WEBVTT
Kind: captions
Language: en-GB

1
00:00:00.320 --> 00:00:04.460
If you ask an English teacher marking an essay,
then counting words is simple. It is the number

2
00:00:04.460 --> 00:00:08.090
that comes up when you hit the “Word Count”
button in whichever program you’re writing

3
00:00:08.090 --> 00:00:11.879
with. For English at least, word counting
algorithms generally count a word as any string

4
00:00:11.879 --> 00:00:15.869
of ‘letter’ characters. Each string of
those, separated by some other character?

5
00:00:15.869 --> 00:00:17.210
That’s a word.

6
00:00:17.210 --> 00:00:20.619
The trouble is, while that might be good enough
for checking that an essay is roughly the

7
00:00:20.619 --> 00:00:23.900
length you want it to be, it’s going to
count plenty of things that aren’t words

8
00:00:23.900 --> 00:00:29.089
too -- and it doesn’t get to a more fundamental
question: what actually is a word?

9
00:00:29.089 --> 00:00:33.329
First of all, there’s the problem of meaning.
If you accept the definition that a “word”

10
00:00:33.329 --> 00:00:40.420
is some letters surrounded by a gap, then
“xnopyt”, “aaaaaaajjjjjjjjj” and

11
00:00:40.420 --> 00:00:44.719
"hrrkrkrkrwpfrbrbrbrlablblblblblblwhitoo'ap" are all words, despite being pretty much meaningless.

12
00:00:44.719 --> 00:00:49.079
You certainly can’t rely on dictionaries
to define what’s a “real word”, either.

13
00:00:49.079 --> 00:00:52.039
There are plenty of words that people use
every day that haven’t made it into standard

14
00:00:52.039 --> 00:00:56.640
dictionaries -- okay, maybe that one isn’t
used every day any more -- but they clearly

15
00:00:56.640 --> 00:00:57.840
have meaning.

16
00:00:57.840 --> 00:01:03.710
So, okay. Let’s say that a word is some
characters, surrounded by something else,

17
00:01:03.710 --> 00:01:08.670
that conveys some meaning. That’s not a
bad definition on paper. I mean, literally,

18
00:01:08.670 --> 00:01:12.930
that’s not a bad way to define English words
when they’re written down on paper. But

19
00:01:12.930 --> 00:01:14.720
language isn’t just about writing.

20
00:01:14.720 --> 00:01:19.290
That definition, spaces around some letters,
doesn’t handle contractions well at all.

21
00:01:19.290 --> 00:01:22.939
In everyday English speech, there are far
more contractions than you ever see written

22
00:01:22.939 --> 00:01:27.570
down. Depending on your dialect, “I’d-a”
is a perfectly good replacement for “I would

23
00:01:27.570 --> 00:01:32.140
have”, as in “I’d-a thought you’d
known that.” Is that three words? Two words?

24
00:01:32.140 --> 00:01:32.899
Just one?

25
00:01:32.899 --> 00:01:37.189
How about “I’m-a”, as in, “I’m-a
let you finish”? There are a dozen different

26
00:01:37.189 --> 00:01:42.030
ways to even spell that one right now, and
it’s replacing four Received Pronunciation

27
00:01:42.030 --> 00:01:46.840
words, “I am going to”. Except there’s
also a decent argument that “I’m-a”

28
00:01:46.840 --> 00:01:51.240
isn’t even a contraction of those words,
it’s a completely separate word in itself,

29
00:01:51.240 --> 00:01:55.460
imma. Either way, good luck counting the number
of words you’ve actually got there.

30
00:01:55.460 --> 00:02:00.219
And to make it worse, the meanings of words,
and how they’re presented, change over time.

31
00:02:00.219 --> 00:02:05.869
“Today” used to be hyphenated: “to-day”,
and before that, it was two words: “to day”.

32
00:02:05.869 --> 00:02:09.880
When did it become one word, instead of two?
How about “first class”? Does that become

33
00:02:09.880 --> 00:02:13.130
a single word if you hyphenate it instead?
You’ll find that most people will give two

34
00:02:13.130 --> 00:02:17.580
different answers: to-day is one word, but
first-class is two.

35
00:02:17.580 --> 00:02:22.050
That’s while we’re still talking about
English. How about synthetic languages, ones

36
00:02:22.050 --> 00:02:27.320
where, in extreme cases, you can put an entire
sentence’s worth of meaning into one long

37
00:02:27.320 --> 00:02:31.700
string of characters or sounds? Is that still
one word?

38
00:02:31.700 --> 00:02:36.860
And more important than any of that: does
it matter?

39
00:02:36.860 --> 00:02:40.460
Wordcount is a convenient thing to refer to
if you're tracking your progress on a written

40
00:02:40.460 --> 00:02:44.310
project like a dissertation: regardless of
how exactly you define a word, it's clearly

41
00:02:44.310 --> 00:02:49.610
a massive feat to write 50,000 of them coherently.
But when you zoom in on language as a system,

42
00:02:49.610 --> 00:02:55.090
it makes the most sense to ignore words entirely
and talk about morphemes or phrases.

43
00:02:55.090 --> 00:02:58.150
Words might be the most convenient way of
dividing up an English sentence, but they

44
00:02:58.150 --> 00:03:02.620
don’t really have much use to linguists.
And also, here’s one question to leave you

45
00:03:02.620 --> 00:03:09.920
with, that no-one’s tackled yet: are emoticons,
or emoji, words? They’re symbols that convey

46
00:03:09.920 --> 00:03:14.860
meaning, sometimes literal, surrounded by
space, sometimes just as a modifier to a sentence.

47
00:03:14.860 --> 00:03:22.030
If I were to end a written version of this
with a smiley face: should that count as a word?

48
00:03:22.030 --> 00:03:24.030
☺