0:00:06.814,0:00:14.045
The average 20-year-old knows between[br]27,000 and 52,000 different words.

0:00:14.045,0:00:20.053
By age 60, that number averages between[br]35,000 and 56,000.

0:00:20.053,0:00:24.330
Spoken out loud, most of these words last[br]less than a second.

0:00:24.330,0:00:28.535
So with every word, the brain has a quick[br]decision to make:

0:00:28.535,0:00:32.235
which of those thousands of options [br]matches the signal?

0:00:32.235,0:00:36.345
About 98% of the time, the brain chooses[br]the correct word.

0:00:36.345,0:00:41.115
But how? Speech comprehension is different[br]from reading comprehension,

0:00:41.115,0:00:44.375
but it’s similar to sign language [br]comprehension—

0:00:44.375,0:00:48.861
though spoken word recognition has [br]been studied more than sign language.

0:00:48.861,0:00:51.421
The key to our ability to understand [br]speech

0:00:51.421,0:00:54.691
is the brain’s role as a [br]parallel processor,

0:00:54.691,0:00:58.691
meaning that it can do multiple different [br]things at the same time.

0:00:58.691,0:01:01.301
Most theories assume that each word[br]we know

0:01:01.301,0:01:05.771
is represented by a separate processing [br]unit that has just one job:

0:01:05.771,0:01:10.931
to assess the likelihood of incoming [br]speech matching that particular word.

0:01:10.931,0:01:15.139
In the context of the brain, the [br]processing unit that represents a word

0:01:15.139,0:01:19.796
is likely a pattern of firing activity [br]across a group of neurons

0:01:19.796,0:01:21.686
in the brain’s cortex.

0:01:21.686,0:01:23.506
When we hear the beginning of a word,

0:01:23.506,0:01:27.286
several thousand such units [br]may become active,

0:01:27.286,0:01:31.532
because with just the beginning of a [br]word, there are many possible matches.

0:01:31.532,0:01:35.535
Then, as the word goes on, more and[br]more units register

0:01:35.535,0:01:40.666
that some vital piece of information [br]is missing and lose activity.

0:01:40.666,0:01:43.126
Possibly well before the end of the word,

0:01:43.126,0:01:48.090
just one firing pattern remains active, [br]corresponding to one word.

0:01:48.090,0:01:50.828
This is called the ‘recognition point.’

0:01:50.828,0:01:53.648
In the process of honing in on one word,

0:01:53.648,0:01:56.718
the active units suppress [br]the activity of others,

0:01:56.718,0:01:58.838
saving vital milliseconds.

0:01:58.838,0:02:03.635
Most people can comprehend up to [br]about 8 syllables per second.

0:02:03.635,0:02:06.965
Yet, the goal is not only [br]to recognize the word,

0:02:06.965,0:02:10.415
but also to access its stored meaning.

0:02:10.415,0:02:14.195
The brain accesses many possible meanings[br]at the same time,

0:02:14.195,0:02:16.875
before the word has been fully identified.

0:02:16.875,0:02:22.018
We know this from studies which show [br]that even upon hearing a word fragment––

0:02:22.018,0:02:23.298
like ‘cap’ ––

0:02:23.298,0:02:26.798
listeners will start to register multiple [br]possible meanings,

0:02:26.798,0:02:31.970
like captain or capital,[br]before the full word emerges.

0:02:31.970,0:02:35.120
This suggests that every time we hear a [br]word

0:02:35.120,0:02:38.480
there’s a brief explosion of meanings in [br]our minds,

0:02:38.480,0:02:43.291
and by the recognition point the brain [br]has settled on one interpretation.

0:02:43.291,0:02:46.221
The recognition process moves more [br]rapidly

0:02:46.221,0:02:50.821
with a sentence that gives us context [br]than in a random string of words.

0:02:50.821,0:02:55.009
Context also helps guide us towards the[br]intended meaning of words

0:02:55.009,0:02:59.009
with multiple interpretations, like ‘bat,’[br]or ‘crane,’

0:02:59.009,0:03:03.009
or in cases of homophones[br]like ‘no’ or ‘know.’

0:03:03.009,0:03:07.393
For multilingual people, the language[br]they are listening to is another cue,

0:03:07.393,0:03:12.706
used to eliminate potential words[br]that don’t match the language context.

0:03:12.706,0:03:16.706
So, what about adding completely new [br]words to this system?

0:03:16.706,0:03:20.706
Even as adults, we may come across a [br]new word every few days.

0:03:20.706,0:03:25.109
But if every word is represented as a [br]fine-tuned pattern of activity

0:03:25.109,0:03:27.439
distributed over many neurons,

0:03:27.439,0:03:31.992
how do we prevent new words from [br]overwriting old ones?

0:03:31.992,0:03:34.322
We think that to avoid this problem,

0:03:34.322,0:03:39.085
new words are initially stored in a part [br]of the brain called the hippocampus,

0:03:39.085,0:03:42.693
well away from the main store of words[br]in the cortex,

0:03:42.693,0:03:46.063
so they don’t share neurons[br]with others words.

0:03:46.063,0:03:49.073
Then, over multiple nights of sleep,

0:03:49.073,0:03:54.470
the new words gradually transfer over[br]and interweave with old ones.

0:03:54.470,0:03:57.990
Researchers think this gradual[br]acquisition process

0:03:57.990,0:04:01.354
helps avoid disrupting existing words.

0:04:01.354,0:04:02.774
So in the daytime,

0:04:02.774,0:04:07.304
unconscious activity generates explosions[br]of meaning as we chat away.

0:04:07.304,0:04:12.305
At night, we rest, but our brains are [br]busy integrating new knowledge

0:04:12.305,0:04:14.125
into the word network.

0:04:14.125,0:04:17.655
When we wake up, this process ensures [br]that we’re ready

0:04:17.596,0:04:20.696
for the ever-changing world of language.