The modern interpretation of Jespersen's theory of grammar
著者（英） Yumi Kawamoto
journal or
publication title
Doshisha literature
number 35
page range 101‑116
year 1992‑03‑10
権利（英） English Literary Society of Doshisha University
URL http://doi.org/10.14988/pa.2017.0000014759
THE MODERN INTERPRETATION OF JESPERSEN'S THEORY OF GRAMMAR
YUMI KAWAMOTO
I
J aspersen investigated grammars of many languages and set forth extensive grammatical theory with a wide range of data in many languages including English, Danish, German, French, Spanish, and Latin. Many of his original ideas have been discussed by succeeding linguists: some accepted and developed his ideas, whereas others simply rejected or ignored them. For example, structural linguists, who came to the front just after J espersen, severely criticized him. On the other hand, many of the generative grammarians, who have been led by Chomsky, have developed arguments in favor of J espersen .
. . . O. Jespersen argues, on what we will reconstruct as transformational grounds, that "the doctor's arrival" is different in structure from "the man's house," despite superficial similarity, because of its relation to the sentence "the doctor arrives." This observation, which, I think, is entirely correct, is criticized by E. A. Nida as "serious distortion and complication of the formal and functional valuesd
There are, in fact, a number of correspondences between Jespersen's theory of grammar and the theory of generative grammar. Even the current generative approach, which is usually called the Principles-and-Parameters theory, or the GB (Government and Binding) theory, seems to be taking advantage' of some of J espersen' s ideas. This paper aims to explore the validity of J espersen's insight in modern synta'ctic theory. The immediately
ClOl]
following'section is to present a rough sketch of
J
espersen' s grammatical theory: three ranks, junction, and nexus, Section III is to discuss the modern interpretations of Jespersen's nexus, and the following section is to investigate the conceptual parallelism between J espersen' s grammar and generative grammaL Through the discussion of the relation between the two theories of grammar, the contemporary values of Jespersen's theory of grammar will, I hope, become apparent.II
Jespersen's grammatical theory is based on rank, junction and nexus, He strictly distinguishes the traditional semantic classification into parts of speech from functional classification into ranks, Thus-, in a combination of words such as a cannon ball, cannon and ball are both classified as nouns in terms of parts of speech, but their functions are quite different: ball expresses the central idea of the word group, while cannon functions simply as a modifier that defines or specifies the central word ball. He referred to the central word of a word combination as a primary, to a word that defines or specifies the primary as a secondary, and a word that defines or specifies the secondary as a tertiary, Thus, in a noun phrase such as the furiously barking dog, dog functions as a primary, barking as a secondary, and furiously as a tertiary,
Moreover, J espersen discusses the relation of the two word groups the furiously barking dog and the dog barks furiously, He points out that the two word groups express the same ideas, and that the words that constitute them have the same rank in each, with dog primary, barks and barking secondary, and furiously tertiary, The former word group he calls junction, and the latter nexus,
He does not overlook the fundamental difference between junction and nexus, Junction is based on an incomplete ideas, he notes, which makes us
expect a continuation. Nexus, on the other hand, exhibits a complete meaning, having subject and the predicate, where the combination is based on predication.
III
In the analysis of sentences such as(1) I like boys to be quiet.
Sweet (1891)2 treated only boys as the object of the verb like, and regarded to be quiet as a grammatical adjunct to the object boys. Sonnenschein (1923)3 considered the sentence a double object construction, like He asked me a question, where both boys and to be quiet are the objects of the verb like.
]espersen criticizes Sweet by pointing out that the sentence (1) is not semantically the equivalent to the sentence (2), where the relative clause counts simply as a semantic adjunct to the object boys,
(2) I like boys who are quiet.
so that to be quiet in the former is not simply an adjunct to the object4 He also argues against Sonnenschein, claiming that the double object construction (3a) has the same meaning as the sentence (3b), where the dative object is omitted, whereas the sentence (1) is totally different in meaning from (4), where one of the objects is omitted as in (3b).5
(3) a. He asked me a question.
b. He asked a question.
(4) I like to be quiet.
Instead, Jespersen argues that the whole word group boys to be quiet constitutes the object of the verb like, which he specifically .refers to as nexus-object6
This analysis is quite consistent with the recent generative analysis of an infinitival complement clause, which is assumed to be the object of the matrix verb as a whole and to consitute an Ip7. Thus, The sentence (1) has the following representation with boys and to be quiet being constituents.
(5) IP
like CP
IP
boys VP
to be quiet
Moreover, J espersen extends the concept of nexus to constructions where the primary is missing as in (4). In (4), the object of like should be the nexus
~ to be quiet, if we consider the nexus boys to be quiet the object of like in (1).
Although the primary, i. e., the subject of to be quiet, is not explicitly expressed, J espersen notes, there should be an understood subject, namely, I in this case8 The same consideration holds in constructions with a gerundive complement:
(6) a. I like [his traveling].
b. I like [ traveling].
In both (6a) and (6b), the bracketed word group counts as the nexus-object. In
105 (6a), his is a primary and traveling a secondary. Correspondingly, J espersen notes, in (6b), although the primary is unexpressed, there should be the understood primary, namely, I in this case9
.The recent generative approach has been mainly concerned with the analysis of null NPs10 Sentences such as (4) and (6b), where the embedded infintival subject is not explicitly expressed, are assumed to have the following representation, having a null pronoun PRO(noun) as the embed- ded infinitival subject:
(7) a. I like [PRO to be quiet].
b. I like [PRO traveling].
It is reasonable to assume that J espersen's analysis that posits a primary, though it is unexpressed, triggered the study of null NPs by recent generative grammarians. As J espersen notes, though it is not phonetically realized, the embedded subject (in J espersen's term, the primary in the nexus-object) is present as a PRO.ll A PRO is a kind of pronominal, which is phonetically null, and the theory that relates a PRO with its antecedent is called the Control Theory.
Furthermore, Jespersen notes that understood primary in (8a-b) is not identical to that in (1) and (6b).
(8) a. To travel is not easy nowadays.
b. Traveling is not easy nowadays.
In (1) and (6b), the unexpressed primary of the nexus-object refers to the element in the main clasuse, that is, 1, whereas in (8a-b) the primary of the nexus cannot be related to any element in the main clause, and hence the generic interpretation12 Taking over ]espersen's analysis, the recent
Control The~ry distinguishes obligatory control as illlustrated in (1) and (6b) from arbitrary control as shown in (8a-b).13 Determining which environments require or permit obligatory control and which require or permit arbitrary control is the keystone of the study of the Control Theory.
J espersen further remarks that in passive constructions such as (9), though the grammatical subject is he, semantically the subject of is believed is not simply he but the whole nexus he to be guilty, since what is believed is not he but the fact of his being guil ty14
(9) He IS believed to be guilty.
It is evident that this idea is the theoretical basis of transformation, especiallY NP-movement. In the generative framework, (9) is assumed to have the following D-structure representation.
(10) be believed [he to be guilty].
Note that the bracketed word group is exactly the same as the one that J espersen claims to be the semantic subject of is believed He ii:t the embedded infinitival clause moves up to the matrix subject position through NP- movejent, and hence the generation of (9).
Raising predicates such as seem and . likely are treated in the same way as passive constructions by Jespersen. In (lla) and (llb), what is supposed and what is thought to be likely is his working hard and his coming, respectively, not simply he.l5
(11) a. He seems to work hard.
b. He is likely to come.
107 This idea is also compatible with recent analysis of xaising predicates, where NP-movement is assumed to be involved as well as passive constructions.
The D-structure representation of (lla-b) would be as follows:
(12) a.
b.
seem [he to work hard].
be likely [he to come].
He in the infinitival embedded clause moves up to the matrix subject position, resulting in (11a-b).
Jespersen points out that the word group (13a) is semantically equivalent to the nexus with a finite verb (13b)16
(13) a. the doctor's arrival b. The doctor arrived.
The doctor counts as the subject and arrival as the predicate. The doctor and arrival are connected in essentially the same way as the complete sentence (13b).17 The same is true for what Jespersen calls the predicative substantives exemplified with (14a):
(14) a. the doctor's cleverness b. The doctor is clever.
The relation of the doctor and cleverness in (14a) corresponds to that of the doctor and clever in (14b)18 The proposal that emerges from such considera- tions, then, is that expressions like (13a) and (l4a) must be classified into nexus. J espersen calls such expressions nexus-substantives 19 This insight has led generative grammarians to assume Nominalization Transformation, though today most generative linguists are skeptical whether the transform a-
tion is indeed present.20 At any rate, it is worth noting that what are now spoken of as derived nouns exemplified with (13a) and (14a) must be distinguished from nouns like the doctor's watch. This idea of relating derived nominals such as (13a) and (14a) to clausal expressions leads to the DP hypothesis on nominals set forth by Abney (1987)21 From the fact that nouns normally require specifiers (quantifiers, determiners, or possessive NPs, etc.) as shown in (l5a-c),
(15) a. *1 read book.
b. I read no book.
c. I read the book.
d. I read my book.
the DP hypothesis claims that all nominal expressions are projections of DP, where an NP occurs as a complement of the head D, on analogy to clausal expressions, which are projections of IP, the head of which selects a VP as the complement. The structure of (13a), which was treated as a nexus- substantive by Jespersen, would be as follows in the generative approach equipped with the DP hypothesis:
(16) DP
SPEC
the doctor D
's
D'
NP I N'
N I arrival
Note this structure exactly corresponds to the one the clausal expression (13b) has22
(17) IP
SPEC
the doctor
AGR I'
VP
V'
I
V I arrived
One can eas~ly imagine that ]espersen's idea of nexus-substantives, which tried to relate nominals to clauses, played a vital role in the development of the DP hypothesis.
IV
The preceding section has sketched how J espetsen's grammatical theory, especially his idea of nexus, has affected succeeding linguists. In this section, J espersen's influence on conceptual grounds will be shown.
J espersen remarks that formulas and free expressions in languages must be distinguished from each other.23 The former are handled simply by memory or repetition of what one has once learned, whereas the latter are created through patterns or types which have come into existence in one's subconsciousness as a result of one's having heard the speech of many others. Thus, while one learns expressions such as How do you do? and Beg your pardon? as a unit, one can create sentences that are different in some one or more respects from anything one has ever heard or uttered before, through the process of constructing patterns or types.
Although structual grammar, which flourished just after J espersen constructed his theory and just before Chomsky wrote Syntactic Structures in 1957, completely rejected Jespersen's idea because structural linguists assumed that language acquisition took place entirely by stimulus, response, and reinforcement based on behaviorism under the leadership of Skinner, Chomsky adopted it. If the acquisition of language was nurtured only by experience, i. e., by hearing adults' speech and having the errors corrected by adults, it would mean, like parrots, we could speak and understand just the words, phrases, and sentences that we have come across before. This is evidently not true. Unlike parrots, we can produce and understand new sentences and make judgment about their well-formedness. All native speakers have the ability to form, interpret, and pronounce sentences that they have not encountered before:
The most striking aspect of linguistic competence is what we may call
111 the 'creativity of language,' that is, the speaker's ability to produce new sentences, sentences that are immediately understood by other speakers although they bear no physical resemblance to sentences which are familiar.24
In fact, this idea of language faculties has been the basis for generative grammar, and even the current framework of the Principles-and-Parameters approach takes its stand on this idea of
J
espersen's .. . . what is actually represented in the mind of an individual even under the idealization to a homogeneous speech community would be a core grammar with a periphery of marked elements and constructions25
[Italics mine.]
A core grammar corrensponds to
J
espersen's free expression, and a periphery to a formula. Thus, a core grammar is a grammar determined by experience through fixing the values of a particular language, which are specifically called parameters. A periphery is a marked element that must be learned one by one.The notion of paramenters has been already captured by
J
espersen.Remarking on the pleonastic it26, he observes:
It became an invariable custom to have a subject before the verb, and therefore a sentence which did not contain a subject was felt to be incomplete. In former times no pronoun was felt to be necessary with verbs like Latin pluit, ningit 'it rains, it snows,' etc.; thus Italian still has piove, nevica, but on the analogy of innumerable such expressions as I come, he comes etc., the pronoun it was added in English it rains, it snows, and correnspondingly in French, German, Danish and other languages:
il pleut, es regnet, det regner. It has been well remarked that the need for this pronoun was especially felt when it became the custom to express the difference between affirmation and question by means of word-order (er kommt, kommt er?), . . 27
Languages like English and French must have explict subjects, while those like Latin, Italian, and Spanish need not. The former type of language must fill the subject position with an explicit element that hqs a phonetical content even in the case of verbs rain and snow, which are thought of as a kind of predicate th~t takes no argument. Therefore, the position must be filled with pleonastic elements it or il. The latter type of language, on the other hand, dose not have such dummy elements. Moreover, in the former type of language, the inversion of subject and verb is not usually allowed for declarative sentences since it is kept for questions, whereas the latter type permits verb-subject order in declarative sentences. In order to capture the correspondence with respect to null subjects, pleonastic elements, and inversion of subject and verb, the Principles-and-Parameters theory posits a pro-drop parameter: the variation in the above mentioned properties between languages amounts to a single choice between a pro-drop language or a non- pro-drop language. Once the value of the pro-drop parameter is specified, it applies to all other phases that the parameter is concerned with, and there is no need for a long list for individual properties specifying whether the language in question allows null subjects, whether there are pleonastic elements, and whether the verb-subject order is permitted in declarative sentences.
Another important conception that seems to have affected generative grammar is the notion of universal grammar.
J
espersen raises the question of whether universal grammar is present. Although he avoids asserting yes or no clearly, his concluding remarks of The Philosophy of Grammar reveal that one of the aims of his investigation has been to explore univerval grammar.My endeavour has been, without neglecting investigation into the details of the languages known to me, to give due prominence to the>
great principles underlying the grammars of all languages, and thus to make my contribution to a grammatical science based at the same time
on sound psychology, on sane logic, and on solid facts of linguistic history.28
Chomsky claims that there should be universal grammar, or "the essence of human language.,,29 Human beings are equipped with knowledge of universal grammar as a part of their genetic endowment, so that children are able to acquire the grammar of their native language so rapidly despite of the fact that data they receive are unstructured, random, and deficient.
Jespersen remarks that he has dealt with "the great principles underlying the grammars of all languages" as exhibited in the above quotation. What is worth noting is that Chomsky has also been dealing with the principles underlying universal grammar:
What we expect to find, then, is a highly structured theory of DG [universal grammar] based on a number of fundamental principles that sharply restrict the class of attainable grammars and narrowly constrain their fOrm. 30
v
It is surprising to know that the essence of generative grammar had already been in embryo about 40 years before Chomsky wrote Syntactic Structures. It can even be said that present generative grammar would not have been born if it had not for J espersen's theory of grammar. The most striking fact that distinguishes J espersen's grammar and generative grammar from other linguistic theories is, I think; that the former aims to study not individual languages. For J espersen the aim of study is to make clear the grammar of all languages through dealing with various languages. His aim resides in not constructing a grammar of a particular languages, but finding principles underlying all languages. For Chomsky, the ultimate aim of study is not to describe the grammar of particular languages, either. He has been exploring
abstract grammar that regulates all languages and principles that govern the human mind. Other grammarians, on the other hand, have been attempting to construct grammars of particular languages. American structural linguists, for example, made an effort to set up grammars of American Indian languages by collecting a corpus of data. Functional linguists have been trying to capture the function of certain expressions in a certain speech community.
It is not sensible to judge which approach is better and which approach is . worse. Every scientific field possesses both all-inclusive and intensive approaches, and both approaches contribute to scientific achievement. In economics, for instance, there is macroeconomics, which studies economy extensively, and there is also microeconomics, which scrutinizes individual economic problems. It is important to recognize that either
J
espersen and Chomsky's approach, which pursues "the philosophy of grammar," namely, grammar underlying all languages, or other grammarians' approach, which carefully examines individual languages, is valid and necessary for the clarification of natural language and the human mind.NOTES
N. Chomsky, "A Transformational Approach to Syntax," ed. Archibald Hill, Proceedings of the Third Te= Conference on Problems of Linguistic Analysis in English (Austin, Texas: University of Texas Press, 1958) 124, footnote 2.
2 H. Sweet, A New English Grammar Part 1, (Oxford, 1881).
3 E. A. Sonne~schein, A New English Grammar (Clarendon, 1923).
4 O. Jespersen, The Philosophy of Grammar (London: George Allen & Unwin, 1924) 117.
5 Jespersen 118.
6 J espersen 122.
7 Following, Chomsky, Barriers (Cambridge, Mass.: MIT Press, 1986), full clauses are assumed to be the category CP, with a complementizer C and a propositional content, the category IP.
8 Jespersen 142-143.
9 J espersen 143.
10 Null NPs are NPs that do not have phonetical content. In the Principles-and- Parameters approach nominal expressions, including null NPs, fall into one of the three basic categories, namely, anaphors, pronominals, and R-expressions. In the list below, the item on the right in the parentheses is a null NP.
(i) anaphors (lexical anaphors, NP-traces) pronominals (pronouns, PROs) R-expressions (names, variables)
11 In the early stage of generative grammar, sentences like (7a-b) were assumed to be derived through the process of Equi·NP Deletion, which erased the embedded subject identical to the matrix subject (d. P. S. Rosenbaum, The Grammar of English Predicate Constructions (Cambridge, Mass.: MIT Press, 1967). However, this idea was claimed to be problematic because it would predict that (i) is derived from (ii) and the two sentences have the same meaning, though they, in fact, do not have the same interpretation.
(i) Everyone wants everyone to kiss Mary.
(ii) Everyone wants to kiss Mary.
Now the current theory posits a null NP, PRO, as the embedded subject, which has pronominal features (see note 10, also).
12 J espersen 143.
13 As fof"the two kinds of control-obligatory and arbitrary, see Edwin Williams,
"Predication," Linguistic Inquiry 11 (1980): 203-238; Yumi Kawamoto, "The Formal Nature of Control Relations," MA thesis of Doshisha University, 1991.
14 Jespersen 119-120.
15 Jespersen 12.0.
16 Precisely speaking, sentences like The doctor arrives and The doctor will arrive also correspond to the expression the doctor's arrival, but tense is ignored in the discussion here.
17 Jespersen 115, 120, and 136.
18 Jespersen 136.
19 Jespersen 136.
20 There are arguments against incorporating Nominalization in grammar to
derive derived nominals. See N. Chomsky, "Remarks on Nominalization," eds. R.
Jacobs and P. S. Rosenbaum, Readings in English Transformational Grammar (Waltham, Mass.: Ginn & Co., 1970).
21 1;. Abney, "The English Noun Phrase in Its Sentential Aspect," diss., MIT, 1987.
22 For the further study on structural correlation between nominals and clauses, see Yumi Kawamoto, "On the Status of Gerunds," Studies in Comparative Culture 17 (Kyoto: Japan Association of Comparative Culture, 1991): 44-59.
23 Jespersen 18-24.
24 N. Chomsky, Topics in the Theory of Generative Grammar (Hague: Mouton, 1966) 11.
25 N. Chomsky, Lectures on Government and Binding (Dordrecht: Foris, 1981) 8.
26 Pleonastic it is the one without some sort of "referential function."
27 J espersen 25.
28 Jespersen 344.
29 N. Chomsky, Reflections on Language (London: Temple Smith, 1976) 29.
30 N. Chomsky, Lectures on Government and Binding, 3-4.
