In many languages, conditionals are reported to be closely related to the temporal domain. For example, in languages like German and Japanese, when and if can be of the same morphological identity; in English, when and if are sometimes interchangeable to mark hypotheticality. Specifically, studies for different languages
show that while both conditional and temporal clauses can mark conditionality, the latter usually express lower degree of hypotheticality than the former. The difference lies in the different speaker’s attitude or epistemic stance associated with these two clauses: conditional connectors like if are usually endowed with a neutral/negative epistemic stance, and therefore they represent higher degree of hypotheticality; on the other hand, temporal connectors like when are associated with a positive epistemic stance; accordingly, they show lower degree of hypotheticality.
Reilly (1986) approaches the close relationship between conditional and temporal clauses from the point of their acquisition. He shows that when temporal and conditional clauses in English share a variety of characteristics (Reilly 1986: 312): (a) they both link simultaneous or sequential events, often implying a causal relationship;
(b) they both can occur either pre- or post-main clause; (c) in both constructions, the different semantic types are distinguished by the auxiliary verb. Furthermore, he discusses the semantic overlap and non-overlap of when and if. The highest degree of semantic overlap occurs in cases where there is a regular co-occurrence relationship between two events, for instance:
(2.27) If/When Jamie drinks cranberry juice, he gets a rash.
(Reilly 1986: 313)
(2.27) is the so-called ‘generic conditional’,4 which signals a regular co-occurrence between ‘Jamie drinking cranberry juice’ and ‘Jamie getting a rash’. When could be replaced by whenever to reflect this regular relationship. In addition, since if-clauses refer to a possible instance of such regular co-occurrence, they are also acceptable in these cases.
4 For the discussion on ‘generic conditionals’, please refer to Fillmore 1990: 152; Dancygier 1998:63-65; and Athanasiadou and Dirven 2002.
The semantic overlap decreases for conditionals which refer to future and past events, as in
(2.28) If
the strawberries are in, we’ll make fresh strawberry pie.
When (2.29) If
it rained last year in Egypt, the Nile Delta flooded.
When
(Reilly 1986:313)
There is a significant difference lying between if/when in the sentences above;
specifically, they express different degrees of expectation or certainty: when implies certainty, or at least the speaker’s expectancy of the occurrence of the event expressed in the antecedent clause, whereas if signals the speaker’s supposition of antecedent event. Therefore, in predictive when temporals like (2.28), the antecedent is expected to occur, and in past temporals like (2.29), the antecedent is in fact known to have occurred. However, for both predictive and past conditionals, the speaker is supposing the antecedent, i.e. the antecedent is only a possibility.
Reilly then summarizes that there is semantic overlap for when and if where they refer to situations occurring, having occurred, or predicted to occur, in the real world. The interchangeability of conditionals and temporal adverbs increases with the degree of certainty that a speaker has concerning the antecedent. Therefore, it appears that the more regular the co-occurrence relationship between the events in the antecedent and the consequent, the more interchangeable the when and if structures. In addition, when-clauses are restricted to refer to facts and reality, whereas if-clauses suppose the possibility of states or events in potentially real as well as irrealis situations.
Reilly’s discussion clearly shows that there is indeed semantic overlap between
conditional and temporal clauses: the interchangeability of conditionals and temporal adverbs increases with the degree of certainty a speaker has concerning the protasis.
On the other hand, the discussion also elucidates another important notion: there indeed exists some semantic difference between conditional and temporal clauses.
This difference lies in one crucial distinguishing feature: the speaker’s certainty toward the protasis event or state. When the speaker believes it to be fact, when structure is preferred; when the speaker is merely supposing the possibility of its existence, if is employed. In other words, while when expresses a positive epistemic stance, if shows a neutral/negative epistemic stance.
Akatsuka (1986) also goes deep into the notion of ‘speaker attitude’ to account for the semantic difference between conditional and temporal clauses. Discussing the identity of if and when in languages such as Japanese and German, Akatsuka asserts that if can be distinguished clearly from when as long as we take ‘speaker attitude’
into consideration: only when the speaker is uncertain about the realizability of p that we get the ‘if p’ reading in these languages; on the contrary, when the speaker explicitly commits him/herself to the factuality of p, we get the ‘when p’ reading.
Therefore, in the following example:
(2.30) Japanese:
Syuzin ga kaette ki-tara, tazune masyoo husband SBJ. returning come if/when ask will ‘If/When my husband comes home, I’ll ask’
German:
Wenn mein Mann zuruck kommt, werde ich fragen When/If my husband back come will I ask ‘If/When my husband comes home, I’ll ask’
(Akatsuka 1986: 344-45)
When the speaker is not sure whether her husband will come home or not, the sentence such as (2.30) is a conditional expression. On the contrary, when the speaker takes for granted that her husband will come home, it is a temporal expression.
By pointing out the role that ‘speaker attitude’ plays in the interpretation of conditional and temporal clauses, Akatsuka elucidates an insightful concept that conditionals and temporal clauses are in fact very different in nature: in a conditional expression, the speaker does not commit him/herself to the factuality of p, while in a temporal clause, the speaker regards p as a fact. It is this difference in ‘speaker’s attitude’ that evidently distinguishes when-reading from if-reading in the languages where when and if are of the same morphological identity.
Grundy (2000) also touches upon the issue of the semantic overlap/difference between when and if from the pragmatic point of view. He explains the semantic difference between if and when by means of the pragmatic concepts ‘presupposition’
and ‘implicature’: while temporal clauses introduced by when typically give rise to presuppositions, real conditionals introduced by if give rise to implicatures. In cases where when marks a temporal frame, the embedded when-clause provides a background against which some event is highlighted. In other words, the proposition contained in the temporal clause is identified as a presupposition, as the example When I started this book, I thought I’d never finish it (Grundy 2000:124), where the proposition I started this book is presupposed. On the contrary, conditional conjunctions such as if give rise to implicatures of possible existence in respect of the situation described in the if-clause5. The sentence If you see her, tell her to call me
5 Grundy (2000) distinguishes two types of if-clause in terms of their status as a presupposition or an implicature. Backward-looking conditionals are counterfactual and give rise to a presupposition of non-existence in respect of the situation described in the if-clause, as in the following example: If you had sent me a Christmas card last year, I would have sent you one this year (Grundy 2000:125). The sentence presupposes that you did not send me a Christmas card last year. Forward-looking conditionals, on the other hand, give rise to implicatures of possible existence in respect of the situation described in the if-clause, like the sentence If you see her, tell her to call me back, where two implicatures are inferred: possibly you see her/possibly you do not see her.
back will give rise to the implicature possibly you see her/possibly you do not see her, and it is usually the wider context that helps the addressee to understand which of the two possibilities is the likelier.
However, temporal and conditional clauses in English do not always have such a clear-cut distinction. Like German wenn, which can have both a temporal and a conditional meaning, English when too sometimes has a conditional meaning. This is where temporal clauses and conditional clauses come to have certain degree of semantic overlap. In cases where when is used to mark a conditional meaning, the presupposition usually associated with the temporal clause would no longer hold.
When uttering the sentence When you get the chance, you should come to Durham (Grundy 2000: 130), the speaker does not seem to presuppose you will get the chance.
Similarly, in the sentence Come to Durham whenever it suits you (Grundy 2000: 130), the presupposition it ever will suit you is not obtained. These two examples suggest that the presupposition supposedly associated with temporal clauses is in fact pragmatic matter—only when the speaker is certain about the actuality of the protasis does it give rise to a presupposition; when the speaker is uncertain about the actuality of the protasis, the presupposition is not retained. In other words, in the cases where when-clauses do not trigger presuppositions, they behave as conditionals.6
Following this line of reasoning, we find that though both if-clauses and when-clauses can both mark conditionality, there exists a pivotal difference between the two clauses: while if-clauses give rise to the implicature possibly p/possibly not p, when-clauses, on the other hand, usually give rise only to the implicature (,or the so-called ‘pragmatic presupposition’) possibly p. In other words, if-clauses do express
6 Note that the term ‘presupposition’ Grundy employs in the paragraph above is ‘the presupposition in the traditional sense’, i.e. the presupposition which cannot be cancelled. It is true that when when-clauses are used to mark conditionality, they do not trigger presuppositions in the traditional sense; however, they still trigger ‘pragmatic presuppositions’. Pragmatic presuppositions behave more like implicatures in that they are cancellable.
higher degree of hypotheticality than when-clauses do.
Grundy gives an interesting instance to illustrate the difference between if-clauses and when-clauses. Grundy says that in fire-warning notices in lifts, we prefer If there is a fire, do not use the lift to When there is a fire, do not use the lift (Grundy 2000:120). The reason is that when presupposes the existence of p, but if does not. Therefore, in this case where p is undesirable, the presupposition-neutral if is preferred to avoid the presupposition of the existence of the fire.
Based on the fire-notice example, Grundy further proposes that there is a scale of presupposition—from clearly presuppositional at one end to clearly non-presuppostional at the other, with different linguistic structures employed:
When there is a fire — presupposition triggering In the event of fire — ? presupposition triggering In case of fire — ? non-presupposition triggering If there is a fire — non-presupposition triggering
(Grundy 2000:120)
We relate this scale of presupposition to the notion of hypotheticality. When gives rise to the implicature (or the ‘pragmatic presupposition’) possibly p; In the event of and in case of give rise to the implicature (or the ‘pragmatic presupposition’) possibly p or the implicature possibly not p; If, in this case, does not give rise to pragmatic presupposition, but only brings about the implicature possibly not p. These different implicatures indicate different degrees of hypotheticality. When shows low degree of hypotheticality because the speaker holds a positive epistemic stance toward p. In the event of and in case of exhibit medium degree of hypotheticality since the speaker holds a neutral epistemic stance toward p. If signals high degree of hypotheticality because the speaker in such case may intend to hold a negative
epistemic stance toward p.
This scale, then, shows clearly that tense-variation is not the only way to express different degrees of hypotheticality in English conditionals. Various lexical markers such as when, in the event of, in case of, and if also exhibit different degrees of hypotheticality.
Like Reilly, Akatsuka and Grundy, Givon (2001:301-25) notices the correlation of when- and if-clauses in English. In discussing the epistemic modalities, he approaches the distinction of realis (not involving hypotheticality) and irrealis (involving hypotheticality) in communicative-pragmatic terms. He redefines the four traditional epistemic modalities (necessary truth, factual truth, possible truth, and non-truth) by communicative equivalents, as shown in the following table:
Table 2.2 Epistemic modalities in Givon (2001)
Epistemic modalities
logical tradition communicative equivalent a. necessary truth presupposition
b. factual truth realis assertion c. possible truth irrealis assertion d. non-truth negative assertion
(Givon 2001:301)
According to Givon, the logical tradition is defected in that it treats modality as a property of proposition detached from their communicative context. He holds that, on the contrary, the communicative-pragmatic interpretation of the four modalities is mainly based on the epistemic states and communicative goals of the participants. In the case of presupposition, the proposition is taken for granted to be true and left unchallenged by the hearer. In the case of realis assertion, the proposition is strongly
asserted to be true; though challenge from the hearer is possible, the speaker has evidence to defend his/her belief. In the case of irrealis assertion, the proposition is weakly asserted to be either possible, likely or uncertain. In other words, the speaker is not ready to back up the assertion with evidence; and challenge from the hearer is expected. In the case of negative assertion, the proposition is strongly asserted to be false. In the face of the challenge from the hearer, the speaker has evidence to back up his/her belief.
In English, many types of adverbial clauses fall under irrealis scope. Based on different degrees of epistemic certainty, Givon provides a four-point scale for irrelais Adv-clauses in English, as shown below:
Certainty scale for irrealis Adv-clauses in English Highest certainty
a. Irrealis ‘when’
When she comes, we will consider it.
b. Irrealis ‘if’
If she comes, we will/may consider it.
c. Subjunctive ‘if’
If she ever came, we would/might consider it.
d. Counterfact ‘if’
If she had come, we would have/might have considered it.
Lowest certainty
(Givon 2001:324)
This scale is related to our study in two ways. First, this scale can be interpreted in terms of hypotheticality, with irrealis when indicating the lowest degree of hypotheticality and counterfact if denoting the highest degree of hypotheticality.
Second, the scale shows the semantic overlap and difference between if and when in English. When, when used to deal with future events that are yet to occur, does not trigger presupposition in the traditional sense (cf. Grundy 2000: 124) and therefore
falls into irrealis scope. In other words, it expresses conditionality as if does. However, irrealis when is still different from irrelais if in terms of hypotheticality. Though not triggering presupposition, irrealis when gives rise to the implicature possibly p, so it is associated with higher epistemic certainty/lower hypotheticality than irrealis if, which gives rise to implicature possibly p/possibly not p.
To sum up, Reilly (1986), Akatsuka (1986), Grundy (2000), and Givon (2001)’s discussion show that conditional and temporal clauses do have certain degree of semantic overlap, evidenced by the fact that in English when sometimes has a conditional meaning, and also by the fact that in languages such as German and Japanese, if and when show morphological identity. Despite the semantic overlap, however, there exists one important difference between if- and when- clauses when both of them are employed to mark conditionality: they are associated with different
‘speaker attitude’. When- clauses usually show speakers’ positive epistemic stance and therefore express lower hypotheticality; if- clauses, on the other hand, are often associated with speakers’ neutral/negative epistemic stance and thus indicate higher hypotheticality.
2.5 Studies on Mandarin Conditionals 2.5.1 Li and Thompson (1981)
Li and Thompson point out that in Mandarin, the conditional clause may be introduced by a forward-linking adverb that signals ‘conditional’, such as rugu 如果, jiaru 假如, jiashi 假使, yaoshi 要是, or it may be expressed by forward linking with no linking element, like the sentence baba qu, wo gen ta qu 爸爸去,我跟他去 (‘If father goes, I’ll go with him’) (Li and Thompson 1981: 633). Based on English conditional sentences, they explain three types of messages that conditional sentences can express. The first one is ‘reality’, describing a conditional relation between two
propositions referring to the real world, like If you heat water to 100 degrees, it boils (Li and Thompson 1981: 646). The second one is ‘imaginative hypothetical’, which expresses a proposition which could be true in some imaginary world, such as If I saw the queen, I’d bow (Li and Thompson 1981:646). The third type is ‘imaginative counterfactual’, depicting a proposition that could have been true but was not, like If you’d listened to me, you wouldn’t have suffered (Li and Thompson 1981:646). With regard to these three types of conditionals, Li and Thompson indicate that while English makes grammatical distinction among these three types in the auxiliary verbs and tense and aspect markers, Mandarin has no such grammatical distinctions.
Therefore, the sentence ruguo ni kan dao wo meimei, ni yiding zhidao ta huaiyun le 如果你看到我妹妹,你一定知道她懷孕了 can receive all types of interpretations: (1) Reality: If you see my younger sister, you’ll certainly know that she is pregnant. (2) Imaginative hypothetical: If you saw my younger sister, you’d know she was pregnant.
(3) Imaginative counterfactual: If you had seen my younger sister, you would have known that she was pregnant. Without grammatical distinctions, which type of message is conveyed by a Mandarin conditional construction is inferred by the hearer from the proposition in the second clause and from his/her knowledge of the world and of the context in which the sentence is being used.
Li and Thompson make it clear that in Mandarin conditionals, an overt linking element is not obligatory. In addition, they elucidate the importance of the ‘context’
on the interpretation of Mandarin conditionals. Furthermore, they point out the fact that Mandarin conditionals do not employ auxiliary verbs and tense and aspect markers to differentiate degrees of hypotheticality. However, they do not try to account for how hypotheticality is differentiated in Mandarin conditionals. As native speakers of Mandarin Chinese, we believe that the presence/absence of conjunctions and also the choice of conjunctions are the main indicators for hypotheticality in
Mandarin conditionals.
The study then inspires us in two ways: first, when exploring Taiwanese conditionals, the sentences with no conditional conjunctions cannot be neglected, because Taiwanese is like Mandarin in that an overt linking element is not obligatory in the conditional construction; second, Taiwanese conditionals, again like Mandarin conditionals, do not make grammatical distinctions in the auxiliary verbs and tense and aspect markers to differentiate degrees of hypotheticality. Following this, we will investigate in our study how hypotheticality is differentiated in Taiwanese conditionals.
2.5.2 Su (2004)
Su (2004) explores the correlation of forms and their meanings (communicative needs) of Chinese conditionals from an interpersonal and interactional perspective. By identifying and calculating Chinese conditional types7, she finds that when language use is geared toward communication, we humans tend to convey primarily our subjective opinions through our language. Though conditionals are used for semantic
Su (2004) explores the correlation of forms and their meanings (communicative needs) of Chinese conditionals from an interpersonal and interactional perspective. By identifying and calculating Chinese conditional types7, she finds that when language use is geared toward communication, we humans tend to convey primarily our subjective opinions through our language. Though conditionals are used for semantic