Eve Sweetser, From etymology to pragmatics: Metaphorical and cultural aspects of semantic structure. Cambridge studies in linguistics, Cambridge University Press, 1990. 174 pp. $44.50. ISBN 0-521-32406-8.
Reviewed by Per Aage Brandt, Center for Semiotic Studies, Aarhus University, Fin- landsgade 26B, DK-8200 Aarhus N, Denmark.
An utterance is content, epistemic object, and speech act all at once.
(Eve Sweetser, p. 75)
In our understanding of language in general, there seems to be a schema for lexicalization the sense of which is that the act of lexicalizing something is the act of presenting it as an established category of human thought. If a lexical item exists, in other words, it must exist as some part of a frame and must correspond to some part of a schema.
(Charles Fillmore, quoted exergually by Eve Sweetser)
1. Domains and meanings
Eve Sweetser's From etymology to pragmatics is a carefully written, well-cen- tered, and clearly argumented contribution to cognitive linguistics. It has one major theoretical proposal to make, and does so by mirroring this in a number of empirical issues. The central theoretical idea is that natural language semantics is based on a general pattern which orients lexical semantic change through history (etymology), lexical polysemy, and the pragmatic ambiguity of utterances: namely, a metaphori- cal transfer of form (meaning, in my phrasing) between differently functional seman- tic domains, from the sociophysical real-world domain (D1), to the mental, epis- temic-objective and emotional-subjective domain (D2), to the speech act domain (D3) - that is, from one 'concrete' domain to two 'abstract' domains. The functional transfer of form thus follows a path on a general semantic map from the real-world D1 to the mental D2 to the speech act D3 (let us say: D1 ---) D2 ---) D3), and this mechanism, which is operative both diachronically and synchronically, is a general
282 Book reviews / Journal of Pragmatics 25 (1996) 281-302
property of language as based on cognition, and not a property of words or sentences taken separately. The possible meanings of words and sentences depend on such a general, underlying, natural transfer structure. This idea discards feature-analysis as an overall strategy for semantic studies, and also opposes generative semantics. Both assume that meaning can be accounted for as an inherent property of the analyzed items, and in both cases, the structure is expected to be located in the analyzed items, and not in the general structure they emerge from. The author's transfer analysis rep- resents a stronger ontology and a holistic viewpoint: "linguistic structure is, then, as logical and objective as human cognition, no more and no less" (p. 17).
The four central chapters of the book (the author's first) deal with perception verbs, modality, conjunctions, and conditionals. The introduction portrays the state of the art, and particularly stresses the importance of metaphor as a general cognitive processor. For instance, Sweetser says, "physical-domain verbs frequently come to have speech act and/or mental-state meanings, and mental-state verbs come to have speech act meanings, while the opposite directions of change do not occur" (p. 19). This phenomenon might express the inherent unidirectionality of a metaphorical connection. Also, both speech acts and mental states are metaphorically treated as types of travel through physical space; but speech acts are also metaphorically treated as exchanges of objects between the interlocutors (the objects thus becoming containers for meaning), or are metaphorically understood as forms of combat. By contrast, mental activity is seen as the bodily manipulation of objects rather than as a form of exchange or combat. Some metaphors are shared, and others are particular to the target area. The abstract areas are structured by metaphor systems whose com- ponents seem to follow the outlined universal tendency.
I would like to add that this view (which is close to Lakoff's, who, however, to my knowledge, has never proposed to canonize the underlying triad of domains) also leads to a modification of the ballistic metaphor of metaphor itself. The receiver-D is not simply a 'target'; instead, because it is 'structured by' a system of metaphors, it actively attracts certain forms by virtue of its own structure. The 'target' is an attractor, and the transferred forms are not only pushed, but also pulled, by inherent formal properties of the receiving D. If, by metaphor, 'A is B', then an A-form attracts a B-form, and I would theoretically single out this dynamic relationship as schematic. This interformal dynamism might explain why poetic metaphors easily invert the orientations of the proposed triad: if a text works out a special view of an A-form, then different schemes in it are activated, and different transfers occur.
2. Sense-perception verbs
The essential point of the chapter on sense-perception verbs is that vision, hearing, and smelling~tasting~feeling are sources that map regularly, but differently, into the abstract Ds. Vision verbs commonly develop abstract senses of mental activity, e.g., senses of knowledge, intellection, mental 'vision', or imagination (D1 ~ D2). Hear- ing (cf. Danish lystre, 'obey') seems regularly connected to intention, attention, respect, and obligation, i.e., to intersubjective (speech-act-forceful) meanings (D1 --~ D3). However, some verbs invert the direction, e.g., French entendre, English rec-
Book reviews / Journal of Pragmatics 25 (1996) 281-302 283
ognize. The proximal perception verbs - smell, taste, touch - come to mean emo- tional events and evaluative preferences. The author's domain classification here is not quite clear (D2 or D3?), but I think it would be reasonable to treat these 'aes- thetic' meanings as naturally conflating I epistemic and speech act properties (D1 --~ D2/D3). Let me discretely add that the empirical data of this chapter (pp. 32-33) are surprisingly scanty for a book with etymology in its title.
3. Modality
The chapter on modality is highly interesting. Modal verbs are seen as rooted in D 1, so root modal meaning here refers to concrete settings of force dynamics. The author discusses this idea of Talmy's (1981, 1988), and proposes a slightly different version of the basic structures, 2 in which these are not only physical, but also socio- physical. This allows for an analysis of both causal and intentional doings in terms of forces and barriers. She even writes: "I prefer to view modality as basically refer- ring to intentional, directed forces and barriers" (p. 52). This issue seems to me to be of extreme importance for the understanding of cognition. If intentionality is basic, then Gestalt perception and naive physics cannot be the general and sufficient cog- nitive source of meaning, as the concept of embodiment might suggest. Rather, root modality and perception of time might be closely related, and grounded in elemen- tary social experience - in something like intentional face-to-face perception (wait- ing, expecting, etc.).
May, as a root modal, refers to a potential but absent barrier (as in Talmy's analy- sis of may and letting in causation). The potentially barred entity is then an object, a flow, or a human doing, and the barrier is rendered absent by physical causation or human permission. 3
"I feel that Peter is right" is almost epistemic (an indication of an intuitive inference) and neverthe- less somewhat forceful in communication, as an invitation for the interlocutor to share the speaker's per- sonal intuition or to take the conclusion as based on admiuedly incomplete premises. 2 Eve Sweetser's earlier work (1982) on forces and barriers coincided with the French semiotician A.-J. Greimas' work on modal meaning in the 1980's, and with the dynamic analyses unfolded within the catastrophe-theoretical paradigm proposed then by followers of Ren6 Thom, such as Petitot (1985, 1989a,b) and myself (Brandt, 1992, 1994, 1995). Sweetser's and Talmy's analyses have, in fact, inspired a whole generation of cognitively oriented semioticians, to whom dynamism and schematism seem more important than metaphor in itself, because modal topologies and schemes can explain essential formal standard properties of 'deep semiotics', and can explain how actants and actantial functions structure temporal meaning, as studied in the narratological tradition. I think that these semio-cognitive develop- ments might fit into the contemporary discussion about basics and grounding in semantics. 3 While letting assumes that the formerly barred process now happens, the. permissive may does not stress the pressure of the barred process; it just opens a passage for it. In French, pouvoir is the verb used for permission ('Tu peux venir'). So, instead of a triad can-may-must, French has a dyad pou- voir~devoir. The 'can' and the 'may' meaning share topology in French, and probably in a more general sense. There is an actant and two 'possible' places for it in the scenario. My hypothesis is that when the barrier between them is low, the actant 'can' pass from one intitial place to the other. When the barrier is high, it 'cannot'; the raising of the barrier is the content of the negation. If the barrier is lowered again, it 'may' pass - and French has 'can' once more. The prerequisite for this reading ('can'=low barrier) is that the actant is equipped with some proper energy that is sufficient to overcome the low barrier.
284 Book reviews / Journal of Pragmatics 25 (1996) 281-302
Must, as a root modal, if it had to follow may's pattern, would take a barrier restricting "one ' s domain of action" (p. 52) to a single act. But then we would need a special device for narrowing the scope of 'action' (like pouring something through a funnel). And negation, as Sweetser points out, works differently on may and must:
may not leaves the barrier in place, hence producing a prohibition, whereas must not
rather indicates the raising of a barrier, yielding a stronger prohibition, and compared to positive must, a prohibition instead of an urge. I f intentional grounding is accepted, the basic meaning of must can be described, according to Sweetser, as "an order to do something, a positive compulsion rather than a negative restriction" (p. 52). But are we then not in D3 instead of D1 ? No, because the meaning is still descrip- tive (as in, ' I must stay home tonight'; ' these people must do what they do ...').
Can, as a root modal, indicates a third dynamic factor, which is neither a barrier nor a force, but rather a "positive ability on the part of the doer", a human potential analogous to potential energy in physics. Sweetser writes: "Let us view can as being the equivalent of a full gas tank in a car, and may as the equivalent of an open garage door" (p. 53).
Ta lmy ' s (1981, 1988) analysis of force dynamics has the advantage of integrating three modal meaning types or values into one image-schematic whole. Overlapping meanings are then easy to explain. But the barrier seems hastily generalized. Sweet- ser 's analysis, on the other hand, has the alternative advantage of focusing on intu- itively satisfying definitions and of admitting intentionality, but it takes us back to a less natural coexistence of three seemingly independent schemes: barrier, force, and ability. I think that both are valuable analyses, and that both semanticists are essen- tially right about the basic dynamics of modality. All we have to do in order to obtain a sensitive and natural analysis is to combine the Sweetser-schemes into a Talmy-coherent model, by using a dynamic topology, such as the following.
We could imagine that there is a doer (agonist, actant), or a sort of mobile con- tainer, which is controlled in three ways: firstly, by stationary barriers (repellors) in its surroundings; secondly, by outer forces (attractors) 4 locally centered in the same surroundings (typically, behind the barriers, but sometimes on the same side of them as the doer); and thirdly, by instrumental inner forces (muscular strength, intellectual knowledge, moral faith, etc.) constituting the doer 's ability. The doer is then an inner dynamic system controlled by an outer dynamic system. None of the components can be defined without reference to the others. A barrier presupposes a process of doing (hence, a force). The variable quantum of the barrier is only critical if com- pared to the forces - inner and outer - by which the doer would overcome it. Instead of requiring the plain absence or presence of a barrier, this dynamic whole only requires a critical change in its quantum. Can, may, and must meanings are naturally used together in descriptions of one and the same situation, which is what the
4 The meaning of such a force is an attractive doing. Here, an observation: to think of a doing is to think of it as a doer would think of it, i.e., as an attractive doing. The meaning of any doing, however horrible it may be, is an inseparable alliance between its content and a prototypical doer's wanting to do it. This may be, for example, why some Germans seem to have problems with simple, plain, narrative thinking about what the Nazis did.
Book reviews / Journal of Pragmatics 25 (1996) 281-302 285
dynamic setting refers to. Further developments of this idea can be found in Brandt (1992, 1994, 1995).
Sweetser's convincing main point here is that root modals are transferred from D 1 to D2 - most clearly in past contexts - yielding epistemic meanings: possibility cor- responds to may, and necessity corresponds to must. This is due to shared topologi- cal structure (cf. p. 60; what the structure is, is therefore a crucial issue). There is an inferential 'doer', namely, thought itself, as a sort of body in the mind; and the process of reasoning is the 'doing'. There is, in fact, a dynamic 'situation', namely, the probldmatique, as the French would say. Counterarguments are barriers. Now, if the 'concluding' is the attractive doing, what are the arguments, or premises? 5 Are they simply 'permissions' or 'compulsions' to conclude? I think Sweetser thinks so. But good arguments must weaken the counterarguments in order to be 'good'. So they lower some barriers. But it would be counterintuitive to say that they dissolve these counterarguments completely; they just weaken them, making it possible for the cogito-doer to overcome them. Otherwise, there would be no discussions in the epistemic world. Counterarguments are respected precisely when they are overcome. This phenomenon shows, at least to this reviewer, that good pro-arguments are instrumental forces analogous to the root can meanings. Instead of muscular strength, we get rhetorics; instead of skill, we get propositionally formed knowledge. Pro-arguments are analogous to abilities. But then, according to the mapping, their modal presence should be rendered by can. In fact, it is rendered by the distal form could, not by the positive can. Tile author is aware of this problem. In a note, she writes: "I would love to be able to explain why some of the root modals map better into the epistemic domain than others". - shall? positive need and can? Consider: * 'My conclusion can be true'; *'I need to conclude ...'; *'I shall think that ...'; *'Shall we think like this?' In some cases, negation and interrogation help. The explanation following from our reformulated dynamic analysis might simply be that the doer is still a human subject, epistemic or not, and that the lifting of a root sub- ject to the epistemic status is a precarious affair. Instead of being filled with moral and bodily motives, the subject's inner dynamics must be filled only with premises. Modal reference to the subject is naturally different from modal reference to other parts of the scenario (the must and may parts). Negative and interrogative cans refer, in fact, to possible 'stoppers' outside the premise-stuffed epistemic subject.
In the third dimension (D3), modals have speech act meanings. In this case, the three basic verbs above are all active, and the standard test, I propose to say, is that we can add '... because I say so' to such modal utterances if they are positive. ( 'The show must go on', says the show director to the actors after a breakdown.) Here, the doer is typically the interlocutor, and the modalized nominal is directly related to the doer. The speaker, on the other hand, appears as a lowered barrier, or rather, as the one who controls and lowers the barrier, as in you may, where a per-
"... in the world of reasoning, we wish to have our conclusions forced or restricted by premises ...", so, in a sense, in D2 we do not want to be ' free ' , as we certainly do in D1. I would say that 'free thought ' nevertheless is a meaningful notion, and that it refers to our access to good arguments by which bad counterarguments can be overcome.
286 Book reviews /Journal of Pragmatics 25 (1996) 281-302
mission is created. The speaker may also appear as a force, as in you must, where an obligation is created. And finally, the speaker may appear as an 'abilizer', as in you can (suggesting something like, 'I hereby give you the strength or the skill to do what you want - just follow my instructions or believe in me' - hypnotizing is a good case of this). In this latter case, an ability is created. The joint presence of speaker and interlocutor in characteristic dynamic positions within a dynamic whole argues strongly, I think, for such a Talmy-Sweetser-Brandt version of modal dynamics. The way in which speech act modality behaves (can is active) seems further to indicate that bifurcation (D1 ---) D2/D3) is the best description of the transfer path.
4. Conjunctions
Causal conjunctions, both positive (because) and concessive (although), apply in the three areas. They refer to real-world motivations and causes (D1), and to argu- mentative doings (D2), and they create speech acts (D3). Sometimes it is difficult to sort out the reference types, but this is due to the general fact that utterances are semantically symphonic according to the D-transfer system.
I shall not go through the causational series presented on pp. 76-86 by the author as evidence for the regular ambiguity existing also in this field. The analysis of therefore, since, so, and despite, and the note on while, support the D-transfer hypothesis and work convincingly.
What would be interesting here is the question of shared topology. Modality, as we have seen, can be traced back to a dynamic topology explaining the transfers by shared schematism grounding the 'metaphors'. I now wonder if the notion of metaphor is at all a good means of understanding the transfers we are discussing. Good examples of metaphor are examples of open class transfers ( 'This man is a monster'); but here we are concerned with closed class structures, and there are no figurative and inferential effects perceived when reading according to the triple semantic system. I feel that (sic !) treating it as a matter of metaphor hides the foun- dational question for us. There are intelligible reasons why sense perception verbs map regularly into D2 and D3 in the way they do. Seeing yields epistemic meanings because it is thing-oriented (as thinking also is), and proximal feeling yields emo- tional meanings because it is flow-oriented (as emotions also are). Heating yields meanings related to interpersonal communication because speech, as a D 1-phenom- enon, is prototypically perceived by hearing.
Is there a causational schematism behind because, although, etc.? The answer unavoidably seems to be, as Talmy suggested, that there is a dynamic scenario strongly related to the modal one. Causes are forces. But in the pure causational set- tings, only plain barriers and forces - i.e., repellors and attractors, not energetic 'abilities' - seem to dominate the topology. Causational expressions prefer conclu- sive outcomes. Therefore, can invests only the effect, not the cause. ('I'11 do it, because I can', for example, remains strange, and if said, it suggests a hidden cause behind the ability - viz. 'I must do everything I can do'.) On the other hand, causa- tional settings may contain multiple conflicting forces. The resulting outcome is then
Book reviews / Journal of Pragmatics 25 (1996) 281-302 287
determined by their relative impact on the de-intentionalized doer. Here, we get 'causality', in the philosophical sense, as opposed to (intentional) finality.
And is based on a simple D1 meaning: "setting things side-by-side additively" (p. 90). There is an iconic unfolding of this schematism still in D1, yielding tempo- ral sequencing and cause-effect relations; and there are, by transfer, epistemic and speech act uses. Truth values emerge only in D2, because what are set side-by-side here are syllogistic parts.
Or is not basically a logical operator, but rather, in D1, a set of options and "an expectat ion that only one of the expressed options will in fact be the correct one" (p. 93). In D2, or becomes a natural equivalent of the formal logical operator, but it is still intentionally linked to the dynamics of selecting syllogistic parts: 'right' and 'wrong'. In D3, the threat by or is a clear case ( 'Do it, or I'll kill you').
But presents a side-by-side setting "contrary to our expectations" (p. 100). In D2, there is thus what Sweetser calls a clash of epistemic parts, and in D3, a clash of rel- evances (as in 'George likes mu shu pork, but so do all linguists', where at first, the fact is interesting enough to be said, but then later it is not). The author wonders why but seems absent from D1. She argues: "but what does it mean to say that A and B 'clash' or 'contrast' in the real world? How can discordance or contrast exist outside of the speaker's mental concept of harmony or non-contrast? In a sense, if two states coexist in the real world (and conjunction with but does present both conjuncts as true), then they cannot be said to clash at a real-world level". Here, it occurs to me that the author forgets that the real world of this theory includes intention, and there- fore also expectation (cf. the analysis of the semantics of or). Suddenly, the real world becomes so real that only causality occurs in it. Expectations can certainly be facts about D 1 - otherwise, will could not have become a future mark.
Let me add that but, in fact, has D1 meanings. This is the case, for example, with trying: 'He tried hard, but had to give up'. This use is not epistemic, since we do not expect all triers to succeed. There is, instead, a modal conflict between ability and barrier, and a temporal sequence from can to cannot. 6 As can implies an unsettled outcome (that is, a realistic alternative, or a bifurcation in the real world of real pos- sibilities, as C.S. Peirce would say), the cannot segment of the sequence simply implies that the alternative degenerates, and that - in the case of trying - the implic- itly focused branch of the bifurcation closes: its content becomes counterfactual, a defunct possibility, a dead end. For strong philosophical reasons, by the way, possi- bility must spring from the real world, and may not be thought of as only epistemic. If the latter were the case, intentions would once more have to leave the real world, and we would be left with an incoherent, positivistic list of 'states of affairs'. The present and future would be expulsed from the real world, and the remaining dis- joined skeleton could not be the source of any cognitive forms at all.
Hence, instead of 'contrary to our expectations', I think we should write, 'contrary to our hopes or fears', or schematically: a temporalized setting, B after A, with A leading to a bifurcation BC, and with C on a branch that first is open, then closed
6 Whi l e t ry ing, one says : ' I can ... '; w h e n g iv ing up, one says: ' I cannot ...'.
288 Book reviews / Journal of Pragmatics 25 (1996) 281-302
(causing disappointment or relief) (see Fig. 1). The implication would be that origi- nally, B and C were both possible; but now B happens instead of C: that is, A, but B (not C).
A
fluctuat
mergitur
Fig. 1. Fluctuat nec mergitur.
5. Conditionals
The natural language uses of conditionals are not identical with logical if-then conditionality. For example, the sentence ' I f Paris is the capital of France, (then) two is an even number ' is logically fine, but linguistically not so fine. The author thinks - and justly so, I believe - that sufficient condition is the basic definition in a D1- sense: the realization of the content of the protasis leads to the realization of the con- tent of the apodosis. (But Paris does not lead to equal numbers.) Also, D2-readings follow this principle: "Natural-language epistemic conditionals must have some pre- sumed relationship between the two clauses, just as any conjoined clauses in natural language are assumed to be related" (p. 117). Speech act readings (D3) such as 'promise ' and ' threat ' refer to a sort of ethical link between protasis and apodosis: the former conditions a positive giving or a negative taking away affecting the inter- locutor. The theory works.
So-called 'given conditionals', 7 where the protasis no longer presents a hypothet- ical circumstance but rather a 'g iven ' fact, express admission in the protasis, and conclusion or insistence in the apodosis. They are D2 and/or D3, but not D1. The if formula, Sweetser explains (p. 131), only displays the reasoning sequence, or builds up the pragmatic justification. Despite the givenness of the protasis, it is taken as a hypothetically hypothetical part of the mental process. It is a sort of have-been con- ditional. There is intuitive evidence for this, but still the issue seeems unclear. The problem is, I think, that the implied schematism has not been worked out. The elab- oration of this schematism should include at least a remark on the structurally grounded semantic overlappings of but and if, and a remark on the highly interesting translinguistic relationship between the conditional if and the interrogative if.
Let us consider the bifurcation ABC proposed above. It yields a basic A, but B (instead of C), or A, but B (since C was no longer possible). It also yields a basic tf A, (then) B or C (both possible). The if simply means that there is a bifurcation ABC
7 These would include, we could add, the French 'si Pierre a fait cela, c'est que ...' ('if he did that, it's because ...'), indicating both epistemic understanding and speech act 'excuse'.
Book reviews / Journal of Pragmatics 25 (1996) 281-302 289
when B or C are considered (e.g., feared, hoped for). 8 A is presented as linked by the bifurcational scheme to a not yet realized or produced B or C. These states are futuric, and A appears as their past. The present is viewed as the past of the B/C future, and is thereby rendered as undecided as the alternative B/C. Therefore, A, but B presupposes, in fact, if A, (then) B or C, and then supplies a confirmation of A and B from the perspective o f the lost and absent C. This is why the but-expression is emotional, whereas the basic/f-expressions are not (although, again, the ' g iven ' con- ditionals are). Emotions are realistic.
These D1 meanings work schematically forward: the ~f is on A. But since the morpheme if refers to the total scheme, and thereby displays the mental process of the thinker or speaker in D2 and D3, as Sweetser says, it can also occur on B or C, at least outside D1. This is the case in interrogative expressions: e.g., ' I don ' t know i fB (or not) ' , ' I wonder if . . . ' , French: ' je ne sais si ...', ' je me demande si ...'. Here, instead of a protasis if, w e get a sort of semantic apodosis tf: e.g., since A, I wonder if B (and if, in fact, B, it's because of A). Let us compare the basic conditional (see Fig. 2) with the interrogative (see Fig. 3).
if A (then)
B
C
Fig. 2. The basic conditional.
(since A) I wonder i f / B
~ ( o r not)
Fig. 3. The interrogative.
8 Sweetser (p. 142) writes: "Natural-language conditionals express a relationship and a dependency, not only between the truth values of the two clauses, but between their contents as well. Von Wright's (1975) interpretation of conditionals as rooted in the idea of possible causal intervention captures this content relationship very effectively (cf. also Lewis 1975)". In fact, yon Wright's possible causal inter- vention supposes that something is wanted (say, C, in the triad ABC) and that somebody wants to inter- vene in order to obtain it. Therefore, an accessible object of intervention is sought for (A), such that it causally makes C accessible as well. Now, by doing A in order to obtain C, the doer admits the risk of obtaining B instead. Actions are never sate, but they are meaningful if that which is the wanted result is at all among their possible outcomes. One can always try. From the intentional point of view, 'doing' is 'trying to do'.
290 Book reviews / Journal of Pragmatics 25 (1996) 281-302
Sentences embedding interrogative /f-clauses are D2 (wonder) or D3 (ask), not DI . This is due to the embedding verb. The underlying semantic fact might be that interrogative ~ are special results of a transfer o f the basic conditional tf in D 1.
The mysterious even i f might also fit into this schematization. Sweetser proposes that even i f A should mean that "certain extremely unfavorable circumstances [X] will nonetheless be sufficiently favorable to ensure Y" (p. 136). So Y will occur even i f it shouldn't . This reading needs some support. Let us assume that A is a very weak possibilizer of B: that is, normally, A leads to C, not to B; but B is still a real- istic possibility, however slight). Now, C is energetically discarded for other rea- sons; therefore, B is forced to happen, since there is no longer an alternative (see Fig. 4). 9
even if A B ~ since not
C
Fig. 4. Even if ...
My intuition is that the since not C part is always to be found ('I'11 do it, even if I don ' t want to ' [because I have to, since I have no alternative]). The closing of the alternative C channels the force of doing towards B.
This schematism leads us finally to consider natural syllogistic reasoning. ~° Con- ditionals are binary (protasis-apodosis), whereas syllogisms are ternary, as in ' i f Socrates is human, then he is mortal (since humans are mortal) ' , or ' i f humans are mortal, then so is Socrates (since he is a human) ' . The conditional sufficiency is a third content, a third part of the epistemic process. Binary conditioning seems to express a natural tendency to suppress one o f the syllogistic premises. But the impli- cated part remains semantically active. It can easily be identified. The general struc- ture includes premise A, premise not-C, and conclusion B (see Fig. 5).
if S. is a human then
S. is mortal
L " K ~ S. is immortal
Fig. 5. L (law): humans are not immortal.
9 Here, we have, in fact, a funnel-like barrier setting - the special case of must that Talmy generalizes. But in our analysis, it remains a special case. ~0 If (even if) Sweetser is right about the difference between formal logic and natural reasoning, and I have no doubt that she is, cognitive linguistics should nevertheless consider the possibility of grounding logic, as such. Observed binary causality - one cause, one effect - is not a sufficient source, since sound natural reasoning combines causal experiences far beyond single observations.
Book reviews / Jourr.,al of Pragmatics 25 (1996) 281-302 291
As mentioned, it can also be phrased as in Fig. 6.
so is S.
then /
if humans are m o r t a l / < ,
Fig. 6. L: even a human called Socrates cannot escape this destiny.
We see that if the protasis contains a design for a law, then the anti-apodosis (the not-C) represents a metalaw protecting it against exceptions. So the basic and gen- eral structure seems to be: A-minor premise, not-C-major premise, B-conclusion. Why not consider this structure as a root form (D1)?
Let us finally consider a case of natural reasoning: ' I f it rains tomorrow, I ' l l prob- ably have to use your car ' (see Fig. 7).
(and if I still must go ...)
If it rains tomorrow
/ use your car p" robably have to ...
~ (rather not)
I'll \
(use my bike and get wet)
(use a raincoat and my bike)
Fig. 7. If it rains tomorrow ...
The coalescent branches yield a negative major premise, e.g., ' I want to avoid using my bike and getting wet ' . This is understood as the motor of the utterance, and it leaves an alternative still open as its conclusion: i.e., 'your car, or my bike and raincoat? ' . It calls for D3-negotiation, but it combines two conditional paths, one causal (rain --4 getting wet, unless), the other intentional (scheduled main doing --~ preferred instrumental doing, unless ...), in an objective epistemic combination. In D1, structures like this one might be grounded in dilemmatic experiences, in which rain-like barriers oppose travel-like doings as forces, and there are no cars and no raincoats. The coalescent outcomes then appeal to the subject 's inner strength (abil- ity to 'undergo ' and endure existence anyway). Once more, if intentionality is admit- ted in D1, the theory of oriented transfer gains further support.
292 Book reviews / Journal of Pragmatics 25 (1996) 281-302
6. Conclusions
As the author stresses in her last chapter, "we cannot escape [sic: funnel!] using the general concept of multiple domain-structures in our analysis of linguistic mean- ing" (p. 147). There are three macrodomains, and these account for semantic change, polysemy, and ambiguity, by a universal transfer tendency by which forms pass from a sociophysical real world (D1) to an epistemic and/or emotional world (D2) and to a speech act world (D3) (hence: D1 --~ D2/D3). Semantics, then, should study the general grounds of these three occurring types of meaning, not only the (also occurring) local semantic features of words or generative sentence superstruc- tures. The morphological equipment of natural languages - their closed class inven- tories - seems shaped for referring symphonically to these three worlds, and for facilitating the transfer of form from macrodomain to macrodomain.
There is now, I strongly feel, a need for a critical philosophical and scientific study of the meaning and the implications of such a cognitive triple world (or 3D) ontology. Linguistics, phenomenology, semiotics, neurobiology, and other disci- plines should contribute to its elaboration or modification. The idea is convincing, and the evidence is strong.
I am less convinced that the best thing to do is to argue, as Sweetser does, "for the necessity of metaphorically structuring domains in terms of each other" (p. 147). Such a metaphorical interstructuring takes place between microdomains within any macrodomain, as well as between macrodomains themselves (as when we see one machine in terms of another, or see organs in terms of machines, or see organs in terms of offices, or see biological processes in terms of speech acts, or see causal events in terms of intentional events [or the reverse], etc.). It is clear to me that this is not what is going on when forms are transferred according to Sweetser's very convincing overall system. We have seen that schematic structures are widely shared by the three macrodomains, and that they are generally rooted in D 1-experience (perhaps prehistorically). But we have also seen that they get specified by the macrodomains. Hence, they must be active, and not simply passive 'targets' of metaphor ballistics. Whenever we create new ideas, or address each other in new ways, we actively use available alternative schemes; only then can good, ' juicy' metaphors eventually emerge. They must be called for. When we act, think, have emotions, communicate, etc., we are not just hit or controlled by blind metaphors. Metaphors, as fixed, frozen, generic alliances of figurative form and schematic struc- ture, are themselves guided by non-frozen uses of schemes in, and on, situations to which we apply our consciousness. The metaphorical A is B actually means A is X, and B is an instance (example) of schematic X-ness. Hence, A itself is inherently ambiguous across the three macrodomains, as is B; but X is connected to neighbor- ing schemes that form the network of things in which we participate. There is a sin- gle, participative, shared, intersubjective life-world that we can refer to in the seman- tic worlds D 1, D2, and D3. Words and sentences alone cannot tell us what we mean.
Book reviews /Journal of Pragmatics 25 (1996) 281-302 293
References
Brandt, Per Aage, 1992. La Charpente modale du sens. Pour une sEmiolinguistique morphogrnEtique et dynamique. Aarhus: Aarhus University Press and Amsterdam: Benjamins.
Brandt, Per Aage, 1994. Dynamiques du sens. Aarhus: Aarhus University Press, Brandt, Per Aage, 1995. Morphologies of meaning. Aarhus: Aarhus University Press. Greimas, A.-J., 1983. Pour une thEorie des modalitEs. In: Du sens II, 57-102. Paris: Editions du Seuil. Lakoff, George, 1987. Women, fire, and dangerous things: What categories reveal about the mind.
Chicago, IL: University of Chicago Press. Lewis, David, 1975. Causation. In: E. Sosa, ed., 1975, 180-191. Petitot, Jean, 1985. Morphogen~se du sens. Pour un schEmatisme de la structure. Paris: Presses Univer-
sitaires de France. Petitot, Jean, 1989a. ElEments de dynamique modale. Poetica et Analytica 6: 44-79. Aarhus: Aarhus
University Press. Petitot, Jean, 1989b. ModUles morphodynamiques pour la grammaire cognitive et la sEmiotique modale.
Recherches sEmiotiques/Semiotic Inquiry 9(1-3): 17-51. Montreal: Canadian Semiotic Association. Sosa, E., ed., 1975. Causation and conditionals. Oxford: Oxford University Press. Sweetser, Eve E., 1982. Root and epistemic modals: Causality in two worlds. In: M. Macauley and O.
Gensler, eds., Proceedings of the eighth annual meeting of the Berkeley Linguistics Society, 484-507. Berkeley, CA: Berkeley Linguistics Society.
Talmy, Leonard, 1981. Force dynamics. Paper presented at the conference on Language and Mental Imagery, May 1981, University of Calitomia at Berkeley.
Talmy, Leonard, 1988. Force dynamics in language and cognition. Cognitive Science 2: 49-100. von Wright, Georg H., 1975. On the logic and epistemology of the causal relation. In: E. Sosa, ed., 1975,
95-113.
H e n k Haverkate, Kees Hengeveid and Gijs Mulde r , eds., Aproximaciones pragmalingtifsticas al espafiol (Difilogos hispfinicos, 12). Amsterdam and Atlanta: Rodopi, 1993. 219 pp.
Reviewed by Victoria Escandell-Vidal, Departamento de Lengua Espafiola, Univer- sidad Nacional de Educaci6n a Distancia, Senda del Rey s/n, E-28040 Madrid.
In November 1992, the Department o f Spanish Studies of the University of Amsterdam organized a round table under the heading 'Pragmalinguist ic Approaches to Spanish ' . Five of the six contributions now collected were presented at this meeting, which was fol lowed by a general debate, also partially transcribed at the end of the book. With the analysis of Spanish phenomena as the c o m m o n denominator, the papers included in this volume explore a wide range o f topics: dis- course operators and connectives (Garrido), uses of echoic questions (Dumitrescu), the contrast between pronominal and empty subject in a pro-drop language (Martin Rojo and Meeuwis), pragmatic uses of non-standard conditionals (Montolfo), expressive and commissive speech acts (Haverkate), and the notion of indirectness (Mulder).
The volume starts with Joaqufn Garr ido 's paper, 'Operadores epistrmicos y conectores textuales ' ( 'Epis temic operators and textual connectives ' ) . The author focuses on the forms incluso ( ' even ' ) , todavfa (,'still', 'ye t ' ) , and ya ( 'a l ready') .
