Download Count/Mass Distinction: A Grammatical and Perceptual Perspective and more Study notes Linguistics in PDF only on Docsity!
P. Schlenker - Ling 1 - Introduction to the Study of Language UCLA, Winter 2006
Meaning I: Truth conditions and Reference^ Introduction to Language^ -^ Lecture Notes 6B
☞ meaning? be able to determine, given any situation, whether the sentence is true or false in that situation. Thus a theory of Goal: (^) We extend to the study of To know the meaning of a sentence meaning the methodology we applied in earlier lectures to, the linguist will say, is at least to know its truth conditions syntax. But what is, i.e. to meaning (=a semantic theory) should at least provide systematic rules for co sentences of the language. But how can truth value to various elements of a syntactic tree. We give a very simple example of such a procedure in this that be done? By giving systematic rules that assign a reference and amputing the truth conditions of the lecture. We then provide a case study in the analysis of reference: the count/mass distinction, i.e. the distinction between words like determiner a, and (iii) water cannot be pluralized; and words like, which (i) can occur in the singular without a determiner, (ii) cannot be preceded by the chair , which have none of properties (i)-(iii). We discuss the relation between this grammatical distinction and the perceptual difference between solid objects and substances.
1 What is meaning? The question might be too complex to be tractable. But the semanticist observes that in order Knowledge of Meaning as Knowledge of Reference and Truth conditions to
know the meaning of a sentences its truth conditions. fruitful strategy in contemporary linguistics. Reducing the question of meaning to the question of truth conditions has proved a very, one has at least to know under what conditions the sentence is true , i.e. to know 1.1 In earlier Lecture Notes, we observed that speakers Syntactic Knowledge vs. Semantic Knowledge of English can in principle distinguish between an infinite number of grammatical sentences (as in (1) (^) b. Tha. The rightmost person in the first row is asleep.e person immediately to the left of the rightmost person in the first row is asleep (1)) and of ungrammatical sentences (as in (2)): c. The person behind the person immediately to the left of the rightmost person in the first row is asleep ... (2) Syntax (as we studied it in past lectures) had as its goal to discover the rules that allow speakers to make such (^) b. *Thea. *Rightmost person the in the first row is asleep. to the left of the rightmost person in the first row is asleep distinctions. However this cannot be the whole story. Consider again the gramm know something more about them sentences can convey information!). And just as we saw in syntax that speakers cannot simp - they know under which conditions each of them is true (this is the reasonatical sentences inly have memorized (1). Speakers all the sentences that are grammatical (there is an infinity of them, and speakers are finite creatures), it is certain that they have not memorized their truth conditions either as well. To see the point more clearly, consider the following series of grammatical sentences: - for the simple reason that there is an infinity of these
P. Schlenker (3) a. John could swim - Ling 1 - Introduction to the Study of Language UCLA, Winter 2006 (^2) b. John's father could swim c. John's father's father could swim d. John's father's father could swim It is clear that no two sentences in this series are true under exactly the same conditions. And yet given any^ e. John's father's father's father co^ etc. uld swim situation this list is true.^1 a speaker of English may (given enough time and patience) determine whether any given sentence in 1.2 How can the speaker achieve this? Providing a systematic answer to this question would lead us far beyond the An Example of How it all Works present course. But we can give an the sentences in (4) Syntax: A grammar that produces the sentences in (3). Here it is: idea of the endeavor by first considering a toy grammar, one that produces al (3) l IP I' VP → → → I VP NP I' Vi NP PN →→ PN, NPJohn 1 's N N I V → (^) i →→ could fatherswim It may be noted that this gr The innovation concerns the ruleammar is similar but not quite identical to those that we discussed in earlier lectures. NP which contains an instance of recursion of NP (reminder: recursion is the situation that occurs when a node o → PN, NP 1 's N, f a given category offered by the rule is applied, we end up with a configuration such as the following, where recursion of NP is indicated by circles: - here NP- can be embedded within another node of the same category). Thus when the 2nd^ option
(^1) This is a simplification. We need to restrict attention to situations in which there is an individual named John, etc.
P. Schlenker - Ling 1 - Introduction to the Study of Language UCLA, Winter 2006 (^4) (5)
each of the sentences in John's father,^ With this background in mind, we can now define some semantic rules that will give truth conditions for etc. refer to. The other part determines whether a sentence of the form (3). The rules have two parts: one set of rules determines what expressions such as x could swim is true, where John, x is any expression. Here are the semantic rules in question; (6) Semantics : Some rules that give the truth conditions of the sentences in (3) (These rules are defined for constituency trees) (i) Reference (ia) (ib) For any expression x, John refers to John x's father refers to the father of what x refers to. (ii) Truth For any expression x, x could swim is true if and only if what x refers to could swim. Let us see how this works in a very simple example. Suppose that John's father was Bill, and that Bill's father was Sam.
- could swim. But rule (ia)According to rule (ii) (Truth), John refers to John, so according to rule (ib) [John's father][[John's fath [John's father]er] [could swim]] refers to Bill. Hence refers to the father of what is true just in case what in the end the sentence is true just in case Bill John refers to; and according to John's father refers to could swim - which is the correct result.
's father John^ 's^ father^ could^ swim
P. Schlenker - If we wish to apply the rules in - Ling 1 - Introduction to the Study of Language (6) to the tree in (^) (5), we need one more step when we compute the reference oUCLA, Winter 2006 (^5) f the subject Bill. Applying rule (ib) to father of what [John's father]'s father [John's father] [John's father]'s father refers to, i.e. to the father of Bill, i.e. to Sam. Hence in the end the sentence. We already established in the preceding example that, we can now establish that the latter expression refers to the [John's father] refers to [[[John's father]'s father] [could swim]] It is worth noting that this little procedure, which has only three rules, can be a is true just in case Sam could swim. pplied to sentences of arbitrary complexity. In fact, it suffices to give the truth conditions of all the sentences in Note: We can see here that syntactic trees have an advantage that was not discussed in previous lecture (3). s: they give us a very easy way to define semantic rules (=rules of semantic interpretation). In particular, one of the major syntactic constituents, NP, turns out to also be a major semantic constituent, since rule (ib), which computes the reference, is defined for NPs. This fact is very general: the constituents created by the syntax are the natural units
for the computation of reference and truth conditions. 2 A Case Study in Reference: the Count/Mass Distinction
between properties of this distinction, and then discuss its possible relations to categories^ In this section we study an important gra count nouns (e.g. book or chair ) andmmatical distinction in the domain of reference. We distinguish mass nouns (e.g. water). We first establish the grammatical of perception. Differences between English and Chinese are discussed at the end of these Lecture Notes [the entire discussion is largely inspired by Gennaro Chierchia's 'Language, Thought and Reality after Chomsky', ms., Milan]. 2.1 • Count terms (e.g. 'book) are those that are compatible with numeral determiners such as 'two', 'three', etc. They The count/mass distinction in English have three related properties: (a) They cannot appear in the singular without a determiner (b) They may appear in the singular preceded by 'a' or 'every' (c) They may appear in the plural. This is illustrated in the following paradigm: (7) a. *John has read book b. John has read a book c. John has read books Other examples: table, chair, car, child... By contrast, mass terms are th have the opposite properties from count nouns:ose that are incompatible with numeral determiners such as 'two', 'three', etc. They (a) They can appear in the singular without a determiner (b) They cannot appear in the singular preceded by 'a' or 'every' (c) They cannot appear in the plural.
P. Schlenker they cannot possibly have elaborated a theory of solid objects from experience. M. Hauser has - Ling 1 - Introduction to the Study of Language UCLA, Winter 2006 (^7) pushed this line of inquiry further, showing that in fact primates like rhesus endowed with a similar theory of discrete objects vs. substances (cf. Hauser 1996xx)." monkeys are (ii) It also appears that the distinction between objects and substances guides children in the acquisition of words. As Chierchia writes to summarize results due to Spelke, Carey and Soja: "Going back to children, it is highly plausible that such knowledge, which children appear to be endowed with at birth, guides them as an identification criterion for novel objects vs. substances they encounter; and, later on, such a knowledge guides them in acquiring language. For example upon encountering a class of solid objects, say bottle openers, the child identifies some key properties of the objects (say, shape and function); then generalizes it to other objects of same sort (forming the concept of a uniform class of objects, bottle openers in general). Upon the encountering, for instance, a new paste or a new powder, one again identifies some of its key properties (in this case it won’t be shape but, say, texture a then one generalizes such properties to other portions of the same substance (...). Knowing thatnd what one typically does with it); things are set up in this way (i.e. that they are naturally sorted in substances and objects) makes identification and nam two categories accordingly." [Gennaro Chierchia, 'Language, thought and reality after Chomsky',ing easier. When language comes in, common nouns will naturally fall in To be a bit more specific, here is the experiment t^ draft, University of Milan-Bicocca] hat Chierchia refers to. Soja, Carey and Spelke tested 2 year- olds who had not mastered the which (a) a solid object, or (b) a substance was shown to them. The experimenter woul question and say: 'This is my blicket' [note that the linguistic context does not disambiguate between count linguistic form of the count/mass distinction. They were put in a condition ind point to the thing in and mass here; for instance you may say: 'this is my chair' (count) or 'this is my wine' (mass)]. The child was then shown sample; while the other sample had the two further samples, one of which had the same texture but a (^) different shapesame shape but a, as illustrated below: different texture as the original
P. Schlenker - Ling 1 - Introduction to the Study of Language UCLA, Winter 2006 (^8) Object Trial Substance Trial Named Stimulus
Test Stimulus
Children overwhelmingly generalized on the basis of texture [Nancy Soja, Susan Carey, Elizabeth Spelke, in the case of substances. ‘Ontological categories guide young children’s inductions of shape in the case of solid objects , and on the basis of word meaning: Object terms and substance terms’. 2.3 Does the count/mass distinction reflect categories of perception? Cognition, 38 (1991): 179-211] perception: count nouns refer to solid objects, while mass nouns refer serious flaws.^ From this one might be tempted to conclude that the count/mass distinction simply reflects categories of to substances. However such a theory has (i) First, the count/mass distinction applies to nouns that refer to abstract notions, which are not perceived to begin with. Here are some examples: (10) a. generosity (mass) b. kn c. pride (mass)owledge (mass) vs.vs.vs. virtues (count)beliefs (count)prejudice (count) (ii) Second, it is possible to find pairs of a count and of a mass noun which apparently refer to exactly the same^ d. fun (mass)^ vs.^ joy (count) thing:
Children generalize on the basis of shape Children generalize on the basis of texture
P. Schlenker 2.4 A Hypothesis: Mass nouns are intrinsically plural. - Ling 1 - Introduction to the Study of Language UCLA, Winter 2006 (^10)
can we between^ The foregoing considerations suggest that the count/mass distinction is a specific featur explain (7) (book) and the behavior of mass nouns relative to count nouns, or do we have to take the contrasts we saw (8) (water) to be primitives of our theory? Interestingly, these contrasts can be explainede of grammar. But if of water and mass nouns. Some linguists (e.g. Chierchia, mass nouns are taken to be intrinsically plural. , i.e. as something like a plural. Of course there still has to be (^) op. cit. The idea is that) have suggested that count nouns (whether singular or water some is understood as something like difference between plural count nouns the parts plural) give information about what the elementary (='atomic') parts are, whereas mass nouns don't (in fact information about th plural but they do not provide information about the elementary parts e elementary parts is necessary in order to count things). Thus mass nouns are intrinsically - i.e. they do not 'care ' what the elementary parts are (by contrast, 'book' or 'books' come such information is provided by 'water') with substances, we naturally use a mass term. The word learning experime_._ As a result, whenever we cannot find any elementary parts, as is the case equipped with information about what the individual books are; nont that was reported above shows that children know this (i.e. when they cannot find any elementary parts, as is the case with a substance, they treat the word that refers to it as mass, not count). How does the hypothesis that mass nouns are intrinsical The crucial observation is that (i) The may appear without a determiner plurals behave exactly like mass nouns: ly plural account for the contrast between (7) and (8)? (ii) They may not appear with the det (iii) They may not take an (additional) plural marker (12) a. John has read books erminer 'a' or 'every'
If we put together the facts of^ b. *John has read a books^ c. John has read books-s (7), (8) and (12), we obtain a particularly clear picture, in which singular count nouns behave in one way, and mass nouns and plural count nouns behave in another way. In this fashion the behavior of mass nouns can be taken to have been explained (at least in part). Singular Count Nouns Plural Count Nouns John has read bookJohn has read books John has read a bookJohn has read a books John has read booksJohns has read books-s Mass Nouns John has drunk water *John has drunk a water *John has drunk waters 2.5 • There are interesting differences among the world's languages in the treatment of the count/mass distinction. In The Count/Mass Distinction in Chinese Chinese, every noun is mass. This immediately raises a question: how do Chinese speakers ta things? In the same way as English speakers count before 'cattle' to specify what is the unit into which the reference of heads of cattle, (^) i.e. by having a measure word ('head') come cattle is to be divided. This is possible evenlk about numbers of though cattle is a mass term, as shown by the following tests:
P. Schlenker (13) a. cattle - Ling 1 - Introduction to the Study of Language UCLA, Winter 2006 (^11) b. *a cattle c. *cattles (14) In Chinese, then, the only way one can speak about numbers of things is by sticking a measure word or 'classifier' a head of cattle two heads of cattle right before measure word (from Chierchia, 'Reference to Kinds Across Languages', 1998): the noun. This gives rise to the following kinds of examples, where CL stands for 'classifier', i.e. (15) a. yi ‘one (grain of ) rice’ one (^) li CL (^) mi rice b. liang ‘two (grains of ) rice’ two li CL (^) mı rice c. yi one ‘one (piece of ) table’ zhang CL zhuozi table d. liang ‘two (pieces of ) table' two zhang CL zhuozi table
