Semantics

semantics

Semantics is the study of meaning in language. It is a sub-discipline of the science of semiotics which, roughly speaking, is the study of meaning in general. There is, of course, a distinction between meaning and meaning expressed via language.
For example, which of the following represents meaning expressed through language? Click on the table when you have an answer.

semantic

Whatever our personal views about whether dogs and cats understand (or use) language or whether true language is a phenomenon confined to humans, in what follows, we will be discussing spoken or written language not the many other ways in which meaning can be expressed. We will not, for example, be dealing with body language but what follows does apply to the various sign languages used by people around the world.
There is a guide to language evolution, linked below, which contains some consideration of whether language is confined to humans or may be said to be used by other animals.

Words, lexemes and lemmas (or lemmata)

The guide to morphology, linked at the end, takes trouble to distinguish between a word and a lexeme and explains why many tests for identifying what a word actually is fail. You should go to that guide for more detail.

Here we will simply note that the term word applies in this guide to any meaningful item in English which can be made by applying the normal rules of morphology. By this definition, the following are lexemes even when they contain more than one word:
The European Union, Mars, washing machine, happiness, undoability, redefinability, colourlessness, go on, look forward (to)
Not all of these will be found in a standard dictionary, of course, but that doesn't stop them from carrying a single significance and functioning as words in English.

Strictly speaking, a lexeme is often defined as a set of related forms of the same word so, e.g., speak, speaking, spoken, spoke are all various forms of the single lexeme speak.

A lemma is what dictionaries use as headwords from which other forms are derived. You would not, for example, expect all the forms of the lexeme speak to have a separate entry in a dictionary. You would simply go to the lemma and from there discover, perhaps, that the past tense of the verb is irregular.

What does mean mean?

There are three fundamental forms of meaning.

Sense
This refers to a word's general significance.
For example, we can probably agree on the sense of the word coin. The word, as a noun, means a small metal unit of currency. This is the word's sense or its denotation.
One word can have a range of senses in which it is used. For example, coin can be a verb (with a number of meanings) and the word bank can refer to a business that deals with money, the side of a river, a verb meaning rely and a verb meaning deposit money. All of these are senses (or denotations) of the word. For more, see the guide to polysemy, linked in the list of related guides at the end.
Reference
This refers to the actual thing I mean (or the action etc. but for simplicity's sake, we'll focus on nouns). In other words, it is this instance of the word's use.
For example, if I say, The coin on the book, I know (and so do you) that I am referring not to coins in a general sense but to a particular one I have in mind. However, just by looking at the words, we have no idea what exactly I am referring to and in a different setting the referent (thing referred to) may be different.
Connotation
This refers to a second level of meaning above denotation and is often personally or culturally determined. For example, the words quack and doctor carry the same sense (medical practitioner) but mean very different things. The same can be said of a whole range of near synonyms such as youth-teen, child-brat, newspaper-rag, speech-sermon, dirt-filth, police officer-copper-cop and so on.
The distinction between denotational and connotational meaning goes under a variety of guises in the literature including extension meaning vs. intention meaning, definitional information vs. contextual information and core meaning vs. encyclopaedic knowledge.
The last of these is the route taken below.

Fuzziness

Before we leave a discussion of what words mean we need to deal with the borderlines of meaning. For example, we may know what we understand as the distinction between related verbs such as climb and scramble but identifying exactly at which point a scramble becomes a climb and vice versa is quite difficult and will probably vary between speakers of the language.We may equally be comfortable distinguishing, for example between these related terms:
    claw vs. talon
    tooth vs. fang
    tree vs. bush
    hill vs. mountain
    fire vs. blaze
    red vs. pink
    ill vs. poorly
    walk vs. run
    expect vs. anticipate
and so on.
However, precisely where we draw the line is clearly debatable.
Partly, we are discussing here something called troponymy. Because, for example, walk and run (as well as amble, stroll, wander, gallop, trot and so on) are ways of moving on legs. In other words, they have a relationship of a special kind of hyponymy.
The guide to lexical relationships, linked at the end, has a little more on hyponymy and troponymy.

Often, we decide where the borders between meanings are on the basis of:

contrasting the meanings with other related word so, for example, we may distinguish between red and pink by deciding whether it is a colour more closely approximating to rose or crimson.
the context in which it occurs, preferring fang for the tooth of a large carnivore and tooth for that of a smaller animal or person.
comparison with a prototype so we decide where the line is drawn between a mountain and a hill by simply comparing the size, characteristics and shape of what we see to a well-known prototype of a mountain such as Mont Blanc.

Meaning is, therefore, partly dependent on sense relations or categories of related meaning. This is the meat and drink of semantics.

Some words, of course, do not exhibit any obvious fuzziness and they are mostly those which are carefully defined in certain fields or are associated with particular objects. Examples of such words are:
    aspirin
    carbohydrate
    planet
    triangle
    logarithm
    DVD
    keyboard
etc.
However, among experts even more than among lay people some terms may be the object of heated debates and the precise point at which a place such as Pluto is downgraded to that of dwarf planet, for example, was only settled in 2006 and many non-astronomers will still refer to it as the ninth planet of the solar system.

Two distinctions of meaning

Core meaning

Allied to the distinction between denotation and connotation is the concept, variously described and classified, of some sort of core or basic meaning of a word and its meaning for whomever reads or hears it.
Here we need to distinguish, as Aristotle did some time ago, between a word's essential meaning and its reference meaning. Aristotle referred to essence: the essential qualities of a thing without which it loses its identity.

For example, if I hear:
Look. There's a dog in the garden
I will not randomly peer through the window trying to understand what is meant but will immediately be alert to spot a four-legged, furry object which matches my mental picture of dogness. I have probably never seen the dog in question but I will immediately recognise it because I have a set of mental constructs which applies only to dogs. I will, therefore, not waste my time looking at trees, ponds, unicorns or elephants because that is not what I am primed to see.

What I am looking for is something which contains the semantic components of the word dog. I.e., it is animate, furry, four legged and of a certain size (a somewhat variable component).

How we do this is slightly mysterious but to try an explanation, here's an example. Given a resource like this,
cup
if someone says:
Please give me that cup
it will not be difficult for most people to recognise the object which is being spoken about from a range of possible containers and other objects in sight even if the cup in question has never been seen before. To do that, people need a mental prototype of a cup which distinguishes it from other sorts of objects. And to do that, we assign characteristics (either positively or negatively) which we use to distinguish cups from other objects. For example, we may select the following as part of the essence or core meaning of the term cup:

Quality for cups	Yes	No	Maybe
designed to hold liquids
made of china or porcelain
made of glass
for hot liquids
roughly cylindrical
plastic
of any colour
transparent
with a handle
holding less than .5 litres
with a saucer

However, if someone says,
Please give me that glass
we will recognise it by a different set of prototypical features:

Quality for glasses	Yes	No	Maybe
designed to hold liquids
made of china or porcelain
made of glass
for hot liquids
roughly cylindrical
plastic
of any colour
transparent
with a handle
holding less than .5 litres
with a saucer

When it comes to more closely related concepts, speakers of a language will often disagree about what to call something.
For example:
Would you like a cup or a mug of coffee?
Did she stroll or amble here?
In the first of these examples, we have two hyponyms which bear some relationship to a hypernym such as food container.
In the second of those two examples, we encounter a relationship between words called troponymy. That is to say, both the words amble and stroll are ways of further defining the concept of walk. However, exactly how the nature of strolling differs from that of ambling, wandering, sauntering, rambling, mooching and so on is quite obscure and native speakers of a language are unlikely to agree completely about where to draw the lines.
For a little more on these two concepts, see the guide to lexical relationships, linked in the list at the end.

What we are discussing here is a word's semantic space which concerns the prototypical features of a word which determine the limits of its use. Children, learning their first language, often make errors in this regard, calling, for example, all four-legged creatures horses or all adults apart from mother and father, uncle and so on. Quite rapidly, children learn to identify the prototypical features of lexemes and limit their use to within conventional, culturally determined boundaries. This goes on through life and most native speakers are frequently faced with uses of words outside the fields in which they have previously been encountered.
Unfortunately, semantic space varies across languages and adults already have a set of prototypical features which they employ frequently to distinguish, for example between a cup and a glass (using the ideas we discussed above such as with or without a handle, transparent or opaque, porcelain or not etc.).
For example, in German, the word vegetable does not include the term potato and in Swedish, the word wood for the material and for a small group of trees is not the same although it is, for example, in French.
Applying the conventional prototypical features that one has acquired with one's first language to a second is, therefore, often perilous.
Moreover, determining the boundaries of closely connected words such as railing and fence, shade and shadow, job and career, position and post and so on is difficult and requires repeated exposure to the lexemes in context. Even native speakers may disagree about the determining features which distinguish a tree from a shrub or a leaflet from a tract, for example, and the distinguishing features of such terms are very unlikely to be parallelled in the learners' first language(s). This is not a phenomenon confined to nouns and the distinctions between stroll, saunter and wander and chilly, cool and fresh are not immediately apparent from context.
The core meanings of railing and fence are virtually identical (involving concepts of a barrier delineating spaces) so only secondary features (such as the materials from which it is made, the purposes to which it is put and its design), which may not always be present, can be used to disambiguate. Hunting the internet for images of either will often throw up the same images in different categories.

To further complicate matters, individuals will vary in the characteristics they see as essential and those which are optional. Between cultures, variations will be even more obvious with German speakers, for example, often excluding potato from the general category of vegetable.
It is also the case that in German, any container for hot drinks which does not have a saucer may be described as a mug [Becher] but that definition will not function too well in English for a small cup with a handle.
Semantic space can differ very dramatically across languages.

The selection will be made on the basis of choosing the object which ticks most of the boxes and excluding those with forbidden characteristics.
What we do, in the jargon, is to carry out a semantic or componential analysis of, in our example, the terms cup and glass. When we do that, if we are faced with two similar objects (or actions, adjectives or whatever), we may use the optional characteristics above to make our choice. So, for example, we may exclude a container made of glass which has no handle in favour of a glass one with a handle when we are searching for the cup in question.

Alert readers will, however, have noticed a snag with an analysis like this: the distinctions are fuzzy at the edges. Some cups are made of glass, some glasses are made of plastic and so on. Aitchison, 1987, suggests that in addition to the prototypical features of the object, people also apply those characteristics with which they are most familiar so that, for example, speakers from certain cultural backgrounds may insist that all cups have handles and come with saucers but for others that distinction may not apply.

Encyclopaedic knowledge

In addition to all this, there is a distinction between the core meaning (what we have discussed up to now) and what is sometimes called encyclopaedic knowledge or extension meaning which refers to what we know about the word from our general knowledge of the world.
Schmitt, 2000: 27, provides the example of the term bachelor, the semantic analysis of which would be +human, +male, +adult and -married or, in other words, an adult, unmarried, male human. The issue here is that while the semantic definition holds true, there's a lot we know about bachelors which is not included. As Schmitt points out, this knowledge includes the fact that bachelors are often young, date women and have exciting lifestyles and none of that information can be captured by a purely semantic analysis because it forms part of the schema that native speakers have, in certain cultures, concerning the state of bachelorhood. As Schmitt points out, the core definition of bachelor would include a divorced, middle-aged man with several children or a male who is unmarried, but living with his partner but our encyclopaedic knowledge might act to exclude such people from the definition.
The issue is twofold:

Core semantic features of a word will be limited and can be taught by exemplification and helping learners to notice where the word starts and stops and how the concept is distinguished from similar concepts (as in the difference between cup and glass discussed above).
Encyclopaedic knowledge is open ended and varies between cultures and individuals within cultures. The schema which for each person is activated by encountering a word such as farmer, for example, will be variable and some individuals (and some cultures) may exclude, for example, someone who has a smallholding on which she works part time or a large landowner who lives in the city and rarely visits a farm.

Encyclopaedic knowledge, incidentally, is part of the ability to recognise impossible collocation. For example, if we encounter words such as lacy, silky, smooth, velvety etc., our knowledge of the world tells us that they cannot be applied to any noun in the whole English lexicon (of which there are over 85,000). One cannot have, for example,
a velvety rhinoceros
simply because the animal doesn't come that way.
Equally, of course, we know that verbs such as assert or enjoy cannot have inanimate subjects because tables and houses etc. do not do these things. Poetically, we may allow many kinds of odd combinations for effect, of course.
For more in this area of what is called suppositional or pre-suppositional meaning. See the guide to collocation, linked below.

Kinship terms and the packaging of information

Core meaning and encyclopaedic knowledge is well exemplified when considering kinship terms.
For example, in English the terms are relatively simple. We can, for example, say:
    He is my grandfather
    That is my niece
    That is her aunt
and so on.
From our encyclopaedic knowledge, we know that grandfather is a male parent of someone's parents and we can define other kinship terms in the same way: female offspring of a brother or sister, sister of a parent etc.
However, when we ask what information is encoded in the terms, the situation is somewhat rough and ready. When someone says, for example:
    That is my nephew
we know only that the person in question is one generation removed (down the family tree) and is male. We do not know:

whether the person is related to the speaker by blood or marriage
whether the person is older or younger than the speaker
whether the person is the son of a sister or a brother
whether the person is the son of an older or younger sibling
whether the person is a boy or a grown man

Equally, someone may say:
That is my uncle
and again, the details of the person and the precise family relationships that exists remain unclear. In many English-speaking societies, the man in question may not even be related in any way to the speaker but the term uncle simply denotes an older male friend of the speaker's parents. Even less informative in this respect is:
That is my cousin
in which we do not even have access to the information about the person's sex.

This is because languages encode the meanings that are socially important and English-speaking culture, presumably, has seen no need historically for the relationship details to be made clear.
Other languages will encode different data and make any or all of the relationships (blood vs. marriage, older or younger, male or female, connection to other family members and so on) clear in the term that is chosen. Some, such as Sudanese, will be much more informative and have words for cousin which distinguish between
    father's brother's children
    father's sister's children
    mother's sister's children
    mother's brother's children
Tamil, an ancient South Indian language, too, has specific kinship terms rendered by only two lexemes in English:
aunt: chitthi (mother's younger sister), periyamma (mother's elder sister), mami (uncle’s wife)
uncle: chithappa (father's younger brother), periyappa (father's elder brother).
Other languages may be even less informative than English. Some Hawaiian languages, for example, do not distinguish between sibling and cousin at all.

Kinship terms are just one of the areas in which languages choose to encode what is culturally important to them. Some languages will have single words to describe events and objects which need a whole sentence in other languages to explain. For example, some languages, will have separate verbs for go which distinguish whether the verb means go uphill, downhill, on foot or on horseback and others will have separate nouns to describe the same animals at different stages of their development and usefulness.

Schemata

Schemata (singular schema) are sometimes called frames or scripts.
The significance is that they activate our encyclopaedic knowledge through context and that has very obvious implications for the classroom and the teaching of lexical meaning.

For example, if one considers a simple sentence such as
    The glasses were broken
it is likely that, without any other context, most people will have their schema concerning glass drinking vessels activated. However, given the context of being able to read a label and the sentence
    My glasses were broken
it is likely that the schema which will now be activated concerns spectacles.
Given, as another example, the sentence:
    The party was united
vs.
    The party was noisy
it is likely that wholly different schemata will be activated simply by the difference in the adjective used to describe party.

Because encyclopaedic knowledge, on which schemata work, is variable from individual to individual and the core meanings of words are not easily translated across languages or between cultures, it becomes very important for teachers to provide adequate data to allow learners to notice (or be told) what is and is not included in a concept represented by a lexeme.
At the outset, it may be enough to define, for example, the verb harvest by reference to well-known examples such as gathering wheat, rice, vegetables etc. but that may not be enough to distinguish the word from the idea of pick (as in flowers, blackberries or other wild fruits) and it almost certainly won't be enough to allow learners to understand that harvesting data is possible but picking data has a wholly different meaning.
Word meaning is learned incrementally with the learners' understanding of limitations and inclusions in the meaning of a lexeme being refined as more data become available.
It is, therefore, as important to help learners understand what is excluded from a word means as it is to convey what is included.
For example, the comparatively simple idea conveyed by the word steps (which might be explained as a series of flat surfaces on which to climb up or down) needs to be refined by excluding carpeted wooden examples inside houses but which will include metal external fire escapes while probably excluding internal stone fire escapes.

The concept of register (i.e., the field, in Halliday's 1978 analysis) in which a word occurs is critical in this area.
If, for example, we know that we are listening to or participating in a conversation about a parliamentary debate, we will almost certainly understand the word minister differently from how we would understand it if the conversation concerns church employees.

Little by little, we are now moving from word and sentence meaning to utterance meaning and here we trespass on the territory of pragmatics. So be it.
However, before we move on, here's a summary of the story so far:

summary

Compositionality

You will readily see that knowing the senses of the individual lexemes in a clause or phrase usually allows you access to the overall meaning. For example, knowing the meaning and/or function of
    completely, problem, the, is, clear, me, to
You can arrive at the meaning of the sentence:
    The problem is completely clear to me
and
    To me, the problem is completely clear
and even
    Completely clear to me is the problem
That is what is meant by compositionality.

However, there are many instances of language when knowing the sense of the constituent parts of a string of words will not allow you access to the overall meaning. You may, for example, know the meanings of
    family, he, black, of, sheep, the, is
but that will not allow you to understand
    He is the black sheep of the family
This is what is meant by non-compositionality. Idioms and phrasal verbs are familiar examples of non-compositionality but it is not an either-or distinction. Some idiomatic expressions can be understood with access to the concept of metaphor and such items vary in the level of transparency. So, for example
    He put the meeting off until Wednesday
is reasonably transparent with an understanding that the particle off frequently carries the meaning of away from as in, e.g.:
    He took the plate off the table
and application of the meaning of the prepositional phrase helps, too, of course. However,
    He put up with her
is almost completely obscure unless you are aware of the meaning of the multi-word verb put up with (=tolerate).
When it comes to idioms, some are often used quite literally and some can only be understood as metaphors. For example:
    She won by a hair's breadth
is an idiom in English which is fixed so you can't change hair to strand or breadth to width and retain the same meaning. It is, however, reasonably transparent in suggesting a narrow margin of victory.
On the other hand, something like:
    He made it by the skin of his teeth
is less transparent insofar as teeth do not have skin for one thing. Indeed,
    That doesn't cut the mustard
is almost wholly obscure and must be understood as a single idea (not good enough).
There are, in other words, various levels of non-compositionality.
For more about this, see the guide to idiomaticity, linked at the end, which also covers the notion of fixedness.

Use and Usage
Signification and Value

The distinction here is between sentence meaning and utterance meaning and lies at the heart of communicative language teaching. The distinction can be summarised:

Usage means focusing on the meaning attached to something as an instance of language isolated from context. It is its signification (what it means).
Use refers to the meaning of something when used for communicative purpose. This is its value (what it does).
For example:
A: Why don't you see a doctor if you are feeling so ill?
B: Mount Everest is very high and Mercury is the nearest planet to the sun.
B's statement has significance (we know what is meant) but no value (it communicates nothing useful).
(Widdowson, 1978)

Utterance meaning can be quite obscure and include attempts at satire, irony and so on. Two obvious examples are hyperbole and litotes.
Hyperbole is deliberate exaggeration for effect in, e.g.
    There were millions of people at the party
    It weighed a ton
when in neither case do we literally mean millions or a ton.
Litotes is the polar opposite way to get the same effect in, e.g.
    You'll find the centre of London a bit busy on Monday mornings
    I drove to Greece and back from the UK so I've done a few miles
when in neither case is the downtoner (a bit and a few) meant literally. The centre of London is actually extremely busy at this time and there are more than a few miles between Greece and the UK.

Communicative force

Another way to distinguish between sentence and utterance meaning is to consider communicative force.
If for example, someone says:
The food is on the table
there are three possible communicative forces in play:

Locutionary force: The 'basic' or sentence meaning of what is said.
In our example, this would correspond to the meaning that the food is in the place I have specified. No more, no less.
Another way to say this is that you have understood the propositional content of the utterance.
Illocutionary force: The meaning intended or the meaning perceived – the utterance meaning.
In our example, the statement could mean, Please sit down and eat or a number of possible meanings such as It is getting cold.
Perlocutionary force: This refers to the fact that an utterance like this may actually produce a reaction in the hearer. If, in this example, the hearers immediately come to the table, sit and begin to eat then the perlocutionary force of the statement has been demonstrated. In some instances, the simple utterance results in the effect. For example, I now pronounce you man and wife may result in the marriage with no intervening linguistic form.

The distinction between sentence meaning and utterance meaning is important for language teachers, of course. The distinction is also used to divide semantics proper (i.e., the study of sentence meaning) from pragmatics (i.e., the study of utterance meaning) but the distinction is neither clear cut nor universally accepted.

Gricean maxims

Herbert Paul Grice's work is relevant in the area and to the teaching of language using any communicative approach. It bears some analysis here although it is also relevant to other areas in this site.
If one starts from the premise that utterances have some kind of illocutionary force, i.e., an intended meaning which the hearer understands, we need to know something about the principles at work which allow the meaning to be understood.
Grice's work is not, of course, without its critics so the following is, at best, the theory, at worst, simply a hypothesis. There are four main maxims which determine how interactions proceed. Overriding all four is the Cooperative Principle. In Grice's words, this is:

Make your contribution such as it is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged.
(Grice 1989: 26)

The four maxims we follow to achieve this are:

THE MAXIM OF QUALITY:
1. Don't say what you believe to be false.
2. Don't say things for which you have no evidence.
THE MAXIM OF QUANTITY:
1. Be informative enough.
2. Don't over-inform.
THE MAXIM OF RELATION:
1. Be relevant.
THE MAXIM OF MANNER:
1. Avoid obscurity.
2. Avoid ambiguity.
3. Be brief.
4. Be orderly.

The issue here is not that we always follow these maxims but that we are subconsciously aware of them. When any are broken, we are immediately alert to the fact that something other than the sentence meaning is intended.

Example 1:: If in answer to
    What time's dinner?
the response is
    But you promised!
Then, on the face of things, the maxim of relevance has been breached because the response is not obviously relevant to the question. However, if the hearer is alert, she/he may presume that the response means something like
    I'm not cooking because you promised to do it so you tell me when it'll be ready.
Example 2:: If the response to:
    I'll get myself a beer
is
    There's a shop down the road, just by the post-office which sells all kinds of drinks, including beer
Then, on the face of it, the maxim of quantity has been breached because this is just too much information. However, the illocutionary force to which the hearer may be alerted could be
    You should go and buy your own beer and not keeping helping yourself to mine.
Example 3:: If the response to:
    Lend me $200
is
    Pigs might fly
Then it seems that both the maxim of quality and the maxim of relevance have been broken but the response may simply mean
    It is about as likely that I will lend you $200 as pigs flying (i.e., not particularly).
Example 4:: Jokes are rich sources of maxim breaking precisely because they rely so often on unexpected responses to utterances. For example:
    I rang the bell of a bed-and breakfast place and a lady appeared at a window.
    "What do you want?" she asked.
    "I want to stay here," I replied.
    "Well, stay there then," she said and closed the window.
which breaks the maxim of quality at least and probably the maxim of relevance by deliberately misunderstanding what is meant by here.

There's a good deal more on Gricean maxims and their classroom implications in the guide to pragmatics on this site.

Jokes and breaking maxims

Here's a joke for you to try. What maxims are being broken?
Click here when you have an answer.

There is a woman sitting on a park bench and a large dog lying on the ground in front of the bench. A man comes along and sits down on the bench.
MAN: Does your dog bite?
WOMAN: No.
(The man reaches down to stroke the dog which bites him.)
MAN: Ouch! Hey! I thought you said your dog doesn't bite!
WOMAN: He doesn't. That's not my dog.

The woman breaks the maxim of quantity because it is clear that the man is referring to the dog in sight. He has no way of knowing that she has a different dog and it is natural to assume that the dog is hers. She has been economical with the truth and provided too little information.
She is also breaking the maxim of relevance by referring to another dog which is not present.
The whole joke (if it can be called that) is dependent on breaking the cooperative principle.

Other guides

The whole area of semantics (and pragmatics) underlies the theory of language which informs almost any teaching approach. The following guides become clearer in the light of semantic theory and reference to this guide.

Related guides:
suasion	focusing on some key functions in English more easily understood in relation to utterance meaning
CLT	the guide to Communicative Language Teaching
morphology	focusing on how words are constructed and what their individual parts signify
idiomaticity	focusing on (non-)compositionality and notions of fixedness
polysemy	focusing on multiple word senses and other relationships such as synonymy and metonymy
lexical relationships	focusing on lexical relationships such as hyponymy, troponymy and antonymy
collocation	for more on suppositional meaning
language, thought and culture	for some considerations of whether language determines thought or whether the reverse is true
the evolution of language	a guide which contains consideration of how the arbitrary nature of words and their meanings may have evolved
multi-word verbs	focusing on these verbs and also on notions of transparency and derived meaning of particles
function words	the guide to help with recognising different levels of meaning
deixis	although not central to semantic analysis, notions of reference are applicable to this area
pragmatics	this extends many of the considerations of the second half of this guide

If you would like to take an easy test on all of this, click here.

References:
This is a huge area, much researched and written about. Many studies of semantics focus on cross-linguistic comparisons, sometimes of obscure and exotic languages, and are less than helpful for English language teachers. However:
Aitchison, J, 1987, Words in the Mind: An introduction to the mental lexicon, Oxford: Blackwell
Grice, HP, 1989, Studies in the Way of Words, Harvard: Harvard University Press
Halliday, MAK, 1978, Language as a Social Semiotic, London: Edward Arnold
Reimer, N, 2010, Introducing Semantics, Cambridge: Cambridge University Press is an accessible source.
Schmitt, N, 2000, Vocabulary in Language Teaching, Cambridge: Cambridge University Press
Widdowson, HG, 1978, Teaching Language as Communication, London: Oxford University Press is also accessible and to the point.