Pragmatics and Discourse Analysis Chapter 5

Entailment Meaning

1. Where are we on the road map [of Meaning in Communication] ?

In the last lecture, we had a brief look at the issue of deixis, through which we can learn that the interpretation of some words in a sentence is dependent on the rather complicated factors in the actual context of language use. Deitic expressions are like open variables in logic. Unlike anaphora, they are not bound by any antecedents in the discourse. They are directly linked to referents in the non-linguistic context. This provides one support to the claim that the meaning of sentences is underdetermined and needs to be enriched in the light of pragmatic factors. Such a claim has been sometimes referred to as the Underdeterminacy Thesis.


Some more explanations are in order. We say that a linguistic expression may refer to something in the physical or mental worlds. Such an expression is said to have a referent. So the expression "Jiang Yan" refers to the person working at QT506, who may also be referred to as "that man", "the man", "my best friend", "the most hated", "my dissertation supervisor", etc., by different users of language. The expression "the Dragon King in the East Sea" refers to an individual in a fictitious world. The expression "my dream lover" refers to an individual in my mental world. The expression "a man" refers to an unspecified individual; "horses" refers to a group of horses or sometimes to all the horses generically. All the noun phrases can be used in this way, so can many other expressions.


Anaphoric expressions can also refer, but they refer via their antecedents. That is, anaphoric expressions simply take over the referential values of their antecedents. Hence in (1), "he" is anaphorically linked to "Brutus". The latter refers to the person working at QT506. Hence "he" also refers to that person.


(1) Brutus did not love Caesar less. But he loved Rome more.

If in the discourse, we cannot find the antecedent of a pronoun such as "he" [as in (2)], it would be hard for us to tell what is the referent of "he" in the world, actual or fictitious, unless "he" is used with a pointing gesture, that is, as a deictic term, [as in (3)].

(2) He said he wanted to see your teacher.

(3) Mary wanted to visit the library, but he [pointing to a particular man] said he wanted to see your teacher.

(2) would make sense only if both the speaker and the addressee had in mind a particular person as a prominent referent in their background knowledge. Or they had in sight only one male person. In the former case, it should be understood as if there were an implicit antecedent, for example, John. In the latter case, "he" is also used deictically, just like using "this man".


But all the time, we have been loosely making use of the term "sentence meaning". What is it anyway ? How can we characterize it ? Without clarifying this starting point, all the rest investigations on meaning would be standing on a shaky ground. We thus move on to the discussion of entailment meaning.

2. Entailment as the meaning of a sentence

Smith & Wilson (1979), among many others, regarded the literal meaning of a sentence as a set of propositions that are entailed by the sentence. They also considered the meaning of an utterance as a set of propositions that may or may not be entailed by the related sentence. To understand what they meant, we need to reconsider the meaning of "proposition".


Propositions, according to Smith & Wilson, are abstract objects designed to represent semantic structure while ignoring syntactic and phonological form. Hence two synonymous sentences of English may be said to express the same propositions, regardless of their syntactic or phonological differences; and two synonymous sentences, one from English and one from French, may also be said to express the same propositions. [There are more stringencies on the concept of proposition, which we can ignore at this stage. Cf. Ayer (19 ).] Thus (4) and (5) share the same proposition (6). [Here we use underlined English sentences to represent propositions.]


(4) The football game is over.

(5) The football game has finished.

(6) The football game has ended.

The notion of entailment was originally developed for logic. Transplanted to linguistics, the entailments of a declarative sentence are those propositions that can be inferred from it in isolation from any context: that must be true whenever the sentence itself expresses a true claim. As an example, (7) entails (8):

(7) We've just bought a dog.

(8) We've just bought something.

There are no circumstances in which (7) could be true and (8) false: thus by definition (7) entails (8) [and (8) is an entailment of (7)]. However, (9b) does not entail (10), even though (10) is the proposition the addressee takes the speaker to be intending to convey when uttering (9b) [From now on, we present contextual information in curly brackets {}.]:

(9) {A CBS lecturer went to lecture at a PolyU classroom and found a staff from the Department of English who was also going to use that classroom. Apparently, the classroom was doubly booked. }
a. Staff from CBS: One of us needs to take the students to somewhere else.
b. Staff from English: I'm not going anywhere !

(10) You should go to somewhere else.

(10) is not entailed by (9b) because it is obvious that there are many circumstances in which the latter can be true but the former false. So (10) is part of the utterance meaning of (9b), but not part of its sentence meaning. We will address the issue of how (10) can be inferred from (9b) in later lectures.


So far, we have been presenting our arguments in such a way that the meaning of a sentence is shown to entail some proposition, represented by another sentence. It is in fact also possible to say that a sentence entails another sentence. But what we really mean to say is that the meaning of one sentence entails another sentence expressing proposition Y. Or we can say that sentence A expressing proposition X entails sentence B expressing proposition Y.


Still, we need to see how a sentence can -- and normally must -- entail not just one, but a set of propositions. According to Smith & Wilson, the sentence (11) entails all the propositions in (12):

(11) John stole three horses.


a. John stole three animals.
b. John stole three things.
c. John stole some number of horses.
d. John stole some number of animals.
e. John stole something.
f. John did something to three horses.
g. John did something to three animals.
h. John did something.
i. Someone stole three horses.
j. Someone stole three animals.
k. Something happened.


Exactly how these propositions are obtained and why some others do not qualify as linguistic entailments are technical issues that we cannot address here. Interested people can refer to Smith & Wilson (1979) for more details. But the basic procedure is to substitute each constituent in a syntactic tree with some more general terms, i.e. superordinates. That way, the obtained propositions are always going to be more general in meaning, in some respects, than the original sentence. This shows that entailment relations are no more than the reverse of set inclusion, a relation we already learnt from the study of meaning postulates. Of course, any propositions obtained this way are always going to be true if the original sentence is true. Moreover, Each of such propositions can be turned into a question, of which the original sentence, with a proper positioning of stress, would be an answer. It is important for us to know that a sentence with different distribution of stress promotes a different set of propositions to a semantically more prominent interpretation.


The set of propositions that a sentence can entail, conceived from the above points of view, constitute the entailment meaning of the sentence, also called literal meaning or truth-conditional meaning. Such meaning is certainly not all the meaning a sentence can convey, but is taken as the basic meaning a sentence can carry. It is also the kind of meaning we get when we put the meaning of words together according to the structures of syntax.

3. Truth and Truth Conditions

The notion of entailment is crucially based on the concept of truth, which again is a term used in logic and later borrowed to study formal semantics. Basically, a sentence carry information [a set of propositions] which may be tested against the situation in the external, extra-linguistic world. If the real-world situation confirms the state described in the sentence, then the sentence is said to be true, otherwise false. But it really depends on which external situations [i.e. which model] we are testing the sentence against, because we can conceive many models: the actual, the fictitious, the mental, the mathematical, Alice's Wonderland, Utopia etc. Given that there can be an infinite number of constructed models, what really really matters is how we can determine whether a sentence is true or false, given ANY model. In formal semantics, there are sets of truth-conditions for each type of sentence patterns. Thus for a particular sentence with a certain type, say Subject + Predicate Verb, its proposition will be true if the reference of its subject is included in the reference of its predicate. And the proposition will be false otherwise. As, in simplest cases, a predicate verb denotes a set of individuals and a subject can denote an individual, it is possible to decide whether the reference of the subject is a member of the set referred to by the predicate verb. What we have are the truth-conditions for this sentence type. Other types of sentences have other truth-conditions. Thus the meaning of a sentence is partially defined in relation to its truth conditions. And the meaning of each word in that sentence can be defined in terms of its contribution to the truth conditions of the proposition(s) of the sentence. In this way, we can work out the compositional processes of the meaning of the sentence: i.e. how the meaning of words combine with one another in steps to yield the final proposition which can be truth-conditionally evaluated. For semanticists, the ultimate concern is not the truth conditions of the proposition, but the semantic structures of language. But through studying the truth conditions, we can learn a great deal about the semantic structures of language.


We have already pointed out that truth-conditional meaning is only part of the overall meaning a sentence can carry, which is the realm of semantics. Other aspects of meaning carried by a sentence in use, i.e. as an utterance, are studied by pragmatics. But a familiarity with the truth-conditional meaning, or the entailment meaning, of a sentence is the starting point for pragmatic investigations.

命題 (proposition)


語言是複雜思維的工具﹐思想是語言的內容﹐兩者相輔相成。我們不妨設想某人進行了一番思維活動﹐獲得了某些思維結果即思想。思想的客觀的、不包括人的主觀因素的那部份內容﹐被稱為命題(proposition)。也就是說﹐客觀的思想以命題的形式出現。維特根斯坦 (Wittgenstein, 1922: 3.11)認為﹕“在命題中﹐思想得到了可由感官感知的表達。” 赫福德與希斯雷 (Hurford & Heasley, 1983)也指出﹕“命題是思維的產物﹐…… 一般的思想被認為是私下的、個人的、心理的過程﹐而[作為客觀思想的]命題則是公開的﹐因為同一個命題可為多人獲得並理解……”。

    命題雖有內容﹐但無語音、語法外形﹐屬心理範疇﹐所以從物理特性上看﹐命題是個與句子不同的概念。羅素 (Russell, 1940)指出﹕“一個命題可由任何語言來表述。…… 就是在一個語言堣]可以通過多種[句法]途徑來表達同一個命題。”

    Iacona (2003)在綜合前人對命題的論述時,就命題的特性整理出了以下幾點:


       第二,命題與表述命題的語言相對獨立。一個命題當然需要用一種語言的句子表達出來,但它也可以用其他語言的句子乃至同一語言的其他語句來表達。也就是說,一個命題之所以可以由某個句子來表達,只是因為該句子的意義與相關命題正好吻合 —— 句子所表達的意義就是一個由命題組成的集合。所以,命題無所謂是英語的還是法語的,因為它只有語義特徵,不具語言特徵。某個命題可能在某個語言產生之前就存在了,也可以在某個語言消亡之後繼續存在。



話句義(utterance meaning)是說話者在特定語境中所表達的意義。它可能等同於句義﹐但也可能超越句義﹐有額外附加的意義﹐甚至與句義完全不同。話句義屬命題義﹐所以可稱之為話句的命題義或話句的命題內容(the propositional meaning (content) of an utterance)。但命題義不全是話句義﹐因為只有說出的命題才具話句義﹐而我們卻可把命題儲存在記憶之中﹐不說出來。




    (1) A    屋堨i真冷。


    (2) |=  <這間屋子室溫很低。>


    (3) |=  <這間屋子室溫很低。希望你能做些什麼使屋子變暖一點。>



    (4) A    小張在系主任那堨揮A的小報告呢﹗

       B 他可真是我的好朋友﹗


    (5) |=  <小張真是B君的好朋友﹗>

    (6) |=  <小張真不夠朋友﹗>


    (7) S君簡直是一隻狼﹗


    (8) |=  <S君是一隻茹毛飲血、四肢著地行走的、穴居的兇殘的野獸。>


    (9) |=  <S君為人兇狠﹐如同一隻狼。>


    (10)   他喜歡輕鬆愉快的西洋音樂和中國民樂。


    (11)   |=  <他喜歡輕鬆愉快的西洋音樂﹐也喜歡輕鬆愉快的中國民樂。 或者



    (12)   |=  <他喜歡輕鬆愉快的西洋音樂﹐也喜歡輕鬆愉快的中國民樂。>




Ayer, Alfred. (1936). Language, Truth and Logic. Middlesex: Penguin Books.

Hurford, James. and Brendan Heasley (1983). Semantics: a Coursebook. Cambridge University Press.

Iacona, Andrea. (2003). Are There Propositions? Erkenntnis 58: 325-351.

蔣嚴、潘海華 (1998/2004). 《形式語義學引論》, 北京:中國社會科學出版社。

Peccei, Jean Stilwell.(1999). Pragmatics. Routledge.Chapter 2.

Russell, Bertrand.(1940). An Inquiry into Meaning and Truth. London: Unwin.

Smith, Neil. and Deirdre Wilson (1979). Modern Linguistics: the Results of Chomsky's Revolution. Penguin Books. Chapter 7. [Can be used as an extra reading.]

Wittgenstein, Ludwig. (1922). Tractatus Logico-Philosophicus. 維特根斯坦《邏輯哲學論》,北京:商務印書館,1962


back    next    content