|
|
Anaphoric pronouns of metonymic expressions
Beatrice Warren
来源：http://coglingchina.anyp.cn/
Abstract
One peculiarity of referential metonymy is that metonymic subjects and their predicates need not agree as to number. Nor need there be gender agreement between metonymic expressions and anaphoric pronouns. In this respect metonymies differ from metaphors.To account for the apparently erratic syntax in utterances containing referential metonymies, it is first suggested that these involve implicit extensions which together with the explicit metonymic element form referential units. This suggests that metaphorical and metonymic mapping processes are fundamentally different. The focus of the paper is next the question of what determines choice of anaphoric pronouns. It is argued that what is perceived to be topic is an important factor.
1. Introduction
One essential difference between referential metonymies and metaphors concerns their syntactic interaction with surrounding elements in the utterance. Whereas metonymic subjects need not agree as to number with their predicates, metaphors consistently display number agreement. Also in the case of anaphoric pronouns, there are differences. In metaphorical expressions, the pronoun will predictably agree with the figurative meaning (the so-called target), whereas in the case of metonymic constructions, it sometimes agrees with the literal and sometimes with the non-literal interpretation of the expression. Below the question of what determines the choice of anaphoric pronouns in referential metonymy will be in focus. The discussion will begin with a more explicit explanation of the problem (section 2). This will be followed by a review of some accounts of how to regulate choice of anaphoric pronouns in metonymy (sections 3 to 5). The emphasis will be on the particular approach of the present writer, presented mainly in section 5. In the concluding section the different accounts will be summarised.
2. The problem
Consider and compare (1) and (2) below:
(1) The French fries are/*is getting cold.
(2) The French fries is/*are waiting. – metonymic extension
[The one who is having] the French fries is waiting.
In (1) the verb has to be in the plural since the subject is plural. However, the singular verb is possible in (2) provided we allow the noun to be metonymically interpreted which, it is posited, involves some kind of implicit extension of the subject.
It is not difficult to find additional examples which support the view that metonymies involve this kind of implicit extension. The proposition in (3) is not contradictory because of the possible extension of the subject.
(3) In Len's painting, the girl with blue eyes has green eyes.
In Len's painting, [the representation of] the girl with blue eyes has green eyes. (Example from Jackendoff (1985:53))
Consider also (4), which could be an infelicitous utterance by someone observing Ringo Starr looking at his wax doll in Madame Tussaud's museum. A metonymic extension of Ringo reveals the source of the infelicity.
(4) All of a sudden, I accidentally bumped into the statues and Ringo toppled over and fell on *himself.
All of a sudden, I accidentally bumped into the statues and, [the representation of] Ringo toppled over and fell on *himself. (Example from Jackendoff (1992:5))
Consider finally (5), (6) and (7), which contain anaphoric pronouns which are acceptable since metonymic extensions are possible.
(5) The French fries is waiting and (s)he is getting upset.
(6) Cruse, which is a course book, is on the top shelf.
[That which is produced by] Cruse, which is a course book, is on the top shelf.
(7) I need to call the garage. -- They said they’d have my car ready.
I need to call [those who are at] the garage. -- They said they’d have my car ready.
These examples (i.e. (2) through (7)) are in line with Langacker's suggestion that a well-chosen metonymic expression lets us mention one entity and thereby evoke -essentially automatically – another entity that is either of lesser interest or harder to name (1993:30). The above examples also support my suggestion that the traditional view that metonymic expressions involve substitution is misleading: the implicit element does not replace the explicit element; instead the two combine to form a referential unit. Hence my suggestion that referential metonymic constructions have a syntax, in which the implicit element is the head and the explicit element the modifier (Warren (2002)). This agrees with the view that in metaphor the target domain annihilates the source domain, whereas in metonymy mapping involves linking of target with source expressed in particular by Dirven (Dirven 1993). This explains inter alia why selection restrictions are normally superficially violated in metonymies: French fries cannot wait, one cannot talk to garages, etc. At first blush this suggestion appears also to take care of another problem that has been discussed by some linguists. That is, anaphoric pronouns can refer to the implicit element in the construction as shown in (5) through (7), but they can also take as the antecedent the explicit element as in (8). The question is what determines the choice of pronoun.
(8) Ringo was hit in the fender when he was momentarily distracted by a motorcycle. (explicit antecedent)
(Example from Nunberg 1996:114)
One explanation that springs to mind is that either of the two elements: the explicit or the implicit can act as antecedents. Which, is determined by context: French fries cannot be upset but customers can, hence the antecedent is the implicit element in this case and similarly: cars cannot be distracted but Ringo could, hence the antecedent is the explicit element.
(5) The French fries is waiting and she is getting upset
[the one who is having] the French fries is waiting and she is getting upset
(8) Ringo was hit in the fender when he was distracted by the noise.
[that which contained] Ringo was hit in the fender when he was momentarily distracted by a motorcycle.
But this does not hold. Consider:
(9) Ringo was hit in the fender when *it turned left.
The implicit element is not available as antecedent in (9) although context clearly invites such an interpretation.
As already indicated, the problem of anaphora in the case of referential metonymic expressions has been discussed in the literature, more precisely by Stallard (1993), Nunberg (1996) and Ruiz de Mendoza (2000) and in passing by myself (Warren 2002). Perhaps one of these approaches can account for choice of anaphoric pronouns in the case of metonymies? Let us therefore consider these suggestions.
3. Ruiz de Mendoza's explanation
Ruiz de Mendoza claims that there is a matrix-to-subdomain relationship between the explicit and implicit elements in metonymic expressions and that only the matrix domain will be available for "antecedentship". So, if the implicit element is a subdomain of the explicit element, the explicit element will be the antecedent, but if it is a matrix domain of the explicit element, it will be the antecedent.
(10) Nixon bombed Hanoi. He killed many. ("bombers" subdomain of Nixon)
(5) The French fries is waiting. She is upset. ("customer" matrix domain of fries)
This suggestion rests on the possibility of determining what domain includes what other domain. There seem to be no other criteria but intuition to do so. Intuition is an important tool in linguistics but only provided it has intersubjective support. It is debatable whether there is such support in the case of domain boundaries[1]. Consider a few examples. In (11) I would be prepared to accept that the water is the subdomain of the kettle since the kettle is the container, but I have problems with (12). In what sense would water be the subdomain of potatoes? I have no intuitions of this kind.
(11) The kettle is boiling and it is hot. ("water" is subdomain of kettle)
(12) The potatoes are boiling and they will be ready soon. (?"water" is subdomain of potatoes)
Similarly in (13), which is an utterance I have heard myself make when I have talked to my relatives in Finland over the telephone. Is Finland really the subdomain of my relatives?
(13) Finland phoned. They sent their love. (?Finland subdomain of my relatives)
Further, I do not know how Ruiz de Mendoza would handle the fact that in (14a) car would be subdomain of Ringo, but in (14b) it would be the matrix domain of Ringo.
(14a) Ringo was hit in the fender when he [Ringo] was distracted. (car is subdomain of Ringo)
(14b) Ringo was hit in the fender when he [his car] was parked at the university. (car is matrix domain of Ringo)
4. Nunberg's and Stallard's explanations
Nunberg (1996) suggests that we must distinguish between deferred reference (reference transfer) and predicate transfer. Deferred reference is exemplified by (15) uttered by the owner of a car when handing a car park attendant a car key. In this case nominal transfer is possible. Predicate transfer is exemplified in (16). The character of the predicate transfer is approximately as indicated within square brackets.
(15) This is parked out back.
[that which belongs to] this is parked out back. (deferred reference)
(16) I am parked out back.
I [have the property of having a car that is] parked out back. (predicate transfer)
When the noun is extended, the antecedent of an anaphoric element will be the implicit element:
(17) This is parked out back and it won't start. (extended noun: implicit antecedent)
When the predicate is transferred, the antecedent of an anaphoric element will needs be the explicit element:
(18) The man with the cigar is parked out back and he might be an hour. (extended predicate: explicit antecedent)
The implicit antecedent is not possible in this case.
(19) The man with the cigar is parked out back and *it might not start.
Exactly how and why the mental processing involved in predicate transfer is accomplished is not made clear by Nunberg. However, it must be taken to involve the following: The conventional meaning of the predicate is first accessed since it is the incompatibility of the predicate with some nominal element of the utterance that triggers the extension of this noun, an extension which then is somehow incorporated as part of the predicate. This seems somewhat implausible. In this respect Stallard's suggestion is easier to accept, which, if I understand him correctly, is that in some instances of metonymy the argument structure of the verb is shifted so that it accommodates the explicit element as an argument[2].
Another weakness of Nunberg's explanation is that it is not consistent with the fact that there are numerous examples of nouns with conventionalised metonymic senses (tea in the sense of a meal, silver meaning cutlery, ecstasy referring to a drug, box referring to a container, date in the sense of appointment or person encountered at an appointment, etc., etc,), but few, if any examples of verbs having conventionalised metonymic senses of the type illustrated above.
What is interesting about Nunberg's and Stallard's accounts, however, is their intuition that in some metonymies the predicate is about the explicit rather than the implicit element and that choice of anaphoric pronoun is connected to this. Consider (10) again and (20), which may demonstrate this more clearly than Nunberg's examples cited above.
(10) Nixon bombed Hanoi.[3] (He killed many.)
(20) This pot has boiled dry. (It is destroyed.)
It seems clear that (10) and (20) are assertions about Nixon and a particular pot respectively rather than about pilots or some liquid in the pot.
5. Warren's explanation
My explanation is inspired by Construction Grammar and is in line with Nunberg's and Stallard's intuition concerning the role of the predicate vis-à-vis the explicit noun. Construction Grammar posits that abstract syntactic patterns can be associated with meanings of a more or less general kind. A pattern such as Subject-Predicate would express the very abstract notion "Predicate is true of/applies to Subject". In other words, the predicate is the comment about the subject, which is our topic. Compare (21a) and (21b).
(21a) The laces of the boots were neatly tied and they [the laces] were clean.
(21b) The boots [its laces] were neatly tied and they [the boots] were clean.
These examples seem to indicate that that which we perceive to be the topic of an utterance will be the antecedent of an anaphoric pronoun and that the explicit member of a metonymic expression, in spite of its modifying status, can be taken to be the topic, probably because it is in topic position. However, a metonym can assert its topic status only if the proposition expressed can be thought of as being an appropriate comment about the topic. Let us try this out:
(22) Nixon bombed Hanoi. (He killed many.)
What can I tell you about Nixon?- Well, he had people bomb Hanoi.
(23) The French fries is waiting. (She is upset.)
What can I tell you about the French fries?- ?? Well, the customer who ordered them is waiting.
(24) The kettle is boiling (and it is hot.)
What can I tell you about the kettle? - Well, the water in it is boiling just now.
(25) The potatoes are boiling (and they will be ready soon.)
What can I tell you about the potatoes? - Well, the water they are in is boiling.
(26) Finland phoned. (They sent their love.)
What can I tell you about Finland? -?? Well, my relatives who live there phoned.
As we can see, this is in line with Nunberg's and Stallard's intuition that something is asserted about the explicit part of the subject in some metonymies but not in all. The difference between Nunberg's and my explanation is that the implicit element is always part of the nominal subject in my account, whereas it is sometimes part of the predicate in Nunberg's account. One consequence of this difference is that Nunberg would maintain that in (21b) boots and the anaphoric pronoun are co-referential, whereas I would maintain that they are not, but so to speak "co-topical".
(21b) The boots [its laces] were neatly tied and they [the boots] were clean.
Similarly, in (27) Nunberg would maintain that Cædmon refers to the poet when acting as the subject of was the first Anglo-Saxon poet and of fills only a couple of pages, whereas I maintain that it changes reference. It refers to the poet in the former case and to Cædmon's poetry in the latter.
(27) Cædmon, who was the first Anglo-Saxon poet, fills only a couple of pages in this book of poetry.
(The example is from Nunberg 1979:167, ex. 29 p 196)
In other words, my account stipulates that change of reference is permissible but change of topic is not. This would allow (14a), in which the topic is kept but referents differ. It would also allow (14b) in which the topic and the referents are kept, but it would not allow (14c) in which the referents were kept but the topic changed.
(14a) Ringo [his car] was hit in the fender when he [Ringo] was distracted.
(14b) Ringo [his car] was hit in the fender when he [his car] was parked at the university.
(14c) Ringo [his car] was hit in the fender when *it [his car] was parked at the university.
This approach also predicts the use of himself in (28).
(28) Norman Mailer [the writer] likes to read himself/*itself [his writing] before going to sleep. (Example from Fass 1997:388)
Possibly the difference between Nunberg and my account emanates from the fact that I see the implicit element as a complement of the explicit noun and not as a replacement, whereas Nunberg seems to adhere to the traditional view that metonymies involve substitution, i.e. the implicit content replaces the explicit item.
My hypothesis is that basically metonymy is a focussing construction. That is, referential metonymic constructions occur because the speaker is focussing on an attribute of some entity rather than on the entity itself although both are mentally present and conveyed. In linguistic terms, the attribute takes the form of an explicit non-referring item (a modifier) whereas the entity becomes the implicit head (referring item) of the construction. However, referential metonymies are not merely focussing constructions, they are simultaneously frequently topicalization manoeuvers in that non-referring items (i.e. modifiers) can anomalously be made topics. In my view it is this “linguistic twist” that makes metonymic constructions interesting and more than simply abbreviated noun phrases[4]. As an illustration, consider again abbreviated versions of (24 a and b):
(21a) The laces of the boots were neatly tied.
(21b) The boots were neatly tied.
Provided (21b) is metonymically interpreted, it expresses the same proposition as (21a), but its focus is different. In (21a) the focus is on the laces, whereas in (21b) it is on the boots, bringing about the implication supported by its topic status, that, because the laces were neatly tied, the boots as a whole were neat. In other words, the fact that the laces were neatly tied becomes an assertion applicable also to the boots.
6. Summing up
Four accounts of what determines choice of anaphoric pronouns of metonymic expressions have been considered. One of these (Ruiz de Mendoza (2000)) connects the ability of becoming antecedent with domain status. Only elements representing matrix domains may become antecedents. The remaining three accounts (Stallard (1993), Nunberg (1996) and Warren (2002 and this article)) have–at least in my interpretation–in common that they see a connection between choice of pronoun and which entity the predicate is an assertion of. If the predicate can be taken to be about the explicit element in spite of the fact that they are truth conditionally incompatible, this becomes the preferred antecedent, otherwise the implicit element can act as antecedent.
This approach has to explain how the predicate can be literally true of the implicit element and at the same time be perceived to assert something about the explicit noun with which it is often incompatible (pots cannot boil, for instance). Nunberg solves this by claiming that in these cases the predicate does not express its conventional meaning. Stallard suggests that the argument structure of the verb is changed and Warren (i.e. the present author) maintains that the explicit element may be perceived to be the topic of the utterance in spite of its non-referring status provided the proposition expressed can be taken to apply to it.
References
Barcelona, Antonio. 2000. “Introduction. The cognitive theory of metaphor and metonymy” in Barcelona, Antonio (ed.) Metaphor and metonymy at the crossroads. Berlin and New York: Mouton de Gruyter, pp 1-15.
Dirven, René. 1993. "Metonymy and metaphor Different mental strategies of conceptualisation", Leuvense Bijdragen 82:1-28.
Fass, Dan. 1997. Processing metonymy and metaphor. Greenwich, Connecticut and London, England: Ablex Publishing Corporation.
Feyaerts, Kurt. 2000. “Refining the inheritance hypothesis: Interaction between metaphoric and metonymic hierarchies” in Barcelona, Antonio (ed.) Metaphor and metonymy at the crossroads. Berlin and New York: Mouton de Gruyter, pp 59-78.
Jackendoff, Ray. 1975. "On belief contexts", Linguistic Inquiry 1.6:53-93.
Jackendoff, Ray. 1992. "Mme Tussaud meets the binding theory", Natural Language and Linguistic Theory 8, 1-31.
Lakoff, George and Mark Johnson. 1980. Metaphors we live by. Chicago and London: The University Press.
Langacker, Ronald. 1993. "Reference-point constructions", Cognitive Linguistics 4:1-38.
Nunberg, Geoffrey. 1979. "The nonuniqueness of semantic solutions: polysemy", Linguistics and Philosophy 3: 143-194.
Nunberg, Geoffrey. 1996. “Transfer of meanings” in Lexical Semantics, Pustejovsky, James and Boguraev, Branimir (eds.) Oxford: Clarendon Press. (pp 109-132) and in Journal of Semantics 12:109-132.
Ruiz de Mendoza Ibánez, Francisco J. 2000. “The role of mappings and domains in understanding metonymy” in Barcelona, Antonio (ed.) Metaphor and metonymy at the crossroads. Berlin and New York: Mouton de Gruyter, pp 109-132.
Stallard, David. 1993. "Two kinds of metonymy" in Proceedings of the 31st Annual Meeting of the Association for Computational Linguistics (ACL-93), Ohio State University, Columbus, OH, pp 87-94.
Warren, Beatrice. 2002. “An alternative account of the interpretation of referential metonymy and metaphor” in Dirven René and Pörings, Ralf. Metaphor and metonymy in comparison and contrast. Berlin - New York: Mouton de Gruyter, pp 113-130.
--------------------------------------------------------------------------------
[1] I am not the only one to question the methodological soundness of basing explanations on the assumption that we can determine domain borders. Cf. Barcelona (2000:8-9), who points out that if we accept Langacker's characterisation of domains as that which includes the entrenched knowledge a speaker has about an area of experience, then "(t)his will vary in breadth from speaker to speaker and in many cases has no precise boundaries." And consider Feyaerts (2000: 63): "Methodologically speaking, it appears that the notion of domain is too malleable to serve as an adequate criterion in the discussion about the distinction between metaphor and metonymy."
[2] For a more in depth review and comparison of Stallard's and Nunberg's approaches, the reader is referred to Fass (1997: 83-91).
[3] This example appeared first in Lakoff and Johnson (1980:38). It was accepted by Stallard (1993:87) and Ruiz de Mendoza (2000:117) as exemplifying metonymy. However, arguably there is an alternative explanation for the interpretation of (10), which is that the verb has a causative reading: "Nixon caused the bombing of Hanoi". Cf .the standard example of a coerced causative reading of a normally intransitive verb, i.e. The sergeant marched the soldiers: "the sergeant caused the marching of the soldiers".
[4] As an illustration of this observation , i.e. that metonymic expressions are not simply abbreviated noun phrases, consider the following: The assumption that in referential metonymy there are implicit extensions of nouns which are mentally present for speaker as well as interpreter is uncontroversial. Yet, interestingly, these extensions seem backgrounded to the extent that when they are made explicit and inserted as in some of the examples above, they seem to interfere with rather than clarify interpretations.
( Taken from metaphorik.de ) |
|
IP卡
狗仔卡
发表于 2005-1-28 21:31:18
提升卡
置顶卡
沉默卡
喧嚣卡
变色卡
显身卡