WEBVTT
Kind: captions
Language: en-GB

1
00:00:00.110 --> 00:00:02.840
Last year, the Associated Press sent out this
headline.

2
00:00:02.840 --> 00:00:07.140
Now, I was working in a newspaper’s office
that week -- I know, newspapers still exist,

3
00:00:07.140 --> 00:00:11.780
it’s weird -- and so everyone around me
was pretty much plugged directly into Twitter,

4
00:00:11.780 --> 00:00:16.840
and the reaction was astonishing. Flight 17
had gone down only a week earlier, the idea

5
00:00:16.840 --> 00:00:21.150
that another plane, one carrying bodies from
that disaster, could also have crashed? It

6
00:00:21.150 --> 00:00:22.810
seemed unreal.

7
00:00:22.810 --> 00:00:24.670
And it was.

8
00:00:24.670 --> 00:00:29.040
Because what the Associated Press actually
meant was: Dutch military plane, COMMA, carrying

9
00:00:29.040 --> 00:00:34.660
bodies from Malaysia Airlines Flight 17 crash,
COMMA, lands in Eindhoven.

10
00:00:34.660 --> 00:00:38.450
Now there were many problems with that message,
not least the fact that a plane landing uneventfully

11
00:00:38.450 --> 00:00:43.760
is not breaking news, but it’s also a great
example of how ambiguous English can be. The

12
00:00:43.760 --> 00:00:47.720
result is called a “crash blossom”, after
this particularly ambiguous headline from

13
00:00:47.720 --> 00:00:49.829
a few years ago.

14
00:00:49.829 --> 00:00:54.600
Ambiguity is when you have a word or sentence
that can mean two different things. All languages,

15
00:00:54.600 --> 00:00:59.969
except for some incredibly pedantic constructed
ones, have ambiguity. Now we’ll set aside

16
00:00:59.969 --> 00:01:03.420
for a moment pragmatic ambiguity, which is
when you raise your hand in class and say

17
00:01:03.420 --> 00:01:07.250
“oh, can I go to the toilet?” and some
smart-arse of a teacher replies “I don’t

18
00:01:07.250 --> 00:01:09.130
know, can you?” [DRUM STING]

19
00:01:09.130 --> 00:01:14.880
Leaving that aside, English has scopal, lexical
and structural ambiguity.

20
00:01:14.880 --> 00:01:19.180
Scopal ambiguity is when you have two or more
quantifiers in a sentence, and the order in

21
00:01:19.180 --> 00:01:24.490
which they’re applied isn’t clear. “Everybody
loves somebody” is the canonical example:

22
00:01:24.490 --> 00:01:30.590
it can mean that for every person, there is
someone that they love -- or it can mean that

23
00:01:30.590 --> 00:01:35.070
there’s a single somebody that’s loved
by everyone. It’s like the old joke, “In

24
00:01:35.070 --> 00:01:41.909
the UK, a woman gives birth every 30 seconds.
She must be exhausted.” [DRUM STING]

25
00:01:41.909 --> 00:01:45.600
Lexical ambiguity is when you’ve got a word
that can mean multiple things. You might have

26
00:01:45.600 --> 00:01:50.900
heard the famous sentence, “Buffalo buffalo
Buffalo buffalo buffalo buffalo Buffalo buffalo”...

27
00:01:50.900 --> 00:01:55.960
...but my favourite one is from the opening
chapters of the Hitchhiker’s Guide to the

28
00:01:55.960 --> 00:01:59.970
Galaxy, where hyperspace travel is described
as “unpleasantly like being drunk.”

29
00:01:59.970 --> 00:02:01.390
“What’s so wrong about being drunk?”

30
00:02:01.390 --> 00:02:03.360
“You ask a glass of water.” [DRUM STING]

31
00:02:03.360 --> 00:02:08.250
Mind you, that’s not just lexical: that
change in meaning also changes the structure

32
00:02:08.250 --> 00:02:09.039
of the sentence.

33
00:02:09.039 --> 00:02:13.700
And that’s the most interesting kind of
ambiguity: structural ambiguity, which is

34
00:02:13.700 --> 00:02:16.540
what we started with. Let’s have a look
at that “crash blossom” sentence.

35
00:02:16.540 --> 00:02:21.550
JAL is Japan Airlines, by the way, but let’s
take that out to make the sentence a bit simpler.

36
00:02:21.550 --> 00:02:26.000
Okay. Let’s diagram this sentence. If you’re
American, you might actually have been taught

37
00:02:26.000 --> 00:02:30.879
how to do this in school, for some bizarre
reason. This shows you how it was intended.

38
00:02:30.879 --> 00:02:34.370
And presumably the headline writer, having
come up with it, didn’t see it any other

39
00:02:34.370 --> 00:02:38.819
way. Most readers, though, saw this. Why?

40
00:02:38.819 --> 00:02:43.560
There are eye tracking studies, watching how
people read and re-read a sentence, that show

41
00:02:43.560 --> 00:02:48.650
that we build the interpretation of a sentence
in our head as we read it. As we scan left

42
00:02:48.650 --> 00:02:52.750
to right -- in English, anyway -- as we scan
left to right, our brain tries to assemble

43
00:02:52.750 --> 00:02:57.910
concepts out of the words it’s seeing. So:
Violinist linked to crash… yep, okay…

44
00:02:57.910 --> 00:02:59.440
what’s the next word?

45
00:02:59.440 --> 00:03:04.209
If it were “recovers”, then that’s clearly
a verb. It must apply to the violinist, so

46
00:03:04.209 --> 00:03:08.519
our brain settles on that one really quickly.
If it were “site”, then that’s clearly

47
00:03:08.519 --> 00:03:13.010
a noun, it must be part of the noun phrase
“crash site”, our brain settles on that

48
00:03:13.010 --> 00:03:13.590
quickly.

49
00:03:13.590 --> 00:03:20.050
“Blossoms”, though. That can be a verb,
sure, but it can also be a noun. There’s

50
00:03:20.050 --> 00:03:24.940
no clear link, and, heck, it’s physically
closer to ‘crash’ than ‘violinist’,

51
00:03:24.940 --> 00:03:29.709
so most people, it seems, come up with “crash
blossoms” and then go “wait, what”?

52
00:03:29.709 --> 00:03:34.909
And that brings us to these wonderful constructions
called garden path sentences. They’re called

53
00:03:34.909 --> 00:03:38.319
that because they lead you up the garden path,
they trick you, they send you down a false

54
00:03:38.319 --> 00:03:39.580
trail.

55
00:03:39.580 --> 00:03:44.190
Garden path sentences solve the question of
whether we take a whole sentence into memory

56
00:03:44.190 --> 00:03:48.390
before trying to comprehend it, or whether
we build its structure as we go along. We

57
00:03:48.390 --> 00:03:53.140
must do it piece by piece: if you waited to
see the whole thing before working it out

58
00:03:53.140 --> 00:03:55.959
as a big block, then you wouldn't get confused.

59
00:03:55.959 --> 00:04:01.459
For more complicated garden path sentences,
like this lovely one -- go ahead, take a second,

60
00:04:01.459 --> 00:04:06.220
I’d spoil it with intonation if I read it
out loud -- eye tracking studies mean we can

61
00:04:06.220 --> 00:04:10.769
watch people jump back to earlier points in
the sentence and scan around, while their

62
00:04:10.769 --> 00:04:13.120
brain tries to work out what it’s been shown.

63
00:04:13.120 --> 00:04:16.479
Generally, ambiguity is something to be avoided
in everyday communication. Sure, it’s the

64
00:04:16.479 --> 00:04:20.939
foundation of countless jokes, and very occasionally
legal loopholes, but if you’re just trying

65
00:04:20.939 --> 00:04:24.150
to be understood, well, it’s probably not
going to help you out.

66
00:04:24.150 --> 00:04:27.869
But linguists can love ambiguity more than
most people.

67
00:04:27.869 --> 00:04:31.529
More than most -- more than most people. More
than…

68
00:04:32.080 --> 00:04:34.080
[Translating these subtitles? Add your name here!]