This section covers the material in File 7 in Language Files.
Just as syntax is the study of sentence structure and morphology is the study of word structure, semantics is the study of meaning.b) Two issues
- the problem of meaning
- meaning relationships
- among words
- in sentences
c) Problem: Just what is meaning?
That is, just what do we mean by meaning? The issue is trickier than we might think. Here's some ideas. All problematic as explanations in and of themselves.d) Dictionaries
One place people go to find out what a word means is the dictionary. This is problematic. It is, in fact, something that I find irritating. Why?This brings us to the other problem with looking towards dictionaries to understand what "meaning" is. Dictionaries define words in terms of other words; this is circular and can keep you up at night if you've been drinking too much coffee. Here's a trivial example taken from Language Files:
- Because dictionaries only reflect the way people use language; they don't dictate how words are used. Good dictionaries do a stellar job of attempting simply to document how are particular speech community uses words. But some dictionaries attempt to do more. The Royal Academy of the Spanish Language is Spain, for example, has the irritating habit of believing that they can dictate what is and what isn't a word in Spanish. So, for years people have been using the term "hippie" in Spain to mean just what it sounds like. But the academy took over 20 years to "accept" it as a word in Spanish. This is silliness. If native Spanish speaking people are using the word, heck, it's a word in Spanish. By the same token, a dictionary can't dictate the meaning of a word. A dictionary can only attempt to reflect that meaning in terms of other words.
An even more problematic example is this kind of runaround:
- "ectomere: a blastomere that develops into an ectoderm". To understand this definition you also need to understand all the words in the definition.
This is an egregious example of circularity. But it again points to the problem of dictionaries inability to escape words in trying to characterize the meanings of words. In some sense, it seems necessary to view meaning independent of language. Think back to the basic notion of Mentalese that Pinker advocates. Having bashed them, it is important to say that dictionaries are practical and useful for people who already know a language. It's just that they don't really tell us what meaning is in a deeper sense.
- pride: the quality or state of being proud
- proud: feeling or showing pride
e) Mental Images
In an attempt to escape language, some people have argued that meaning can be best understood in terms of mental images. That is, meaning is the mental image conjured up by a word. This seems to work pretty well for examples like this: Big Ben. You probably get a strong mental image of a big brown tower with a clock (I think there's a flag somewhere in my mental image but I'm waivering).Anyway, there are problems with this. Here's a list:
- Words can conjure up completely different mental images in different people without varying so substantially from one another in meaning (example: lecture, speed trap, operating room). My mental image of "lecture" is different from yours, but if I say to you, "I'll see you in lecture today," I'd say we could make a pretty strong case that "lecture" means the same thing for both of us in that sentence.
- Images tend to be prototypical: bird (penguin), dog (chihuahua). Penguin doesn't fit our standard mental image of a bird. But, we still know that penguin is included in the meaning of bird. This is problematic if meaning is simply a question of mental images.
Bottom line. If we want to advocate a theory of meaning based primarily on viewing meaning in terms of mental images, we have to address the serious challenges raised by these problems.
- Some words have no images: (remind, doubt, ponder, thrive). This is obviously problematic. If something has no mental image, does it have no meaning? What's the mental image of "the"?
f) Meaning and Reference
What if we assume that meaning is the thing that a word points to in the world (its referent). Well, here's some of the problems that arise.Not all words point to real things in the world (Santa Clause, Klingon). Simply put, words can point to completely imaginary things and we can clearly know what these words mean.
There's also a logical issue here. Consider the following set of sentences.
The third sentence is not true. Yet, Cassius Clay and Mohammed Ali refer to the same person. Here's another set of examples.
- Cassius Clay is Mohammed Ali.
- Cassius Clay knocked out Sonny Liston.
- ??Mohammed Ali knocked out Sonny Liston.
What's the deal? If Bill Clinton and the President refer to the same individual in the world, why can't we substitute one for the other? That is, why is the last sentence not a paraphrase of the third sentence? Bottom line. If we want to advocate a theory of meaning based primarily on viewing meaning in terms of mental images, we have to address the serious challenges raised by these problems.
- Hillary Rodham Clinton is married to Bill Clinton.
- Hillary Rodham Clinton is married to the President.
- John wants to know if Bill Clinton is the President.
- ??John wants to know if Bill Clinton is Bill Clinton.
g) Sentence Meaning
Okay, so we haven't arrived at a simple understanding of meaning. But, we can see that meaning must have something to do with the way language relates to the world, and mental images and reference must have something to do with this, even if they are not, in and of themselves, the who enchilada. One other issue that we must face is that we not only need to understand how words mean, but also, how combinations of words (e.g. sentences) mean.
h) Meaning and Truth
One way that linguists and philosophers of language have dealt with this problem is to explore the possibility of sentence meaning in terms of an understanding of what it of takes for a sentence to be true. Consider the following sentences.i) Meaning and Language UseOur knowledge of what this sentence means can be characterized in terms of our knowing what it takes for the sentence to be true. The first sentence would be true just in case the individual denoted by the name Jon Bon Jovi had the quality of not having hair on his head. The second sentence would be true just in case the man denoted by the name Bob Dole engaged in the activity of residing in the place denoted by the name Washington.
- Jon Bon Jovi is bald.
- Bob Dole lives in Washington.
Think about how this is useful for sentences like "Santa Claus lives in the North Pole." We can understand what such a sentence means, because we know what it would take for it to be true. We'd need to identify a possible world in which an individual who delivers presents at Christmas with the name Santa Claus engaged in the activity of residing in the place denoted by the name North Pole.
Okay, now remind yourselves of this pair of sentences.
What we see is that the conditions of being president and of being Bill Clinton are not the same. Hence, one cannot be substituted freely, even though they refer to the same person. Calculating sentence meaning in terms of what it takes for a sentence to be true. The truth conditions for these two sentences are quite different. One will always be true (at least while Clinton is alive) while the other will cease to be true in a few months.
- Bill Clinton is the President.
- Bill Clinton is Bill Clinton.
It's worth noting that many sentences can't be viewed in terms of truth conditions. How do we know whether a question is true? A command? The meaning of these other types of sentences require other tools, tools which are discussed in file 8 on Pragmatics, the study of language in context or in use.
j) Relations between words
This section will review meaning relationships between/among words. We're just scratching the surface here, but it is important to see that much of what it means to know words is to know how they are related to other words in our lexicon. This helps to give us a sense that our mental dictionary is not just a messy box full of words. Rather, it is a complexly structured machine.k) Synonymy
Synonymy refers to a situation in which there is a clear meaning overlap between words. This is a gradient phenomenon in that we have a sense that some words are closer synoyms than others. Pinker (and others) argue that there are no true synonyms, that people are very parsimonious in chunking up the meaning space. Anyway, here are some obvious synonyms in English.
- examples: sofa/couch car/automobile cease/stop prohibit/forbid
l) Homonymy
Sometimes, two meanings share the same phonetic form. That is, two words sound identical but are different in meaning. This is homonymy.
- examples: knight/night know/no bore/bore bank/bank bat/bat or/oar
m) Antonymy
Other times words have opposite meanings. This is antonymy.
n) There are 3 flavors of antonymy
o) Semantic features and lexical decomposition.
- contradictory pairs (you're one or the other but can't be both)
- examples: married/single, male/female
- scalar antonyms (you need not be totally one or the other i.e. you can fall between the two extremes of a particular scale)
- examples: hot/cold tall/short happy/sad(temperature, size, emotional state etc...)
Note that relational opposites are different from contradictory pairs. You don't have to be a cop if you're not a robber. But, if you aren't male, you have to be female (genetically). And note that relational opposites are scalar. There's not a scale between cop and robber on which you fall. By contrast, you can clearly locate yourself on a scale between happy and sad or hot and cold.
- relational opposites (pairs that don't represent extremes of some physical scale)
- examples: stop/go patient/doctor over/under cop/robber teacher/student
One way to analyze meaning is to attempt to break the meanings of words down into yet more basic meanings. This activity has a couple of fancy sounding names. One is lexical decomposition, and the other is componential analysis. What they amount to is basically our attempt to study the ways in which the meanings of some words are built up out of the meanings of others. Consider the examples of the words mare, stallion, hen, and rooster that are given to you in the book. One way you can approach understanding what you know about these words is to see that they all share the property of being animals. You could say they have the feature [+animal]. For their part, mare and hen share the feature [+female], while stallion and rooster share the feature [+male]. The point here is that we can decompose the concept of rooster into more primitive elements. The set of that elements can help encode what we know about the meaning of rooster and it can encode ways in which rooster is semantically related to other concepts. Nouns like these lend themselves more straightforwardly to this kind of lexical decomposition into sets of features than do more abstract nouns or verbs.p) Question: Does horse entail mare?With this notion of decomposition in mind, consider the following pairs of words..
If we look at the first word of each pair, we see that if its meaning holds, so does the meaning of the second. The term your book gives for this is hyponymy. In simple terms, if you are a mare, you MUST also be a horse. That is, the set of mares is always a subset of the set of horses, so we can say that mare is a hyponym of horse. Logically, mare entails horse. If you are a dog, you MUST also be a mammal. If you are a woman, you must also be a human. In all of these cases, the first, more specific word of the pair entails the second.
- mare/horse
- dog/mammal
- woman/human
Rules of entailment:
That's how entailment works. This will be useful when we discuss pragmatics, as sentences have entailment relations. For example, if you hear the sentenc "Chip lives in North Carolina" we know that "Chip lives in the United States." Living in North Carolina entails living in the United States.
- A entails B if and only if:
- 1. If something is an A, it much be a B, too (if something's a mare (A), it is also a horse (B))
- 2. If something is not a B, then it cannot be an A (read: if something's NOT a horse (B), then it CAN'T BE a mare (A).
Nope. Because if you are a horse, you are not necessarily a mare. And if you aren't a mare, you still might be a horse!
q) Semantic decomposition again
How might we understand a bit better how entailment works? Well we can decompose the word mare semantically. This means that we can take its meaning apart to look at the elements that make up "mare-ness" (note that this word is odd precisely because [-ness] is a fairly unproductive derivational suffix, but you all know what it means). I'm going to use capitals here to indicate that all of the decomposition below is actually happening in mentalese rather than in English in particular.
- MARE
- HORSE
- FEMALE
- ADULT ANIMAL
Now we can see the nature of the entailment relation. Mares are a subset of the set of horses because HORSE makes up part of the meaning of MARE. If something isn't a horse, we know it can't be a mare. So, mare entails female and adult-animal as well.By the way, this type of semantic decomposition is important. Here's a sentence:
Componentially, if we break down "lost" we see that it might mean something like "be defeated/not win". But defeat, which means "prevail over X or beat X, i.e. to win." By decomposing lost and defeated we can see why this sentence makes no sense.
- UNC defeated Maryland last Saturday, and we lost the game.
r) What is it?
What we just did above also required interpreting a sentence in terms of the combined meanings of its parts. This is semantic composition, not to be confused with semantic decomposition, which is about looking at what sub-meanings make up the meaning of a word like "mare". Here's a definition:s) Some sentences are ambiguous for syntactic reasons. Look at these:
- Semantic composition: the meaning of a sentence is composed of the meaning of its words and by the syntactic structure in which they are combined.
In these cases, we need to know, for example, whether the PP [with bincoluars] part of the NP [an octopus with binoculars] or not. We need to know whether "moving" and "visiting" are adjectives describing "sidewalks" and "relatives" or whether they are acting as progessive verbs. So, where things are attached in the tree matters in how we interpret the meaning of a sentence. We've already seen this in syntax, and that's part of semantic composition.
- Mary saw an octopus with binoculars.
- They are moving sidewalks.
- We are visiting relatives.
t) Idioms
Idioms, in fact, call our attention to compositionality precisely because they are NON-COMPOSITIONAL. Consider these sentences:u) More on compositionalityThe idiom "kick the bucket" isn't about "kicking" and it isn't about "buckets". The meaning of the idiom cannot be derived from the combined meanings of its parts. This is what makes idioms special. For our purposes, we can use idioms to call attention to the generally compositional nature of our sentences. "I lifted the box" is entirely compositional. Even a sentence like "John kicked the bucket" is compositional to a large degree. If we know that the idiom means "die", then we put that together with "John" and we get the meaning of the whole thing.
- Now the cat is out of the bag.
- He kicked the bucket last night.
- Ralph was just pulling my leg.
- I told him to kiss off.
Syntactic structure alone isn't enough to understand compositionality in sentences. A clear illustration of this come is Adj + N combinations. Consider these examples:What's going on here is that although all three are syntactically identical, they do not mean in the same way. The problem is that we have to dig even deeper and look at the different ways that adjectives work.
- blue suit (intersection of the set of things that are blue and the set of suits?) Yep.
- big mosquito (intersection of the set of big things with the set of mosquitos?) No!
- alleged thief ((intersection of the set of alleged things with the set of thieves?) No!
- fake Picasso (intersection of things that are fake with the set of things that are Picassos?) No!
v) Types of adjectives (and end of semantics review)