Автор работы: Пользователь скрыл имя, 22 Мая 2012 в 21:40, курсовая работа
In contemporary semantics a broad distinction is drawn between denotation (referential) approach and language-intrinsic (or language-immanent) approach. This distinction follows from the opposition of two aspects of meaning: denotation and sense. As a rule the analysis of denotation results in the description of specific properties of extralinguistic objects denoted by a word (e.g. B. Pottier’s analysis of the field siege (chaise, fauteuil, tabouret, canape, pouf – chair, armchair, stool, sofa, pouf) is known to result in the distinction of such concrete and unique denotational components as S1 – with back, S2 – with legs, S3 – for a single person, S4 – for sitting, S5 – with arms, S6 – made from hard material).
1. Theoretical background
2.1.1 Time
2.1.2 Possibility, hypothesis, desirability
2.1.3 Participants
2.1.4 Verb agreement
2.2 American Sign Language
2.3 The category of voice
2.4 The category of mood
2.5 The category of tense
2.6 Palmer’s and mind’s discussion on English modality
Annotation
Bibliography
Sensory: evidence gathered
through sense perception, e.g., seen, heard
Event modality
Deontic: speakers express
conditioning factors that are external to the relevant individual
Permissive: permission is
given on the basis of some authority, e.g. rules, law, or the speaker
Obligative: an obligation
is laid on the addressee(s), also on the basis of some authority
Commissive: a speaker commits
himself to do something; the expression may be a promise or a threat
Dynamic: speakers express
conditioning factors that are internal to the relevant individual
Abilitive: expresses the ability
to do something
Volitive: expresses the willingness
to do something
These notional categories
are discussed and illustrated throughout the book.
The illustrative data reveal
many of the formal means for expressing the notional categories in a
variety of languages. According to Palmer, three grammatical categories
predominate in the expression of the notional categories: (1) affixation
of verbs, (2) modal verbs, and (3) particles. Many of the languages
from which Palmer chose data use more than one grammatical category
to express the notions.
This is probably not unusual.
In fact, the two Austronesian languages with which I am most familiar
spread the notions across all three grammatical categories, and the
lexical and morphosyntactic patterns are completely unlike English patterns,
although the similarity of notions is fairly obvious. I would expect
to see a closer correlation of the grammatical means of expessing modality
among related languages.
Palmer discusses the use of
modal verbs and their association with possibility and necessity in
chapter 4. He draws together issues involving epistemic modality, i.e.,
a speaker’s attitude to the truth value or factual status of a proposition
in contrast to deontic and dynamic modality that refer to unactualized
events. Although notionally there is a difference, Palmer explains that
in English and many other languages, the same modal verbs are used for
both types. He gives three English sentences as examples:
(1) He may come tomorrow.
(2) The book should be on
the shelf.
(3) He must be in his office.
He states that each of the
modal verbs in the sentences can express either epistemic or deontic
modality. However, he goes on to say in a later section that there are
some formal differences: deontic must and may can be negated whereas
epistemic must and may cannot be; if may and must are followed by have
in a clause, they always express epistemic modality, never deontic;
another formal difference between may and must is that deontic may is
replaceable by can and would still express deontic modality, but if
replaced by can’t it would then likely express epistemic modality,
i.e., a truth value. This type of illustration and explanation is used
throughout the book.
Palmer discusses the links
between mood and modal systems with particular respect to languages
that express mood formally, or in combination with modal notions. Although
Palmer suggests that there is basically no typological difference between
indicative/subjunctive and realis/irrealis since both are instances
of mood, he does state that there are considerable differences between
the functions of what have been labeled «subjunctive» and «irrealis»
For that reason he deals with them in three separate chapters.
Although Palmer’s notional
categories make sense, I found that it was difficult to process the
grammatical patterns in the language data used to illustrate the categories.
Part of my difficulty may be attributed to the fact that I believe modality
needs to be studied in the context of use, i.e., natural texts, not
isolated sentences; and also, I believe, that a thorough study of all
grammatical expressions of modality and mood must be done within a single
language before the results are compared and contrasted cross-linguistically.
Perhaps the authors of the papers and grammars that Palmer used had
done just that, but the contexts were lost through the excerpting of
sentences to illustrate his notional categories.
In spite of this criticism,
I found Palmer’s categories, his compilation of data from many different
languages, and explanations of terminological usage very helpful in
my own work, as well as thought provoking. I wholeheartedly recommend
the book for your reference shelf, particularly if you are a linguist
or translator who needs to do an in-depth study of modality in a single
language or a cross-language comparison of modality.
In his preface, the author
explains that this volume is a complement to an earlier
volume, titled An Empirical
Grammar of the English Verb: Modal Verbs, published in 1995 by the same
publisher. The present volume clearly aims to provide an inventory of
the various verb combinations within the English verb phrase. This is
done on the basis of the study of a large amount of corpus data. I am
not so sure whether the term grammar in the title is entirely justified,
for even though information about distribution and frequency of occurrence
of the various patterns is provided, the book hardly ever goes beyond
providing this kind of information. The author writes in his introduction
that «all instances of verbs and verb phrases can be explained as cases
of rule-governed grammatical behavior,» but what these rules are is
not explained anywhere. I will come back to this below.
The book definitely has a
number of strong points, but it also has quite a number of serious shortcomings.
Let me first discuss the general contents of the book. It is divided
into seven chapters: «Introduction» «General Categories» «Verb
Forms».
«Verb Phrases»
«Finite Verb Phrases» «Non-finite Verb Phrases» and «Time
Orientation».
In the introduction, the author
explains the inductive approach he has followed.
Basically, there are three
steps: from language to verb patterns and their contexts, from database
creation (i.e., storing the verb patterns identified) to linguistic
analysis of verb patterns, and from the results of these analyses to
a grammar of the English verb. He also states (6) that the grammar is
based on authentic English and that there has been no borrowing from
previous grammars. This is probably also why there is no reference section.
They are useful for a quick first impression. In the prototypes, the
author claims, the users of the book will find the most frequent patterns
they are likely to encounter «in texts or in contact with speakers
of the language» (12). The details, finally, give information on form
(full vs. contracted forms, which the author persistently calls «elided»
forms), meaning, and context. In the latter, contextual information
is given on affirmative and negative contexts, declarative and interrogative
contexts, combination with subjects, combination with verbs, and other
syntactic information.
Some of this information is
certainly useful, but a lot of it is repetitive and could well have
been stated generally. For instance, in the contexts of nearly all the
patterns, it is said that affirmative contexts are far more frequent
than negative contexts and that declarative contexts are far more frequent
than interrogative contexts, the percentages being roughly 90 for affirmative
and 10 for negative and another 90 for declarative and 10 for interrogative,
give or take a point or two. This information could have been formulated
once, under a general heading, after which contexts with a clearly deviant
distributional pattern could have been appropriately highlighted and
commented on, as in the case of be allowed to, which occurs in a negative
context in 23 percent of all cases (400). Incidentally, the author apparently
only considers the occurrence of the word not (or n’t) to be an indication
of a negative context, for on the same page he quotes the sentence nobody
should be allowed to forget it as an example of an affirmative context
(400). This, and the lack of comment, makes the book really little more
than a mere listing of examples of the various patterns distinguished.
«General Categories,»
briefly discusses the concept of time, temporal orientation, and temporal
reference. Temporal orientation can be past, present, or future, while
time reference can be preceding, simultaneous, following, or neutral.
Thus, the sentences below
(listed on page 19) all have past time orientation but have preceding,
simultaneous, following, and neutral reference, respectively, the reference
indicated by the highlighted verb phrases.
Lee, I noticed, had asked
for a Coca-Cola
But what he saw was an ageing
Australian woman
She was glad that he would
be with her
He won because he’s forty
years younger than you
Only time orientation, however,
is indicated for the various verb phrase patterns distinguished «because
of the intricacies of time reference» (19). It makes sense to make
an inventory of the time orientation of the verb phrase patterns because
after all this orientation is somehow expressed by the tense of the
verb phrase (although this does not apply to nonfinite verb phrases).
It makes equal sense to make an inventory of real and nonreal states
or events referred to by the verb phrase patterns because, again, this
is indicated by the tense or modality expressed by the verb phrase.
I find it less natural to make an inventory of restrictive or nonrestrictive
meaning expressed particularly by nonfinite verb phrases (21), for this
distinction is not inherent in the verb phrase itself. Moreover, it
can only refer to a relatively small subset of nonfinite clauses–namely,
those with an attributive function.
The final category that Mindt
distinguishes as relevant to the description of the verb phrases is
the nature of the subject with which they are associated. Mind distinguishes
between intentional and nonintentional subjects (22), but he does not
really explain the difference at all convincingly. He merely provides
a few examples of each, giving the reader the impression that the distinction
more or less coincides with human and nonhuman subjects, for he then
says, «Because of the relation between verb phrase and intentional
and non-intentional subjects, the distinction between intentional and
non-intentional has to be made no matter whether the subject is acting
intentionally or not» (22). Thus, in most patients are taught to do
this, we have an intentional subject, whereas in more techniques are
taught, we have a nonintentional subject. It would seem to me that Mindt
has thought of a category, then found that it is not useful at all in
many cases but has decided to hang onto it in spite of this.
«Verb Forms» is
a very straightforward chapter spelling out the details of verb forms,
verb morphology, inflection, spelling rules, and patterns of irregular
verbs. This chapter concludes with a learning list of irregular verbs,
based on the rank list compiled, one assumes, on the basis of the corpora
listed in the appendix.
Fortunately, there is also
an alphabetical list of irregular verbs.
«Verb Phrases,»
discusses verb phrase types. Mindt uses a three dimensional graphic
representation of verbal elements that can occur in a verb phrase and
the order in which they occur. This model was introduced in An Empirical
Grammar of the English Verb: Modal Verbs (Mindt 1995). The model enables
Mindt to account for a great variety of verb phrase patterns, in which
all kinds of combinations of modals, auxiliaries, so-called catenative
verbs, and main verbs can be combined in specific ways. The main problem
with this chapter (as with the following two) is the justification of
(or rather the failure to justify) the existence of the category of
catenative verbs. The catenative verbs are said to be a group of «chaining»
verbs (which is exactly what the term catenative means) whose function
is apparently to link elements in a verb phrase together. Catenative
verbs do not share any characteristics with modal verbs and very few
with primary auxiliaries. Examples of catenative verbs are seem or begin.
While there are admittedly very good reasons for wishing to distinguish
a category such as catenative verbs, Mindt fails to present any convincing
arguments for this. Moreover, he includes verbs in this category that
should not be included by any standard, such as want, avoid, mean, enjoy,
or be important. Worst of all, in his illustrations of how so-called
catenative verb phrases are distinguished from «noncatenative» verb
phrases, he seems to ignore elements of well-established modern descriptive
grammars of English.
For instance, in the sentence
we want you to come with us (112), the main clause is said to be we
want you and the subclause to come with us, and you is said to be the
object of the main clause and the «semantic subject» of the subclause.
This appears to take us right back to Zandvoortian times and to ignore
the fact that there are very simple constituency tests that would tell
you that in this case, for instance, it does not make sense to ask who
do we want? but it would make sense to ask what do we want? – thus
ruling out you as an object of the main clause.
In the sentence he wanted
to talk to Armstrong (111), wanted to talk is a catenative verb phrase,
with wanted a catenative verb, but in the sentence quoted above (we
want you to come with us), want and to come are separate verb phrases,
with both want and come as main verbs. Mindt argues (471) that the distinction
of the category of catenative verbs reduces the number of nonfinite
verb phrases, implying that this makes the description of sentences
more straightforward. I am not convinced that that is true. Moreover,
an important generalization is missed – namely, that verbs such as
want are simply complemented by infinitive clauses, with or without
a subject of their own.
Another example of a catenative
verb occurs in the sentence the authorities failed to respond speedily
(111). Again, no argumentation is provided. Mindt could have argued
that failed cannot be assigned main verb status (e.g., because failed
basically means no more than did not) and therefore should be looked
on as a catenative verb, making up a single verb phrase with the following
main verb respond.
In the book, we can only guess
what traditional descriptions and which previous grammars he means.
But what he claims is not quite true, of course, for Quirk et al. (1972)
do distinguish a separate category of semi-auxiliary verbs, including
verbs such as seem and happen (in Quirk et al. 1985, these verbs are
also termed catenative verbs, by the way).
I find the discussion of catenative
verbs particularly problematic because Mindt does not provide any proper
syntactic arguments, a state of affairs that leads him to include an
excessive number of verbs in this category. For instance, the verb want
is included (see the example above) on the strength of the argument
that catenative verbs «allow overlap of two meanings within one verb
phrase. This overlap cannot be achieved by modals alone, because a verb
phrase cannot contain more than one modal verb. Thus, Mindt claims,
possibility/high probability can be expressed by might, as in fever
might kill him. Volition/intention can be achieved by will, as in I
will not be a soldier. If we want to combine these two, Mindt argues,
we cannot simply combine might with will, but instead we can combine
might with the catenative want to express volition/intention, as in
they might want to kill us. The flaw in this argument, I think, is that
want does not simply express volition/intention but desire, which is
not the same thing.
Mindt overlooks the rather
basic fact that propositional content is expressed by the lexical verb
in a clause and that all subordinate verbs in the verb phrase do not
add any propositional content, but only such things as modality, aspect,
and so on.
This can easily be tested
by comparing active and passive counterparts, which should express the
same proposition. For instance, on the basis of the sentence pair Harry
kissed Jane/Jane was kissed by Harry, we can equate the following pairs:
Harry has kissed Jane = Jane
has been kissed by Harry
Harry will kiss Jane = Jane
will be kissed by Harry
Harry may have kissed
Jane = Jane may have been kissed by Harry
Harry appeared to kiss Jane
= Jane appeared to be kissed by Harry but not the following:
Harry wanted to kiss Jane
≠ Jane wanted to be kissed by Harry which shows that want adds
propositional content to these sentences and should therefore be looked
on as a lexical, rather than a catenative, verb.
Mindt also distinguishes a
group of catenative verbs followed by present participles, suchas continue,
start, keep, and so on (321 ff.). Here too, a number of verbs are included
that clearly do not belong there, such as consider, enjoy, avoid, mean.
Again, Mindt does not use
a rather simple constituency test to make the distinction.
It would be simple enough
to compare he kept going for ten hours to he enjoyed possessing his
knowledge on his own by applying a pronominalization test, which would
show that it is impossible to paraphrase the former sentence above by
he kept it but perfectly possible to paraphrase the latter by he enjoyed
it, thus giving separate constituency status to the bit that follows
enjoy but not to the bit following keep.
Finally, there is a category
of catenative adjective constructions, such as be able to and be likely
to (404) (incidentally, these are called semi-auxiliaries in Quirk et
al. 1985). Regrettably, Mindt erroneously includes a number of cases
of extraposed subject clauses here, such as it is necessary to go back
in time and it could be important to record facts, where the supposed
catenative verbs are be necessary to and be important to. This kind
of error should not have occurred in a book like this. on nonfinite
verb phrases, a three-way distinction is made between verbal to-infinitives,
verbal to-infinitives preceded by be, and gerundial to infinitives.
This amounts basically
to the clause functions of adverbials, subject complements, and subjects
of NP modifiers, respectively. However, in the first group, we find
the example it’s impossible to be accurate about these things (472
– highlighting Mindt’s). One wonders why be possible to is listed
earlier, as a catenative adjective construction while be impossible
to is apparently something else. Of course, this is again a case of
an extraposed subject clause and should therefore, if anything, have
been listed as a gerundial to-infinitive.
Mindt discusses the patternVERBPHRASE
+ DIRECT OBJECT + TO-INFINITIVE. This makes one think of Zandvoort’s
(1945) Accusative with Infinitive constructions. Again, Mindt is not
very careful here, for here he lists verbs suchas want, ask, tell, allow,
expect, persuade, cause. These are precisely the verbs that grammarians
and generative linguists alike have used over the years to demonstrate
different types of verb complementation, based on the differences in
syntactic behavior of the complements of these verbs.
The concept of meaning, like
other concepts, is not explained but rather exemplified.
This leads to distinctions
that are fairly arbitrary, such as the distinction of the two meanings
of the catenative constructionHAVE (TO) (298), which is said to express
either necessity or obligation. The following examples are given, without
any further comment:
Necessity: I have to speak
to you about Pepita’s education among other things you’ll have to
stand up for yourself Obligation: the man had to retire at sixty one
of us will have to go in the end This leaves one wondering what the
distinction is based on. Is it based on the possibility of paraphrasing
the former two by it is/was necessary that… and the latter by there
is/was an obligation/order…? I do not know, and frankly, the examples
do not even convincingly point in this direction.
In the discussion of the prototypes
of the progressive (254 ff.), Mindt distinguishes four types, expressing
incompletion, temporariness, iteration/habit, and highlighting/prominence,
respectively. He then goes on to describe each of these prototypes in
more detail. Curiously, hardly any of the prototypes are found to be
pure types: they nearly always combine with elements of other prototypes.
So what is meant by the «prototypes» is probably aspects of meaning.
Incidentally, in the discussion of incompletion, it is said that in
30 percent of the cases, it is combined with temporariness (257)7. In
the discussion of temporariness (258), it is said that in 50 percent
of the cases, it combines with incompletion. It is hard to compare these
figures to each other since no absolute figures are provided. Also,
Mindt does not indicate whether there is possibly a difference between
the combination temporariness + incompletion and incompletion + temporariness.
However, what is clear from these examples is that the really typical
progressive form combines the aspects of incompletion and temporariness.
But this conclusion is not in the book.
All in all, there are too
many of these infelicities in this book. What exactly is a verb phrase
is not clarified. The book would have been so much more valuable if
the classifications had been shown to have been made on the basis of
syntactic arguments.
It would undoubtedly also
have meant that certain erroneous classifications would have been avoided.
As it is, the book can be no more than an inventory of examples of English
verbal patterns, which may be used as a resource for course book designer.
In syntax, a verb is a word
belonging to the part of speech that usually denotes an action (bring,
read), an occurrence (decompose, glitter), or a state of being (exist,
stand). Depending on the language, a verb may vary in form according
to many factors, possibly including its tense, aspect, mood and voice.
It may also agree with the person, gender, and/or number of some of
its arguments (subject, object, etc.).
Информация о работе Грамматические категории глагола в английском и русском языках