones without, we may say that the intuitions about the use of this construction are present in the speakers' competence. If there is no systematic correlation between the scores in the acceptability test and the performance frequency, we may deduce that the construction is primarily caused by performance factors. I used a paired comparison design for the acceptability tests of the SIED-, LD- and Repeating constructions. For the first

two constructions we used the design of Greenbaum and Quirk (1970) for elicitation experiments as well.

The results of this chapter are that SIED is rooted in the competence of the speakers. The same is true for LD, but the RC is intermediate between competence and a peformance error.

In the last chapter I offer some conclusions about the constructions.

Vorige Volgende