1
00:00:06,814 --> 00:00:14,045
The average 20-year-old knows between
27,000 and 52,000 different words.

2
00:00:14,045 --> 00:00:20,053
By age 60, that number averages between
35,000 and 56,000.

3
00:00:20,053 --> 00:00:24,330
Spoken out loud, most of these words last
less than a second.

4
00:00:24,330 --> 00:00:28,535
So with every word, the brain has a quick
decision to make:

5
00:00:28,535 --> 00:00:32,235
which of those thousands of options 
matches the signal?

6
00:00:32,235 --> 00:00:36,345
About 98% of the time, the brain chooses
the correct word.

7
00:00:36,345 --> 00:00:41,115
But how? Speech comprehension is different
from reading comprehension,

8
00:00:41,115 --> 00:00:44,375
but it’s similar to sign language 
comprehension—

9
00:00:44,375 --> 00:00:48,861
though spoken word recognition has 
been studied more than sign language.

10
00:00:48,861 --> 00:00:51,421
The key to our ability to understand 
speech

11
00:00:51,421 --> 00:00:54,691
is the brain’s role as a 
parallel processor,

12
00:00:54,691 --> 00:00:58,691
meaning that it can do multiple different 
things at the same time.

13
00:00:58,691 --> 00:01:01,301
Most theories assume that each word
we know

14
00:01:01,301 --> 00:01:05,771
is represented by a separate processing 
unit that has just one job:

15
00:01:05,771 --> 00:01:10,931
to assess the likelihood of incoming 
speech matching that particular word.

16
00:01:10,931 --> 00:01:15,139
In the context of the brain, the 
processing unit that represents a word

17
00:01:15,139 --> 00:01:19,796
is likely a pattern of firing activity 
across a group of neurons

18
00:01:19,796 --> 00:01:21,686
in the brain’s cortex.

19
00:01:21,686 --> 00:01:23,506
When we hear the beginning of a word,

20
00:01:23,506 --> 00:01:27,286
several thousand such units 
may become active,

21
00:01:27,286 --> 00:01:31,532
because with just the beginning of a 
word, there are many possible matches.

22
00:01:31,532 --> 00:01:35,535
Then, as the word goes on, more and
more units register

23
00:01:35,535 --> 00:01:40,666
that some vital piece of information 
is missing and lose activity.

24
00:01:40,666 --> 00:01:43,126
Possibly well before the end of the word,

25
00:01:43,126 --> 00:01:48,090
just one firing pattern remains active, 
corresponding to one word.

26
00:01:48,090 --> 00:01:50,828
This is called the ‘recognition point.’

27
00:01:50,828 --> 00:01:53,648
In the process of honing in on one word,

28
00:01:53,648 --> 00:01:56,718
the active units suppress 
the activity of others,

29
00:01:56,718 --> 00:01:58,838
saving vital milliseconds.

30
00:01:58,838 --> 00:02:03,635
Most people can comprehend up to 
about 8 syllables per second.

31
00:02:03,635 --> 00:02:06,965
Yet, the goal is not only 
to recognize the word,

32
00:02:06,965 --> 00:02:10,415
but also to access its stored meaning.

33
00:02:10,415 --> 00:02:14,195
The brain accesses many possible meanings
at the same time,

34
00:02:14,195 --> 00:02:16,875
before the word has been fully identified.

35
00:02:16,875 --> 00:02:22,018
We know this from studies which show 
that even upon hearing a word fragment––

36
00:02:22,018 --> 00:02:23,298
like ‘cap’ ––

37
00:02:23,298 --> 00:02:26,798
listeners will start to register multiple 
possible meanings,

38
00:02:26,798 --> 00:02:31,970
like captain or capital,
before the full word emerges.

39
00:02:31,970 --> 00:02:35,120
This suggests that every time we hear a 
word

40
00:02:35,120 --> 00:02:38,480
there’s a brief explosion of meanings in 
our minds,

41
00:02:38,480 --> 00:02:43,291
and by the recognition point the brain 
has settled on one interpretation.

42
00:02:43,291 --> 00:02:46,221
The recognition process moves more 
rapidly

43
00:02:46,221 --> 00:02:50,821
with a sentence that gives us context 
than in a random string of words.

44
00:02:50,821 --> 00:02:55,009
Context also helps guide us towards the
intended meaning of words

45
00:02:55,009 --> 00:02:59,009
with multiple interpretations, like ‘bat,’
or ‘crane,’

46
00:02:59,009 --> 00:03:03,009
or in cases of homophones
like ‘no’ or ‘know.’

47
00:03:03,009 --> 00:03:07,393
For multilingual people, the language
they are listening to is another cue,

48
00:03:07,393 --> 00:03:12,706
used to eliminate potential words
that don’t match the language context.

49
00:03:12,706 --> 00:03:16,706
So, what about adding completely new 
words to this system?

50
00:03:16,706 --> 00:03:20,706
Even as adults, we may come across a 
new word every few days.

51
00:03:20,706 --> 00:03:25,109
But if every word is represented as a 
fine-tuned pattern of activity

52
00:03:25,109 --> 00:03:27,439
distributed over many neurons,

53
00:03:27,439 --> 00:03:31,992
how do we prevent new words from 
overwriting old ones?

54
00:03:31,992 --> 00:03:34,322
We think that to avoid this problem,

55
00:03:34,322 --> 00:03:39,085
new words are initially stored in a part 
of the brain called the hippocampus,

56
00:03:39,085 --> 00:03:42,693
well away from the main store of words
in the cortex,

57
00:03:42,693 --> 00:03:46,063
so they don’t share neurons
with others words.

58
00:03:46,063 --> 00:03:49,073
Then, over multiple nights of sleep,

59
00:03:49,073 --> 00:03:54,470
the new words gradually transfer over
and interweave with old ones.

60
00:03:54,470 --> 00:03:57,990
Researchers think this gradual
acquisition process

61
00:03:57,990 --> 00:04:01,354
helps avoid disrupting existing words.

62
00:04:01,354 --> 00:04:02,774
So in the daytime,

63
00:04:02,774 --> 00:04:07,304
unconscious activity generates explosions
of meaning as we chat away.

64
00:04:07,304 --> 00:04:12,305
At night, we rest, but our brains are 
busy integrating new knowledge

65
00:04:12,305 --> 00:04:14,125
into the word network.

66
00:04:14,125 --> 00:04:17,655
When we wake up, this process ensures 
that we’re ready

67
00:04:17,596 --> 00:04:20,696
for the ever-changing world of language.