This post is for you:
If you ever get into debates about words.
If you ever get told how you should or shouldn't use words.
Especially, if you want to tell other people how they should or shouldn't use words.
My first task, however, is to persuade you that you might need some basic instruction about the way that words work. In doing so I face some difficulties.
Your brain is already an expert at words (at least for words in your mother tongue), as you use them all the time, and most of the things I will highlight below are well known to your subconscious. If this weren't the case, you would barely be able to communicate, because your brain needs to apply most of these principles to navigate the millions of ambiguities and complexities that accompany almost any set of words.
You may therefore think that since you use words expertly all the time, you can talk about them expertly. However, these two levels of skill — what your brain can do, and what you can do consciously — are different, and may not be related at all.
I saw this insight first from Hofstadter's Gödel, Escher, Bach: An Eternal Golden Braid (a brilliant, mentally exhausting but mind-expanding book that I keep on plugging from this blog!). In at least one place he points out how your brain can do all kinds of tasks at a mechanical or subconscious level that you cannot do consciously. Take, for example, the task of catching a ball. This requires your brain to do some awesomely powerful image processing and recognition tasks, along with numerical calculus calculations, and probably many other things, and very quickly too. But at a conscious level, given the same tasks (image processing, calculus etc.) you might be completely floored about how to start, or think you know how to do them and fail miserably.
In that example, our subconscious abilities appear to be completely inaccessible to our conscious minds. Only relatively recently have we discovered how to do numerical calculus, with image processing and recognition still being in their infancy, and none of these disciplines was helped by sitting and reflecting on our ability to catch a ball.
When it comes to words, it seems that our subconscious handling of words is quite a bit closer to the conscious level — you can access some of what your subconscious knows about words. This is particularly apparent when learning a foreign language as an adult. However, people still make huge numbers of elementary mistakes when dealing with words. This seems to happen:
When people handle languages foreign to them.
Even in their mother tongue, when they go into the upper, conscious level and argue about the meaning and usage of words.
TL;DR: Your brain is awesome at words, but your conscious mind has a lot to learn, and you need to understand that.
My credentials in writing this post are not what I would like, as I haven't had a huge amount of formal training specifically in linguistics. At school I learnt French and German, and continued with French beyond school including some at University. I'm currently living in Istanbul and 8 months into intensive learning of Turkish, a very different language from the other European languages I've studied. Learning foreign languages doesn't necessarily make you good at understanding linguistics, but it can help make you more aware.
Probably more significant is the language study I've done in relation to working as a Bible teacher. This included learning the basics of Biblical Greek and Hebrew at Bible college, and, vitally, the principles of exegesis. As someone who believes the Bible is God's word, but also a very human book, communicated using the normal rules of human language, I obviously think it is very important to understand language well enough to handle the Bible correctly. And, unlike learning living languages, it is very hard to benefit from most of the automatic understanding of language that our brains can do. This has the unfortunate effect that a lot of Bible teaching can slip into many linguistic fallacies when handling the Greek and Hebrew. The upside is that if you attempt to learn how to handle the ancient Bible texts properly, you necessarily have to learn a lot about linguistics at the conscious level.
A hugely helpful book in this regard is D. A. Carson's Exegetical Fallacies — in fact a lot of this post parallels the first chapter “Word-Study Fallacies”. Also, NT Greek Beyond The Basics by Daniel B. Wallace, has really helped me understand the difference between form and function, syntax and semantics, both in general and in the specifics of Greek, and the vital of role of looking at usage.
I still feel very under qualified to write this post, however, and I'm doing so only because I can't find something similar on the internet that I could point people to (you know, for when someone is wrong on the internet). If you know of something better, or have corrections to this post, please let me know in the comments!
TL;DR: I'm not as qualified as I want to be to write this post, but I can't find a better summary of these things and corrections will be gladly accepted.
The main thing to understand about words, in the context of this post, is this: the meaning of a word is determined by usage. That is, a word means what people intend and understand by it. In fact, it is generally better to talk about the meaning of a specific usage, rather than the meaning of a word.
We can break that down in a number of ways. I'm going to illustrate with fairly uncontroversial examples at first, and leave the more controversial points until a bit later.
Words have multiple meanings
If words are used in multiple ways, then they must have multiple definitions. In practice this is extremely common, which will be very obvious to you after a moment's reflection. Or just crack open a dictionary — it’s often hard to find an entry with just one definition. This is a very basic point, but I still sometimes see arguments that are implicitly assuming that every word only has one meaning.
Words are not defined by their roots
This is a very common genetic fallacy. Don Carson puts it this way:
One of the most enduring of errors, the root fallacy presupposes that every word actually has a meaning bound up with its shape or its components. In this view, meaning is determined by etymology; that is, by the root or roots of a word.
A very closely related idea is that there is a single “root” meaning for every word, which is the “true” meaning of that word, and other meanings are less important or less correct.
While in some cases all current usages can be traced back to a single root, that root is not the “true” meaning of the word, nor does it necessarily determine or even flavour current usages. Etymology is of extremely limited usefulness in determining current meanings of words in living languages.
There are millions of examples of this. My current favourite is “atomic”. Let's analyse the sentence “it was like an atomic explosion” according to this fallacy:
The etymology of “atom” goes back (as far as we can trace it) to the Greek word ἄτομος (atomos) meaning "indivisible", from ἀτέμνω (atemno), from α ("not") and τέμνω (“I cut”). It was used by Greek philosophers as the name of a theoretical particle that was by definition indivisible — the fundamental building block of all matter.
We would therefore conclude that an explosion (and therefore division) of an atom is something that by definition cannot happen. Our sentence must be claiming that “it” was by definition impossible (and therefore presumably didn't happen).
Of course we see this is false. The meaning of the word atom has evolved, and thankfully it’s not that hard to trace. Physicists found a particle that at the time appeared to be indivisible. These particles had different types (carbon, oxygen, hydrogen, etc.), and they could be put together to form molecules, but they resisted attempts to break them down. Using the terminology of the time, they chose to call these species of matter “atoms”. They were premature in their conclusion of indivisibility, as we later found out — the particles are composed of the smaller particles neutrons, protons and electrons — but the name had stuck.
Atoms, and especially the nucleus of atoms, are, however, generally extremely resistant to being split, or being combined, but when we can achieve it (nuclear fission and nuclear fusion respectively), large amounts of energy can be released (depending on the type of atom). “An atomic explosion" is really short for “an atomic bomb explosion” — a bomb that can harness these reactions to release vast amounts of energy. We've arrived at the true meaning of our sentence — “like an atomic explosion” means “extremely powerful or energetic”.
Of course, the original meaning of atom is still alive and well, so it's not a simple case of the word changing from one thing to another. Lisp programmers use atom to refer to an indivisible unit in a data structure. In concurrent programming and database systems, atomicity refers to processes that are indivisible (to outside observers at least) — only the before and after states can be observed.
Ironically, the meaning of the word ‘atomic’, like the particle (mistakenly) named after it, has also been split.
The meanings of words change
This sounds like I'm restating the previous section, but there is a different mistake people can make. In the last section, I addressed “This word used to mean X, so it means X today.”, The corollary, also incorrect, is “this word means X today, so it used to mean that in the past.”
It's here that etymology comes into own to help us out. A failure to appreciate how words change might mean you misunderstand older texts — or worse, miss the joke. For example, when the great 11th century Islamic scholar Al-Ghazali wanted to poke fun at the divisive tendencies of his fellow Muslims, he used these two wonderfully humorous statements:
Muslims are so good at dividing that they can divide the atom. If you see two Muslims, probably they belong to three parties.
If you don't take into account the way that 11th century scholars used the word ‘atom’, you might miss the fact that both these statements are saying the divisiveness of Muslims is at logically impossible levels, not merely physically impressive (which seems a bit unfair to me — I'm quite sure that the history of Christianity could give them a run for their money).
The different meanings of words can be quite independent
It is not always the case, but very often the different meanings live their own independent lives, and are usually used quite separately. Failure to recognise this leads to various mistakes when interpreting communication, especially when not in our native tongues, or when we are over-thinking it.
One of these mistakes is called “illegitimate totality transfer”, or “unwarranted adoption of an expanded semantic field” in Carson's book. In this case, someone interprets a word as if it brings with it all its potential meanings. Normally when we use a word, we are using just one of the meanings.
If I said to you, “You’re dead right“, unless we have somehow found our way into one of those action films in which the hero, despite being in mortal danger, always has the time for a corny pun before dispatching the bad guy, it would be crazy of you to claim that my sentence had violent overtones because of the use of “dead”. Here it means “absolutely”, and that is all.
While this is obvious subconsciously, I've seen people making mistakes on exactly these lines when arguing about words.
The opposite mistake can be made too — a failure to recognise that a word brings with it, by association, other meanings. This is normally a failure by the person speaking. For example, take the word “bitch” meaning “female dog”. Amongst dog owners and breeders, this is a perfectly normal choice of word (at least in the UK). However, suppose you are speaking publicly, and not at a dog owners conference. You'd be well advised to be careful with this word, because its usage as a very strong insult has become so common. So there are several potential mistakes here:
A hearer who gets the wrong meaning — this is a failure at the subconscious level, the hearer didn't have enough background knowledge to automatically pick up the right meaning.
A hearer who understands the usage, but tries to claim the speaker is using language insulting to women, when that was not intended. This is someone whose sensitivity to words at the conscious level is turned up too high.
A speaker who made a bad choice of word that lead to the confusion. This is someone whose sensitivity to words, or knowledge of their audience, is not high enough.
In general, in a given instance, a single meaning of a word will be intended by the speaker or author. There are exceptions, however. Puns and other forms of rhetoric deliberately bring together distinct meanings of a word. Double meanings can also be a great way to appear to be saying one thing (perhaps aimed at one group of people), but actually be saying something else — in other words, a form of deniable deception. But this is the unusual case. Before claiming someone was deliberately using two meanings, you need to be pretty sure.
Language is not like maths
Since word meaning is defined by usage, and humans tend to use words with less than mathematical rigour, you can't apply strict logic to determine what words mean.
For example, take the following pairs: hot and cold; warm and cool.
We instinctively understand these as opposites, and together they form a spectrum — cold, cool, warm, hot. However, think about the meaning of these words when applied to people. All four are commonly used to describe human characteristics, but a “hot woman” is not the opposite of a “cold woman”, and a “cool guy” is not the opposite of a “warm guy”. 
These words have independent lives that don't follow mathematical laws.
There are no true synonyms
Part of this lack of mathematical precision is that no two words ever mean exactly the same thing. This is important, because some arguments about words implicitly depend on switching one word out for a synonym, but this always involves a change — the new word can bring with it different associations from the old, changing the meaning in subtle but crucial ways.
The writing of this post almost got derailed at this point, as I started wondering how quickly you could make a word diverge from its original meaning through the use of synonyms. How many synonyms does it take to make an antonym?, or “what is a the minimum length of a list of synonyms required to transform a word into an antonym?”. I got a little bit excited thinking about this — why, all you would need is a synonym and antonym dictionary in digital form and more hours than I could possibly justify finding an algorithm that could trim the potentially huge search space to get a result in a reasonable amount of time and memory, hmm, what a nice programming puzzle...
“Cleave!”, said my subconscious a few minutes later. To my great relief and disappointment, somehow my brain had been working on a different approach, and in a successful attempt to reclaim some of the free time that my conscious self had just mentally squandered, and to again demonstrate that it is sometimes much more intelligent than my conscious self, it found this word. “Cleave” is a verb meaning both:
To split with or as if with a sharp instrument (as in, cleaving a crystal, or a meat cleaver).
To adhere, cling, or stick fast (as in “therefore a man shall leave his father and mother and cleave unto his wife” Genesis 2:24).
In other words, it is its own antonym, or, as I soon found on wikipedia, an auto-antonym or contronym (and there are better examples, like the verbs ‘to sanction’ and ‘to dust’). For these words, a chain of zero synonyms is required to transform a word into an antonym, and you can’t get fewer than that. So there is your answer.
In other words: if you can't even trust the word itself not to transmogrify into its own opposite before your very eyes, you certainly can't trust synonyms to fully preserve meaning.
Meaning is determined by context
Given all of the above, with multiple meanings for words, and all kind of crazy irregularities, how do our brains sort out through all the mess to work out the meaning of a particular usage? The answer is we use context.
This includes context of various types:
Immediate grammatical form and context
For example, a word being singular or plural is a part of its grammatical form that forms a clue to which of several meanings you could be using. Making a word plural is not a simple mathematical multiplication — there is no problem using “cows” to refer to a group containing both male and females, for example, but to use “cow” for a single male would be a mistake.
I'm also including in this heading the immediately associated words, such as ”compound nouns”. In a compound noun, a preceding adjective or noun changes the meaning of the primary noun. Sometimes the words combine in obvious ways, and sometimes not. For example, a “water meter“ is fairly obviously a meter used for measuring water. However, while a year is a unit of time, a “light year” is actually a unit of distance. (thanks English stackexchange for the examples). Sometimes the second noun drops out altogether. “A microwave” is not “a very small wave”, but a type of oven (physicists would normally talk about “microwaves” not “a microwave” when referring to radiation of wavelength 1mm - 1m).
In English, phrasal verbs are a classic example where adding another word completely changes the meaning, and usually in very unpredictable ways. Knowing the verb “to make” and the preposition “up”, would you expect a non-native speaker to be able to guess the meaning of “to make up”? Or, should I say, one of the many meanings:
to make something up = to invent
to make something up = to put together from ingredients or components (making up a bed, or a drink)
to make something up = to fill up to an amount
to make up with someone = to be reconciled with
to make something up to someone = to do reparations
A different form of the verb can also change the meaning. Consider “tell” vs “can tell”. Sometimes “can” modifies tell in the way you would expect:
“I can tell you my doctor's name but not their telephone number” (I'm am able to tell you their name, but not able to tell you their telephone number, because I don't know it or I'm not allowed.)
However, sometimes a new meaning arises:
“I can tell that you are not happy.”
“I can tell” here does not mean “I have permission or ability to tell (someone) the following”. Rather, it means “From observation, I am able to deduce, and have deduced, the following information”.
This second meaning appears in dictionaries under “tell”, but in reality only appears when the word is combined with “can” or “could” – it is this particular grammatical context that clues us to the meaning.
Also note that a change in person (1st/2nd/3rd person) can again change the meaning. Using the same example — “you can tell” is slightly different from “I can tell”. The former means “the truth is evident, and you ought to be able to deduce it”, rather “you have been able to deduce it”.
This section could go on and on, you get the idea.
Context of surrounding sentence
Further clues to meaning, or things that affect meaning, are found in the surrounding words and sentence. The sentence “I'm not making it up” most probably is using the first meaning of “make up” given above, namely “invent”, although it might be one of the others. A few additional words from the context would make it clear.
The fact that meanings are defined by usage does not mean this is a completely “anything goes” arena, where we can use words any way we want, and no-one can tell us otherwise. A prescriptivist approach is fine, indeed required, in various contexts:
in teaching people the language.
in correcting usage. It is entirely possible to use a word incorrectly — that is, in a way that no-one else, or very few people, use it, which only leads to confusion. (Or, deception if it is done deliberately. If you consciously use a word in a way that your hearers will understand differently, it's called lying).
However, we will have to accept that what is currently ‘(in)correct’ usage can change over time.
in technical contexts, where certain distinctions are vital to maintain.
where certain words have become empty of useful meaning due to ambiguity. For example, Objectivity and Subjectivity in Statistics and Science argues that in the context of statistics, we should prefer terms like ““Transparency,” “consensus,” “impartiality,” and “correspondence to observable reality,” instead of “objectivity and “multiple perspectives” and “context-dependence,” instead of “subjectivity”.
When it comes to living languages, we are not relegated to the role of spectators — we can influence language usage, for good or bad. But ultimately, if we fail to influence others, we cannot claim to be authorities or say that commonly understood usages are objectively incorrect.
I was unsure whether to include this section at all, but in the end I decided this article lacked weight without highlighting some of the ways we go wrong, where it actually matters. I'll try not to call out bad examples using specific quotations — we've all made these kinds of mistakes, and my aim is for more fruitful debates about words in the future.
In this area, it's unfortunate that there is a huge amount of linguistic nonsense. For example, is “guys” a gender neutral word? Even the question is wrong — we should be talking about the meaning of usages, not the meaning of the word, because a single word has multiple meanings.
With the wrong starting point, its not surprising that we go downhill. Due to ignoring multiple meanings, this argument starts with the false dilemma of assuming that a word is either gender neutral or not gender neutral. It continues by pointing to an instance in which the word is being used in a gender specific way, and using that to prove that “guys” is not gender neutral — in other words, ignoring the fact that context changes the meaning of words.
Many arguments ignore the grammatical context. Singular “guy” and plural “guys” are not the same, and second person “you guys” or “hey guys” is not the same as third person “those guys”. Regional and social differences in usage are also ignored.
Now, in response to this, some people might argue that “guys” is not gender neutral because at least one of its meanings is male biased, with no corresponding female biased meaning, and therefore overall it is not gender neutral — and therefore shouldn't be used for mixed groups because it excludes people. I think this argument is still very flawed however:
It still fails to understand the fact that words have multiple separate meanings.
It is guilty of “totality transfer” i.e. insisting that one usage's meaning necessarily affects the other usages. Attempting to add up the meanings of a single word is simply not a meaningful thing to do.
It creates an impossible standard for communication that could be pointless or counter productive.
With this way of handling words (i.e. you should take all the meanings/usages of a word and combine them to get an overall judgement about the qualities of a word), are there any truly gender neutral words? For example, is "doctor" gender neutral? The famous riddle about a doctor and their son shows that for many people “doctor” is more likely to conjure up a male image than a female one. Since meaning is defined by usage (intent and understanding), you could argue that “doctor” is therefore male biased (even if in some usages it is not meant that way). Therefore, “doctor” shouldn't be used for female doctors.
A better example might be “actor”. Often this is both intended and understood to refer specifically to men — for example in the phrase “actors and actresses”. It is never used to refer specifically to women as opposed to men. Therefore, the word is not gender neutral. Therefore, it should not be used for women because it excludes them.
Is that really what we want — to be left without a gender neutral word for actor? As I understand it, the acting profession has gone in the opposite direction, and tends these days to prefer “actor” over “actress” for women.
We should also note that gender bias in perceptions regarding doctors exists despite the word having no obvious etymological gender bias.
So, whether a word is gender neutral is either:
meaningless — because it is usages that are or aren't gender neutral, or
irrelevant — because if we decide that ‘overall’ a word is or isn't gender neutral, it gives us little help in deciding whether we should continue using it in gender neutral ways.
On a positive note, an example of handling this issue really well is Julia Evans' When is guys gender neutral? I did a survey. While there might be various issues with her survey which she points out, she is firstly asking a question (instead of just assuming), and secondly asking the right question: how do people use and understand the word “guys” in different contexts?
And this survey points to a much better argument for not using “guys” to refer to mixed groups, even in the second person plural when it appears to be at its most gender neutral: for some people it would cause confusion, due to regional/social/gender differences in usage. This applies especially when you don't know your audience's background and how they understand words (e.g. the internet). In other words, it is a simple matter of communicating well.
I'm obviously not saying there is no such thing as unhelpful gender biased language. And there are some much more subtle issues and arguments in this area that merit more consideration. But no light is shed on these real issues by arguments that fail to understand the basics of how words work.
A common fallacy in arguing about racist language is related to the issue of there being no true synonyms. For example, you could argue that “black”, “negro” and “nigger” all fundamentally mean the same thing, but this ignores the associations and the fact that some words are used as racial slurs while others are not. To argue that we can use “nigger” because “black” is fine is completely wrong – and it's equally wrong to argue the converse.  Thankfully I think we've seen the back of these kinds of arguments.
But unfortunately they have given way to perhaps more ridiculous ones. For example, despite there being no racial overtones or associations in the history or the current usage, some people have decided to blacklist blacklist. This requires believing that people are really extremely stupid when it comes to handling words — that they cannot separate different usages of the word “black” in their minds. The fact is, we have no problem doing this all the time.
The irony is that, in thinking that our subconscious brains are stupid, it is our conscious selves that go on to perform acts of great stupidity when handling the exact same issue.
Black and dark are used in various contexts with negative meanings (e.g. blacklist, black thoughts, black deeds, Black Friday), while white and light are often associated with positive meanings, not because of racism, but for a range of reasons mainly to do with natural analogies (and also possibly due to various accidents of history).
The natural reasons are not hard to discover. The blackness of night is a good cover for criminal or shameful acts, and a cause of fear for those on the receiving end. The lack of sight when in darkness is a great physical metaphor for the lack of hope when in despair. People across huge ranges of time and culture have used these words this way because of the physical analogies (see 1 Thessalonians 5:5-7, John 3:19-20, Isaiah 9:2 from the Bible for example, dating back 2000 to 2700 years ago). If you feel all of these need to be corrected, you are going to have to ‘correct’ a fundamental way in which our brains are wired to find analogies.
If you try instead to explain these usages as some global, cross-cultural, pan-historical conspiracy to denigrate black people (pun intended), then you will need to deal with usages that go against the pattern — “in the black” (financially sound), “black belt” and “white with rage”, for example. You'll see that all of these work by (non-racist) metaphorical thinking, just like the ones in the previous paragraphs.
Even if a usage does have its roots in racism, it can still be a mistake to think that the words are racist in their current usage. I was surprised to find that “long time no see” was originally imitating a form of pidgin English (either Native American or Chinese). It has become such a part of normal English, that I had never noticed the fact that is ungrammatical, and I have certainly never heard it used in a mocking way. Any insulting overtones have been lost.
The real question to ask about these words is not what their origin is, but “Do these usages contribute to racism now?”. Since our brains are perfectly capable of separating different usages, there is absolutely no reason to think that they would promote racism. For example, when we talk of “feeling blue” (gloomy), or “blue skies” (theoretical or impractical when used as a technical term, or happiness when used as a metaphor), or a “blue movie” (pornography), all these different meanings of blue do not colour the other usages at all, even when they are opposites of each other! In fact, the wittiness of puns and other word games often relies on connecting different usages of the same word. We find these things clever precisely because our brains have segregated the meanings so entirely that we are surprised to find the word is actually the same.
Well, this has turned into a sermon, so I may as well have an explicit “lessons” section.
Given that context is so vital, and given that people in different places use words differently, we should probably exercise caution if we happen to be in a context that, say, brings people together across vast geographical and social contexts, especially if, say, our medium of communication deliberately and artificially makes it hard to get context of more than 140 characters. It might just make things a little bit tricky. That caution applies to both what we say and what we think we've understood.
We should remember our subconscious brains are much better at words than our conscious understanding. We should remember that other people's brains are also much better at words than our conscious understanding. We should remember that the meanings of words change, and are hugely dependent on social, historical, literary and grammatical context.
So we should think twice about telling other people their usage of words is incorrect, or will cause confusion, or induce sexism, racism, fascism or destroy society in some other way. There are plenty of real battles to fight in these areas. We should make sure we are fighting one of those.