WEBVTT
Kind: captions
Language: en

1
00:00:00.340 --> 00:00:02.060
I thought this debate was settled.

2
00:00:02.060 --> 00:00:05.510
I thought that /gɪf/, with a hard G,
had won.

3
00:00:05.510 --> 00:00:06.510
But apparently not.

4
00:00:06.510 --> 00:00:08.390
It is more popular than /ʤɪf/,

5
00:00:08.390 --> 00:00:10.100
but it’s not a landslide:

6
00:00:10.100 --> 00:00:12.650
in a 2014 survey,
only 70% of people said /gɪf/.

7
00:00:12.650 --> 00:00:14.880
And the format’s creator, Steve Wilhite,

8
00:00:14.880 --> 00:00:16.110
argues that it should be /ʤɪf/,

9
00:00:16.110 --> 00:00:18.350
and has been arguing that for a long time.

10
00:00:18.350 --> 00:00:19.690
In fact, when the Webby Awards let him give

11
00:00:19.690 --> 00:00:22.820
one of their famous 5-words-only
acceptance speeches,

12
00:00:22.820 --> 00:00:27.780
he said: “it’s pronounced
/ʤɪf/, not /gɪf/”.

13
00:00:27.780 --> 00:00:29.869
This is the slide he put up with it.

14
00:00:29.869 --> 00:00:32.410
And, uh, yeah, you can see the problem there.

15
00:00:32.410 --> 00:00:36.010
Now, one of the most fundamental principles
of modern linguistics is descriptivism:

16
00:00:36.010 --> 00:00:40.649
there should be no value judgment about particular
words or pronunciations or types of speech,

17
00:00:40.649 --> 00:00:44.420
there must be no Correct Way To Speak
handed down from on high.

18
00:00:44.420 --> 00:00:48.530
We describe how people speak.
If language changes, we change with it.

19
00:00:48.530 --> 00:00:51.989
So the Oxford English Dictionary
accepts both pronunciations,

20
00:00:51.989 --> 00:00:55.180
despite Steve Wilhite calling them
"Wrong… end of story."

21
00:00:55.180 --> 00:00:56.690
Turns out, even if you invented a word,

22
00:00:56.690 --> 00:00:59.260
you don’t get to make the call
on how it’s pronounced.

23
00:00:59.260 --> 00:01:03.160
Whether he likes it or not, /gɪf/ is the
more common pronunciation.

24
00:01:03.160 --> 00:01:05.600
That said: linguistics also gives us the tools

25
00:01:05.600 --> 00:01:08.850
to look at all the other words in English
and figure out, well, okay.

26
00:01:08.850 --> 00:01:12.590
We’ve got this new word.
What other words is it most like?

27
00:01:12.590 --> 00:01:16.779
You might hear the argument that /gɪf/ is
short for Graphics Interchange Format.

28
00:01:16.779 --> 00:01:18.079
Graphics, not /ʤ/raphics.

29
00:01:18.079 --> 00:01:21.429
But that’s not how English treats acronyms.

30
00:01:21.429 --> 00:01:23.609
In JPEG, the P isn’t short for "potography".

31
00:01:23.609 --> 00:01:27.630
And in the same way, laser isn’t pronounced
"lass-uhr",

32
00:01:27.630 --> 00:01:29.859
and scuba isn't "scubb-uh".

33
00:01:29.859 --> 00:01:32.700
So rather than acronyms,
what about other English words?

34
00:01:32.700 --> 00:01:36.659
Well, in the red corner: git, gift, gaffer.

35
00:01:36.659 --> 00:01:39.919
In the blue corner: gin, gist, gel.

36
00:01:39.919 --> 00:01:42.649
Okay, so if you went through the whole dictionary:

37
00:01:42.649 --> 00:01:43.979
which side would have more words?

38
00:01:43.979 --> 00:01:47.380
Well, in 2015, a researcher called
Michael Dow did just that,

39
00:01:47.380 --> 00:01:49.639
only counting the words that have one morpheme.

40
00:01:49.639 --> 00:01:53.609
So, words like "gift": you can’t break that
down into smaller lexical parts.

41
00:01:53.609 --> 00:01:57.869
"Gift-ed" wouldn’t count, you can break
that down into "gift" and "ed".

42
00:01:57.869 --> 00:02:01.749
So just using those really basic words,
it’s a victory for the /ʤɪf/-ers:

43
00:02:01.749 --> 00:02:06.350
there are more words where "g-i" is pronounced
/ʤɪ/ than /gɪ/.

44
00:02:06.350 --> 00:02:08.979
But… the /gɪ/ words are are used more often.

45
00:02:08.979 --> 00:02:11.519
So the actual likelihood of each pronunciation
is about equal.

46
00:02:11.519 --> 00:02:16.090
But if you look only at words that start with
g-i, /gɪ/ is more common.

47
00:02:16.090 --> 00:02:18.349
So now we’re on a question of psycholinguistics!

48
00:02:18.349 --> 00:02:19.349
When we read a new word,

49
00:02:19.349 --> 00:02:23.219
do we consider all the times we’ve seen
those letters before anywhere,

50
00:02:23.219 --> 00:02:25.980
or do we favour ones that are
in the same position?

51
00:02:25.980 --> 00:02:30.790
And even if we do, why do we
pick one over the other?

52
00:02:30.790 --> 00:02:32.349
Unless you’re very young,

53
00:02:32.349 --> 00:02:35.549
GIF is a word that you almost certainly read
before you heard.

54
00:02:35.549 --> 00:02:37.810
And the first time you ever saw
these three letters,

55
00:02:37.810 --> 00:02:40.939
your brain instantly tried to match them to
patterns you already knew.

56
00:02:40.939 --> 00:02:42.780
Gib, gift, gin, git, give.

57
00:02:42.780 --> 00:02:45.840
Maybe /g/ill or /ʤ/ill, both of which are
valid English words,

58
00:02:45.840 --> 00:02:47.960
one's an old measurement,
one's a part of a fish.

59
00:02:47.960 --> 00:02:50.761
/g/illingham or /ʤ/illingham. Both English towns,
a couple of hundred miles apart.

60
00:02:50.761 --> 00:02:56.950
In a split second, your brain went through
everything that it knew about g-i,

61
00:02:56.950 --> 00:03:01.480
decided which pronunciation you thought was
right, and locked that in your head.

62
00:03:01.480 --> 00:03:03.160
The correct answer is, of course,

63
00:03:06.390 --> 00:03:09.209
My co-author Gretchen McCulloch has a great
podcast called Lingthusiasm

64
00:03:09.209 --> 00:03:10.980
that covers all sorts of things like this,

65
00:03:10.980 --> 00:03:12.890
I recommend it, and the link is in the description.