English Semantics

In semantic analysis every proposition contains one predicate and a varying number of referring expressions (noun phrases) called arguments.The predicate may be a verb, an adjective, a preposition, or a noun phrase.
The arguments that accompany the predicate have different semantic functions, or roles, in the proposition. What roles they have depends partly on the nature of the predicate and partly on their own meanings.

The description of a sentence is a syntactic analysis. The description of a proposition is a semantic analysis. A syntactic analysis is an account of the lexemes and function words in a sentence, describing how these combine into phrases, and showing the functions that these lexemes and phrases have in the sentence. There are somewhat different ways of doing syntactic analysis, but generally these sentence functions are recognized: subject, predicate, object, complement and adverbial.
Note that every lexeme and function word is assigned to one of the syntactic functions, subject, predicate, etc., and these functions are listed in the order they have in the sentence.

The semantic analysis deals with meaning, the proposition expressed in the sentence, not necessarily with all the function words in the sentence. In semantic analysis we first separate Inflection from Proposition.

When Inflection—including Tense—is separated from Proposition, we see that the forms of the verb be (am, is, are, was, were) have no meaning. They are clearly part of the syntactic structure of sentences but not of the semantic structure.

An English sentence has certain kinds of modification that, together, we call inflection. Inflection includes tense (the distinction between present walk and past walked, for instance); aspect (are walking, have walked), and modality (may walk, could walk, should walk, among other possibilities). Tense, aspect and modality can be combined, as in were walking, should be walking, would have walked.

A proposition is something abstract but meaningful. It can be expressed in different sentences and in parts of sentences, perhaps with differences of focus but always with the same basic meaning.
And, as you recall, any sentence can be expressed in different utterances, produced by different people at different times and in different places.

2a Helen put on a sweater.
2b Helen put a sweater on.

These are different English sentences, but they convey the same message—they express the same proposition.

3a Richard wrote the report.
3b Richard is the one who wrote the report.
3c The report was written by Richard.
3d The report is what Richard wrote.

We may say that these four sentences also express a single proposition but they differ in focus: 3b and 3c give a special emphasis to Richard, 3d emphasizes the report, and 3a has no particular focus. In the approach taken here, a proposition does not have a focus; a sentence may add a focus and may add the focus in different places and in different ways. The four sentences about Richard embody the same proposition. A proposition, then, can be realized as several different sentences.

A traditional way of defining a sentence is ‘something that expresses a complete thought.’ This definition is a rather strange way of explaining since it assumes that we know what a complete thought is and with this knowledge can determine whether something is or is not a sentence. But surely the procedure must be the reverse. Sentences are more knowable than thoughts. In spite of individual differences speakers of a language generally agree about what is or is not a sentence in their language. Who can say what a complete thought is? Compare these language expressions:

1a We walk in the park.
1b our walk in the park
1c for us to walk in the park

We call the first a complete sentence, and in writing we begin with a capital letter and end with a period. We say the other two are not complete sentences. But all three expressions have the same semantic content, the same relation to an action or possible action performed in a certain place by two or more people, one of whom is the speaker or writer. The difference is grammatical. The first expression asserts something, makes a statement. The other two expressions can be parts of statements, as for instance:

We enjoyed our walk in the park.
It’s not too late for us to walk in the park.

but they do not make assertions by themselves. The formal differences among these three expressions—we, our and us, for example—are a matter of grammar, not semantics.

The semantic content shared by the three expressions is a proposition. A simple statement like We walk in the park expresses asingle proposition, something presented as a fact and therefore subject to verification; generally speaking, one can find out if the proposition is true or false. We don’t walk in the park is the negation of this proposition, and Do we walk in the park? is a question about it.

Truth-conditional semantics is based on the notion that the core meaning of any sentence (any statement) is its truth conditions. Any speaker of the language knows these conditions. If a sentence is true (or false), what other sentences, expressing partly the same,
partly different conditions, can be judged by this sentence? If a given sentence is true, does this make another sentence also true, or does it falsify the other sentence, or is there no truth relation? Matters of truth and logic are of more importance in truth-conditional semantics than meanings of lexemes per se.
We are not yet finished with the dimensions of meaning. Often we derive more meaning from what we hear or read than what is actually in the message. Perhaps this is due to an intuition we have or to the fact that the speaker or writer infers something—hints at some further meaning. In semantics we are not interested in intuitions or hints but we are interested in the instances when the language of the message implicates some additional meaning that accounts for our inference.

We communicate with utterances, and each utterance is an instance of a sentence. But how can we explain what ‘sentence meaning’ is?
Two points are obvious. First, the meaning of a sentence derives from the meanings of its constituent lexemes and from the grammatical meanings it contains. So if you know all the lexical and grammatical meanings expressed in a sentence, you know the meaning of the sentence, and vice versa. Second, at least if the sentence is a statement, if you know the meaning of the sentence, you know what conditions are necessary in the world for that sentence to be true.

Albert Thompson opened the first flour mill in Waterton.

You don’t know whether this sentence is true or not, but you know that if it is true, there must exist (at some time) a person named Albert Thompson and a place called Waterton (presuppositions), that Albert Thompson opened a flour mill, and that there was no flour mill in Waterton before Albert Thompson opened his mill (entailments). You know that if this sentence is true, the sentence Albert Thompson did not open the first flour mill in Waterton is false (a contradiction).

When homonyms can occur in the same position in utterances, the result is lexical ambiguity, as in, for example, “I was on my way to the bank.” Of course, the ambiguity is not likely to be sustained in a longer discourse. A following utterance, for example, is likely to carry information about depositing or withdrawing money, on the one hand, or, on the other hand, fishing or boating. Quite often homonyms belong to different lexical categories and therefore do not give rise to ambiguity. For instance, seen is a form of the verb see while scene is an unrelated noun; feet is a plural noun with concrete reference, feat is a singular noun, rather abstract in nature; and so on.
Ambiguity occurs also because a longer linguistic form has a literal sense and a figurative sense.
Skeleton in the closet can mean ‘an unfortunate event that is kept a family secret.’ With this meaning skeleton in the closet is a single lexeme; with its ‘literal’ meaning it is a phrase composed of several lexemes.

A lexeme is a conjunction of form and meaning. The form is fairly easy to determine: in writing it is a sequence of letters, in speech a sequence of phonemes. But meaning is more difficult to determine.
In homonyms, such as bank ‘a financial institution’ and bank ‘the edge of a stream,’ pronunciation and spelling are identical but meanings are unrelated. In other pairs, numerous in English, such as steak and stake, pronunciation is identical but spelling is different, reflecting the fact that the words were once different in their phonological form. English also has pairs of homographs, two words that have different pronunciations but the same spelling; for example, bow, rhyming with go and referring to an instrument for shooting arrows, and bow, rhyming with cow and indicating a bending of the body as a form of respectful greeting.

Lexicographers and semanticists sometimes have to decide whether a form with a wide range of meanings is an instance of polysemy or of homonymy. A polysemous lexeme has several(apparently) related meanings. The noun head, for instance, seems to have related meanings when we speak of the head of a person, the head of a company, head of a table or bed, a head of lettuce or cabbage. If we take the anatomical referent as the basic one, the other meanings can be seen as derived from the basic one, either reflecting the general shape of the human head or, more abstractly, the relation of the head to the rest of the body.

Dictionaries recognize the distinction between polysemy and homonymy by making a polysemous item a single dictionary entry and making homophonous lexemes two or more separate entries. Thus head is one entry and bank is entered twice. Producers of dictionaries often make a decision in this regard on the basis of etymology, which is not necessarily relevant, and in fact separate entries are necessary in some instances when two lexemes have a common origin. The form pupil, for example, has two different senses, ‘part of the eye’ and ‘school child.’ Historically these have a common origin but at present they are semantically unrelated. Similarly, flower and flour were originally ‘the same word,’ and so were the verbs to poach (a way of cooking in water) and to poach (‘to hunt [animals] on another person’s land’), but the meanings are now far apart and all dictionaries treat them as homonyms, with separate listing. The distinction between homonymy and polysemy is not an easy one to make. Two lexemes are either identical in form or not, but relatedness of meaning is not a matter of yes or no; it is a matter of more or less.

A lexeme may consist of just one meaningful part like these:

arm chair happy guitar lemon shoe

or of more than one meaningful part like these armchair unhappy guitarist lemonade shoehorn The technical term for a minimal meaningful part is morpheme.
Arm, chair, happy, guitar, lemon, shoe and horn are all morphemes; none of them can be divided into something smaller that is meaningful. They are free morphemes because they occur by themselves. The elements un-, -ist and -ade in unhappy, guitarist and lemonade respectively, are also morphemes; they are bound morphemes which are always attached to something else, as in these examples.

A dog barked.
The above is a meaningful sentence which is composed of smaller meaningful parts. One of the smaller parts is the phrase a dog which refers to a certain animal. We call this phrase a referring expression.
A referring expression is a piece of language that is used AS IF it is linked to something outside language, some living or dead entity or concept or group of entities or concepts. Most of the next chapter is about referring expressions. The entity to which the referring expression is linked is its referent. Another meaningful part is the verb bark, which is also linked to something outside of language, an activity associated, here, with the referring expression a dog. We call this meaningful part a predicate. The use of language generally involves naming or referring to some entity and saying, or predicating, something about that entity. The sentence also has several kinds of grammatical meanings. Every language has a grammatical system and different languages have somewhat different grammatical systems. We can best explain what grammatical meanings are by showing how the sentence A dog barked differs from other sentences that have the same, or a similar, referring expression and the same predicate.
Grammatical meanings, then, are expressed in various ways: the arrangement of words (referring expression before the predicate, for instance), by grammatical affixes like the -s attached to the noun dog and the -ed attached to the verb bark, and by grammatical words, or function words, like the ones illustrated in these sentences: do (in the form did), not, a, some, and the. Now let’s return to dog and bark. Their meanings are not grammatical but lexical, with associations outside language. They are lexemes. A lexeme is a minimal unit that can take part in referring or predicating. All the lexemes of a language constitute the lexicon of the language, and all the lexemes that you know make up your personal lexicon. The term ‘lexeme’ was proposed by Lyons (1977:18–25) to avoid complexities associated with the vague word ‘word.’

(a) go, going, went, gone
(b) put up with, kick the bucket, dog in the manger

How many words are there in group (a)? Four or one? There are four forms and the forms have four different meanings, but they have a shared meaning, which is lexical, and other meanings of a grammatical nature added to the lexical meaning. We say that these four forms constitute one lexeme which, for convenience we designate as go. Group (b) presents a different sort of problem. The expression put up with combines the forms of put and up and with, but its meaning is not the combination of their separate meanings. Therefore put up with, in the sense of ‘endure,’ ‘tolerate,’ is a single lexeme. The same must be true of kick the bucket meaning ‘die’ and dog in the manger when it refers to a person who will not let others share what he has, even though he does not use it himself.

First, there is the relation of the lexeme with other lexemes with which it occurs in the same phrases or sentences, in the way that arbitrary can co-occur with judge, happy with child or with accident, sit with chair, read with book or newspaper. These are syntagmatic relations, the mutual association of two or more words in a sequence (not necessarily right next to one another) so that the meaning of each is affected by the other(s) and together their meanings contribute to the meaning of the larger unit, the phrase or sentence. Another kind of relation is contrastive. Instead of saying The judge was arbitrary, for instance, we can say The judge was cautious or careless, or busy or irritable, and so on with numerous other possible descriptors. This is a paradigmatic relation, a relation of choice. We choose from among a number of possible words that can fill the same blank: the words may be similar in meaning or have little in common but each is different from the others. Since we are used to a writing system that goes from left to right, we may think of syntagmatic relations as horizontal and paradigmatic relations as vertical. A compound expression, such as book and newspaper, cautious but arbitrary, read or write puts two lexemes that are paradigmatically related into a syntagmatic relationship.

As children, we learn vocabulary first through specific associations with specific things, actions, and characteristics (reference) and as we learn to recognize different instances of the ‘same’ thing, the ‘same’ event, and so on, we generalize (denotation). Slowly we learn from other members of our speech community and from our personal experiences what associations are favorable and which are not (connotation). And we acquire an implicit knowledge of how lexemes are associated with other lexemes (sense relations). Our implicit knowledge of syntagmatic relations facilitates our perception and identification of what we hear and read, enabling us to correct automatically what we hear and see, or what we think we hear and see, when correction is needed..

Meaning is more than denotation and connotation. What a word means depends in part on its associations with other words, the relational aspect. Lexemes do not merely ‘have’ meanings; they contribute meanings to the utterances in which they occur, and what meanings they contribute depends on what other lexemes they are associated with in these utterances. The meaning that a lexeme has because of these relationships is the sense of that lexeme. Part of this relationship is seen in the way words do, or do not, go together meaningfully. It makes sense to say John walked and it makes sense to say An hour elapsed. It doesn’t make sense to say John elapsed or An hour walked. Part of the meaning of elapse is that it goes with hour, second, minute, day but not with John, and part of the meaning of hour, second and so forth is that these words can co-occur with elapse. Part of the relationship is seen in the way word meanings vary with context. A library is a collection of books (Professor Jones has a rather large library) and is also a building that houses a collection of books (The library is at the corner of Wilson and Adams Streets). A number of English verbs can be used in two different ways —different grammatical association—and then have slightly different meanings.

Here what happened to the window is the same, but in the first sentence broke is equivalent to ‘became broken’ and in the second it is equivalent to ‘caused to be broken.’ (More about this in the next chapter.) Adjectives, too, can have different senses. If you come across some object which you have never seen before, and you wonder about its origin and its purpose, we can say that you are curious about it. But we can also call the object a curious kind of thing. The same term is used for your subjective feelings and for the supposedly objective properties of this item—a curious person, a curious object. A judge makes decisions; if he is guided by personal whim or choice, the judge is arbitrary (dictionary definition: ‘inclined to make decisions based on personal whim’) but we also say that the decision is arbitrary (dictionary definition: ‘based on personal choice rather than reason’). A lexeme does not merely ‘have’ meaning; it contributes to the meaning of a larger unit, a phrase or sentence. Take these phrases with the adjective happy.

When happy combines with a word that has the feature [human], like child and family in the first line, it is roughly equivalent to ‘who enjoy(s) happiness’—a happy child is a child who has or enjoys happiness. In combination with words that have the feature [event] such as accident and experience, its contribution is roughly ‘that produces happiness.’ In combination with words that have the feature [discourse]—story, report—its meaning is roughly ‘containing a happy event or events.’ Each of these words has a range of meanings; each meaning is determined by its linguistic context, just as the meaning of door on any specific occasion is determined by the physical context in which it occurs. The meaning of a lexeme is, in part, its relation to other lexemes of the language. Each lexeme is linked in some way to numerous other lexemes of the language. We can notice two kinds of linkage, especially.

The word dog has a certain denotation, the possibility of entering into numerous referring expressions such as the underlined expressions in the following.
1 This dog is a Dalmatian.
2 My children have just acquired a dog.
3 Several dogs were fighting over a bone.

But how do you feel about dogs? How does a particular society value dogs? Hjelmsle (1971:109–10) pointed out that among the Eskimos a dog is an animal that is used for pulling a sled, the Parsees regard dogs as nearly sacred, Hindus consider them a great pest and in Western Europe and America some members of the species still perform the original chores of hunting and guarding while others are merely ‘pets.’ Hjelmslev might have added that in certain societies the flesh of dogs is part of the human diet and in other societies it is not. The meaning of dog includes the attitudes of a society and of individuals, the pragmatic aspect. It would be wrong to think that a purely biological definition of the lexeme dog is a sufficient account of its meaning. Part of its meaning is its connotation, the affective or emotional associations it elicits, which clearly need not be the same for all people who know and use the word. A denotation identifies the central aspect of word meaning, which everybody generally agrees about. Connotation refers to the personal aspect of meaning, the emotional associations that the word arouses. Connotations vary according to the experience of individuals but, because people do have common experiences, some words have shared connotations.

Languages provide means of expressing different attitudes. The referring expressions that violin and that fiddle can have the same referent—can refer to the same object on a particular occasion—but they do not have the same meaning. They differ in connotation. Violin is the usual term, the neutral one; fiddle-is used for humor or to express affection or lack of esteem. Somewhat similar relations are seen with automobile and car, building and edifice, fire and conflagration and other sets, the members of which have, or can have, the same denotation but differ in the situations in which they are used and therefore have differences in the degree of formality, the style or ‘flavor’—the connotation. (We also need to note here that car, building, and fire have larger denotations than automobile, edifice and conflagration respectively.) The expression of attitudes can be quite subtle. We choose to use one word rather than another. We might, for example, say that Linda is thin, or slender, or svelte, or skinny.

In every language there are words like tree and run and red which seem to have an obvious relation to objects and events and descriptions of things in the world around us. Children learning their native language first learn words in association with observable items and situations and events. This simple fact can give rise to an overly simple idea about what ‘meaning’ is. We are likely to think that a language consists of a large number of words and each of these words has a direct correlation with something outside of language, which is its meaning. And since, if we communicate with one another through language, it must be that we all have the same ‘idea’ or ‘concept’ associated with each word. The best known elaboration of this view was made by Ogden and Richards (1923), who developed a mentalistic theory about meaning, an attempt to explain meaning in terms of what is in people’s minds.

Ogden and Richards called the bond between word and concept an ‘association,’ the bond between concept and object ‘reference,’ and the bond between object and word ‘meaning.’ When we hear or read a word, we often form a mental picture of what the word represents, and so we are apt to equate ‘concept’ with a mental picture. To be sure, it is easier to form a mental picture for some words—DOOR and DOG, for example—than for others such as ORDINARY or PROBLEM or PRETEND. But the idea of a mental picture is misleading. What mental image do you form for DOOR? A revolving door? A folding door? A sliding door, moving horizontally? An overhead door which moves vertically? A door turning on hinges? Is it in a wall, or on a cabinet, or part of a car? Is your image associated with DOG that of a St Bernard or a Pekingese, a mongrel or an Irish Setter? You can picture all of these in sequence but not simultaneously. Clearly the meaning of door or dog is more than what is included in a single image, and your knowledge of these words is much more than the ability to relate them to single objects. You can use these words successfully in a large number of situations because you have the knowledge that makes this possible.

Just as we distinguished between ‘utterance’ and ‘sentence,’ we need to draw a distinction betweeen reference and denotation. Reference is the relation between a language expression such as this door, both doors, the dog, another dog and whatever the expression pertains to in a particular situation of language use, including what a speaker may imagine. Denotation is the potential of a word like door or dog to enter into such language expressions. Reference is the way speakers and hearers use an expression successfully; denotation is the knowledge they have that makes their use successful.

The trouble with a mentalistic theory of meaning is, first, that not all words can be associated with mental images and some words have a range of meaning greater than any single association. The bigger problem with a mentalistic theory is that we have no access to other people’s minds. How can we ever know that we all have the same mental images? If semantics is a science, it cannot operate scientifically by starting with things that are not observable and not comparable. Furthermore, words are not the only semantic units. Meanings are expressed by units that may be smaller than words—morphemes—and meanings are expressed in units—sentences—that are larger than words. The sentences 'The dog bit a man' and 'The man bit a dog', to use a trite example, contain the same words but they don’t express the same meanings.

Furthermore, meaning is more than denotation. People not only talk and write to describe things and events and characteristics; they also express their opinions, favorable and unfavorable. Language furnishes the means for expressing a wide range of attitudes; this aspect of meaning is called connotation. Another aspect is sense relations: the meaning of any expression varies with context, what other expressions it occurs with and what expressions it contrasts with.

Combinations of paralanguage and gestures can communicate something about the mood of the speaker—anger, boredom, nervousness, elation, for instance—and actors work hard to achieve such effects in interpreting the characters they play. But each actor strives to do so differently. We are not impressed by a budding starlet who uses the same inventory of mannerisms as a seasoned actress. Other facets of appearance—clothing, hair style, jewelry, cosmetics, facial hair and what is done with it—have an effect on others, intentional or not. The distance between interlocutors and whether they touch each other or not depends on tacit standards that each of us learns from the culture in which we grow up. Whether we sit on the floor or on chairs, cross our legs at the ankles, over the knees, or not at all—these ‘say’ something about a person’s cultural background but they do not communicate semantically.

Just as we learn a language early in life and largely take it for granted, so we also learn these more peripheral elements of communication and, unless we move into a different society, we assume that they are ‘just natural.’ But gestures, interpersonal distances, the ways the voice is used can be quite different in different societies and thus any of these can have different effects on people of different backgrounds. What is meaningless or mild in one culture may be rude, obscene, or otherwise over-effective in another. Thus a face-to-face communication event contains linguistic and non-linguistic elements like these:

Linguistic:
vocal and verbal—words put together to form utterances (representing sentences)
vocal and non-verbal—prosody, the intonation and accenting with which utterances are spoken

Non-linguistic:
vocal—paralanguage, the “tone of voice”
non-vocal—distances maintained; appearance; gestures; silence.

There are visible signs, gestures, which have a standard, shared meaning, and there are elements of appearance—‘body language’— which possibly create an effect on the observer and therefore on the interpretation of a spoken message. The former, the visible signs, have the capacity to communicate in much the way a word communicates; the latter could only be said to communicate in a secondary sense.

Consider these visual signs:

nodding the head in response to an utterance
crossing one’s fingers
pretending to yawn, with finger tips in front of mouth
holding up a thumb from a closed fist
pinching one’s nostrils closed with thumb and forefinger

Like the audible signs mentioned above, these gestures have recognized, though somewhat vague, shared meanings. The first suggests agreement or affirmation; the second, a hope for success when circumstances are uncertain; the third, boredom; the fourth, determination to make one’s cause successful; the fifth, disgust or displeasure with something (‘It stinks’). Other physical postures and movements—for instance, gestures with hand or whole body, such as pounding on a table with a finger or a fist, and facial gestures like pursing one’s lips, arching the eyebrows, opening the eyes wide, squinting, or fluttering the eyelids— are not conventional signals and do not have meanings in themselves; they may lead an observer to form some particular impression of the speaker, which in turn may have an effect on how the hearer interprets, but they are not in themselves semantic. It is possible that a wink or a broad grin on the speaker’s face may communicate to the addressee a visual message that the verbal message is meant facetiously, not to be taken seriously, but there is never any certainty that such is the intended message.

There are other ways of using the voice, as part of the spoken utterance, which cannot be considered either signs or part of language. These include laughing, giggling, crying (which need not accompany an utterance); whisper, falsetto, a quavering or ‘breaking’ voice; and other elements that are vocal but not verbal: the relative loudness or softness of the voice, high or low pitch, the modulations of pitch from a near monotone to an exaggerated rising and falling, a nasal quality, a rasping sound, the tempo of speech—the speed at which a whole utterance is delivered or the relative timing of syllables, ranging from clipping to drawling. These ways of using the voice cannot be considered signs—they do not signify—but they may be expressive, communicative in a secondary sense. Speakers may want to create a particular effect with their ways of using the voice; listeners may interpret what they hear in particular ways because of vocal features; but if intentions and interpretations coincide, the coincidence is fortuitous. All these ways of using the voice are together called paralanguage. The failure to use language—silence—at a particular juncture can likewise be expressive.

Aside from what we say to one another through the verbal content and the prosody of spoken utterances, we can, in face-to-face communication, transmit less systematic messages to one another by means of audible and visible signs that are not part of language. In addition, our voice or our appearance may have an effect on other participants in a conversation and therefore have an effect on the way our verbal messages are interpreted. Consider first the standardized noises we make, which are written this way (not very accurately):

ps-st sh-sh huh? unh-huh m-m-m b-r-r tsk-tsk

These count as signs. On the whole they are known to all (or at least large portions) of a language community and indeed may be used by speakers of several different languages. In general, the maker of the sign and those who hear it attach the same meaning to that sign; communication occurs. These seven audible signs indicate, respectively, a request for attention; a call for silence; a request for repetition or clarification; a signal of agreement; an expression of pleasure or enjoyment; an indication of coldness; and an expression of shame or shock.