1
00:00:00,560 --> 00:00:04,880
Speaker 1: He's night side with Dan Ray on WBS. He costins

2
00:00:04,960 --> 00:00:08,680
Speaker 1: me radio nice of being Weather says ten seven on

3
00:00:08,720 --> 00:00:09,080
Speaker 1: the Dot.

4
00:00:09,440 --> 00:00:13,920
Speaker 2: And I told you earlier that I I watch YouTube

5
00:00:13,920 --> 00:00:16,599
Speaker 2: a lot. It's there's a lot to learn and it's

6
00:00:16,600 --> 00:00:20,400
Speaker 2: fun to do. It's it's a great place. If I

7
00:00:20,440 --> 00:00:24,600
Speaker 2: had children, I would try to ease them over into

8
00:00:24,760 --> 00:00:27,720
Speaker 2: YouTube to learn stuff while I'm doing their screen time.

9
00:00:28,680 --> 00:00:32,000
Speaker 2: And one of the one of the sights I subscribed

10
00:00:32,040 --> 00:00:35,879
Speaker 2: to is called Rob Words. And this guy Rob Watts,

11
00:00:36,000 --> 00:00:38,640
Speaker 2: he's over and he lives in Berlin now, but he's

12
00:00:38,800 --> 00:00:44,680
Speaker 2: I think he's English. He he goes through like he

13
00:00:44,760 --> 00:00:50,120
Speaker 2: does segments on the meaning, the root meanings, the origins

14
00:00:50,159 --> 00:00:52,400
Speaker 2: of the meaning of the states in the United States,

15
00:00:53,440 --> 00:00:58,240
Speaker 2: and the complete history of the English language. I love

16
00:00:58,320 --> 00:01:03,240
Speaker 2: the language and I love it's idiosyncrasies. One day while

17
00:01:03,280 --> 00:01:10,119
Speaker 2: watching Rob, so you had a guest, a distinguished guest,

18
00:01:10,920 --> 00:01:16,679
Speaker 2: Mark Lieberman, the Christopher H. Brown Distinguished Professor of Linguistics

19
00:01:16,680 --> 00:01:20,520
Speaker 2: and Director of the Linguistic Data Consortium at the University

20
00:01:20,520 --> 00:01:24,839
Speaker 2: of Pennsylvania. And he's a heavy hitter, but he's fun too,

21
00:01:25,360 --> 00:01:28,760
Speaker 2: And we're very glad to have Mark with us to

22
00:01:28,840 --> 00:01:35,200
Speaker 2: talk about initially these things called egg cons and as

23
00:01:35,280 --> 00:01:38,680
Speaker 2: we as you find out what this is, I'll ask

24
00:01:38,720 --> 00:01:40,319
Speaker 2: you to get involved. But you're going to need to

25
00:01:40,360 --> 00:01:44,440
Speaker 2: know what an egg corn is before I can explain

26
00:01:44,480 --> 00:01:47,280
Speaker 2: how to get involved. So, Professor Lieberman, thank you for

27
00:01:47,319 --> 00:01:55,400
Speaker 2: being with us. So can you explain what an egg

28
00:01:55,520 --> 00:01:56,120
Speaker 2: corn is?

29
00:01:58,200 --> 00:02:02,600
Speaker 3: Well, this all started about years ago when I heard

30
00:02:02,640 --> 00:02:06,840
Speaker 3: about a case where a woman had complained to a

31
00:02:06,880 --> 00:02:10,799
Speaker 3: car dealership that her the hood of her car had

32
00:02:10,800 --> 00:02:14,960
Speaker 3: gotten dinged by what she called eggcorns falling on it

33
00:02:15,000 --> 00:02:17,840
Speaker 3: because she parked under an oak tree. So it was

34
00:02:17,880 --> 00:02:22,360
Speaker 3: obviously she was talking about acorns, but she thought sensibly

35
00:02:22,480 --> 00:02:26,720
Speaker 3: enough that they were actually eggcorns. It's a sensible mistake

36
00:02:27,200 --> 00:02:29,919
Speaker 3: because an egg corn kind of looks like an egg

37
00:02:30,280 --> 00:02:33,040
Speaker 3: and it's a kind of a seed, which is one

38
00:02:33,080 --> 00:02:34,079
Speaker 3: thing to corn means.

39
00:02:35,320 --> 00:02:39,320
Speaker 2: And so all her life she thought acorns were called

40
00:02:39,400 --> 00:02:42,519
Speaker 2: egg corns because it sounded that way, and they kind

41
00:02:42,520 --> 00:02:46,520
Speaker 2: of looked like corn on a fried egg or something

42
00:02:46,600 --> 00:02:46,959
Speaker 2: like that.

43
00:02:49,639 --> 00:02:53,280
Speaker 3: Well, or an egg in the shell, I mean, the

44
00:02:54,400 --> 00:02:57,280
Speaker 3: nut part of an acorn actually looks a little bit

45
00:02:57,400 --> 00:03:04,280
Speaker 3: like an eggshell. And then so the question was what

46
00:03:04,360 --> 00:03:06,960
Speaker 3: to call this, and we I did a little bit

47
00:03:07,000 --> 00:03:09,880
Speaker 3: of looking around on the web and discovered she's not

48
00:03:09,919 --> 00:03:11,760
Speaker 3: the only one who ever thought this. A bunch of

49
00:03:11,760 --> 00:03:14,560
Speaker 3: other people did as well, though it's not all that common.

50
00:03:15,440 --> 00:03:19,040
Speaker 3: And but the thing is, uh, it's obviously it's obviously

51
00:03:19,120 --> 00:03:24,600
Speaker 3: a mistake, although it's a reasonable mistake since sakehorn itself

52
00:03:24,639 --> 00:03:28,600
Speaker 3: doesn't mean anything in the contemporary language except the seed

53
00:03:28,639 --> 00:03:30,640
Speaker 3: of a well, seed of an oak tree.

54
00:03:33,040 --> 00:03:33,120
Speaker 4: Uh.

55
00:03:33,520 --> 00:03:37,360
Speaker 3: And my colleague Jeff Pullam suggested, well, you know, often

56
00:03:37,800 --> 00:03:42,240
Speaker 3: when something new comes up, we name it after the

57
00:03:42,480 --> 00:03:45,600
Speaker 3: example that brought it to our attention, So why not

58
00:03:45,720 --> 00:03:48,360
Speaker 3: call the call it? Call things like this an eggcorn?

59
00:03:49,120 --> 00:03:51,200
Speaker 3: And when you start looking around, there are lots and

60
00:03:51,280 --> 00:03:56,680
Speaker 3: lots of other examples of cases in which people uh

61
00:03:57,480 --> 00:04:02,240
Speaker 3: think they're a phrase or a word that's not the

62
00:04:02,560 --> 00:04:08,560
Speaker 3: quote unquote correct phraser word is what that phraser word

63
00:04:08,600 --> 00:04:12,640
Speaker 3: should be construed as in their internal lexicon, so to speak.

64
00:04:13,360 --> 00:04:16,680
Speaker 2: And Uh, this is where I'm want to ask the

65
00:04:17,680 --> 00:04:22,160
Speaker 2: listening audience to share any things that when they were

66
00:04:22,200 --> 00:04:26,320
Speaker 2: a kid or maybe way into life, that that they

67
00:04:26,320 --> 00:04:29,880
Speaker 2: didn't quite hear right and had been saying incorrectly until

68
00:04:29,920 --> 00:04:34,200
Speaker 2: someone corrected them, And that can include music, folks, musical lyrics.

69
00:04:34,560 --> 00:04:38,480
Speaker 2: For example, the big the big example is Jimi Hendrix.

70
00:04:39,000 --> 00:04:41,560
Speaker 2: Excuse me, while I kissed this guy, a lot of

71
00:04:41,560 --> 00:04:45,680
Speaker 2: people thought that's what he was saying. And everybody has

72
00:04:45,839 --> 00:04:48,920
Speaker 2: some word that for a while they were saying wrong.

73
00:04:49,040 --> 00:04:55,440
Speaker 2: For me, mine was murger. I thought murder was murger

74
00:04:55,600 --> 00:04:56,920
Speaker 2: m U G E R.

75
00:04:57,680 --> 00:04:58,160
Speaker 1: And I don't know.

76
00:04:58,200 --> 00:05:01,080
Speaker 2: I must have been like, you know, I'd hear it

77
00:05:01,080 --> 00:05:05,400
Speaker 2: on Perry Mason here on trial for Murger one. And

78
00:05:05,480 --> 00:05:10,080
Speaker 2: it took forever a little too long for me to

79
00:05:10,080 --> 00:05:12,520
Speaker 2: to find out that that was not the case. Now,

80
00:05:12,560 --> 00:05:15,040
Speaker 2: what are some other fun examples of these egg.

81
00:05:14,960 --> 00:05:23,240
Speaker 3: Cons well, excuse me. Famous ones are things like Alzheimer's

82
00:05:23,279 --> 00:05:25,120
Speaker 3: disease for Alzheimer's.

83
00:05:24,560 --> 00:05:28,160
Speaker 1: Disease, Oldzheimer's disease, Okay.

84
00:05:27,960 --> 00:05:32,320
Speaker 3: Yeah, in one foul swoop instead of in one fell

85
00:05:32,400 --> 00:05:37,839
Speaker 3: swoop for all intensive purposes instead of for all intents

86
00:05:37,880 --> 00:05:38,600
Speaker 3: and purposes.

87
00:05:39,560 --> 00:05:40,520
Speaker 1: Yeah, And so.

88
00:05:40,920 --> 00:05:44,039
Speaker 2: Back back up to fell swoop, I thought it was

89
00:05:44,920 --> 00:05:48,000
Speaker 2: I don't know whether it's felled or fell.

90
00:05:48,480 --> 00:05:52,200
Speaker 1: And what is the origin of that phrase that idiom?

91
00:05:52,240 --> 00:05:58,599
Speaker 3: I guess, well, it's typical of eggcorns that the true

92
00:05:58,839 --> 00:06:03,400
Speaker 3: version of the phrase either never meant anything or it

93
00:06:03,440 --> 00:06:07,680
Speaker 3: doesn't mean anything anymore in the contemporary language. So feil

94
00:06:07,920 --> 00:06:13,320
Speaker 3: is basically an obsolete word meaning bad or evil.

95
00:06:14,960 --> 00:06:17,880
Speaker 1: Okay, so what a all right?

96
00:06:18,040 --> 00:06:18,360
Speaker 3: All right?

97
00:06:18,400 --> 00:06:18,800
Speaker 1: I get it?

98
00:06:18,880 --> 00:06:22,280
Speaker 3: Yeah, where what.

99
00:06:22,279 --> 00:06:24,400
Speaker 1: Were some of the others again? Because I don't want

100
00:06:24,400 --> 00:06:24,760
Speaker 1: to just.

101
00:06:25,680 --> 00:06:28,719
Speaker 3: Old Teimer's disease for Alzheimer's disease.

102
00:06:28,839 --> 00:06:32,800
Speaker 2: How prevalent is it that people say old Teimer's disease.

103
00:06:35,200 --> 00:06:37,919
Speaker 3: I'm not sure that I've ever heard it said, but

104
00:06:38,120 --> 00:06:41,320
Speaker 3: if you google it, you'll find lots and lots of

105
00:06:41,360 --> 00:06:44,520
Speaker 3: examples from common sections and things like that, where people

106
00:06:44,560 --> 00:06:46,239
Speaker 3: seem to think that that's the phrase.

107
00:06:47,400 --> 00:06:51,000
Speaker 1: Okay, and you mentioned a couple others as well.

108
00:06:52,520 --> 00:06:58,640
Speaker 3: Well. Another example that's pretty common is ex patriot. The

109
00:06:58,720 --> 00:07:02,280
Speaker 3: word x patriot meanings somebody who lives in a country

110
00:07:02,320 --> 00:07:08,440
Speaker 3: they weren't born in. Instead of uh, rendering that as

111
00:07:08,600 --> 00:07:11,560
Speaker 3: ex patriot as a former patriot.

112
00:07:12,800 --> 00:07:14,560
Speaker 1: So people would just spell it with the letter X.

113
00:07:14,600 --> 00:07:18,600
Speaker 3: You mean, no, they would spell it as ex hyphen

114
00:07:18,720 --> 00:07:21,880
Speaker 3: patriot p A t r I O T, rather than

115
00:07:21,920 --> 00:07:24,480
Speaker 3: e x p A t r I A t e.

116
00:07:26,040 --> 00:07:26,080
Speaker 4: Y.

117
00:07:26,600 --> 00:07:31,080
Speaker 2: He had been x patrioted as opposed to repatriated et cetera.

118
00:07:32,160 --> 00:07:43,960
Speaker 3: Yeah. Another example is uh, old wives tail for old

119
00:07:44,040 --> 00:07:49,200
Speaker 3: wives tale. That's a case where both makes sense. Old

120
00:07:49,240 --> 00:07:52,840
Speaker 3: wives tale makes sense, old wise tail makes sense.

121
00:07:55,720 --> 00:07:56,400
Speaker 1: That's interesting.

122
00:07:57,040 --> 00:08:00,400
Speaker 2: I've never heard that either old wives tail. Everyone has.

123
00:08:00,720 --> 00:08:02,640
Speaker 2: Every one of you out there must have had one

124
00:08:02,640 --> 00:08:05,720
Speaker 2: of these errors. Now, don't be shy. Let us know

125
00:08:06,840 --> 00:08:09,480
Speaker 2: something you thought was said one way was actually said

126
00:08:09,480 --> 00:08:13,560
Speaker 2: another way. I shared my mistake, now you should share yours.

127
00:08:13,600 --> 00:08:16,840
Speaker 2: Let's continue with Professor Lieberman in a moment on WBZ.

128
00:08:18,440 --> 00:08:22,720
Speaker 5: You're on Night Side with Dan Ray on WBZ, Boston's

129
00:08:22,760 --> 00:08:23,400
Speaker 5: news radio.

130
00:08:23,920 --> 00:08:27,320
Speaker 2: Let's continue with the professor Mark Lieberman, a linguistics A

131
00:08:27,440 --> 00:08:30,280
Speaker 2: distinguished linguistics professor.

132
00:08:30,600 --> 00:08:33,160
Speaker 1: By the way, a professor.

133
00:08:32,920 --> 00:08:36,079
Speaker 2: Can you give me a handle on what a linguistics

134
00:08:36,120 --> 00:08:38,600
Speaker 2: professor actually teaches? What kind of courses you might teach us?

135
00:08:38,640 --> 00:08:43,120
Speaker 2: I never I don't believe I took a linguistics course,

136
00:08:43,160 --> 00:08:46,280
Speaker 2: even though I was a communications major. What's our class

137
00:08:46,280 --> 00:08:46,760
Speaker 2: of yours?

138
00:08:46,880 --> 00:08:51,280
Speaker 3: Like? Well, I teach a bunch of different kinds of courses.

139
00:08:51,320 --> 00:08:55,280
Speaker 3: I'm actually actually have positions in not only the Department

140
00:08:55,320 --> 00:08:58,439
Speaker 3: of Linguistics, but also the Department of Computer Information Science

141
00:08:59,120 --> 00:09:08,360
Speaker 3: and the Psychology Graduate Group. And so On the linguistics side,

142
00:09:08,440 --> 00:09:15,360
Speaker 3: I teach the general undergraduate introduction without any corequisites. I

143
00:09:15,480 --> 00:09:21,240
Speaker 3: teach graduate courses in phonetics, that is, the sound patterns

144
00:09:21,240 --> 00:09:26,720
Speaker 3: of language. For On the computer science side, I have

145
00:09:26,840 --> 00:09:34,800
Speaker 3: taught courses on all right, what's basically digital signal processing,

146
00:09:34,920 --> 00:09:37,280
Speaker 3: how to find things in signals?

147
00:09:37,559 --> 00:09:39,000
Speaker 1: You know what, let me like, let me go eat,

148
00:09:39,080 --> 00:09:43,600
Speaker 1: and be more basic, what is linguistics? What are linguistics?

149
00:09:43,679 --> 00:09:44,440
Speaker 1: How do you define that?

150
00:09:46,480 --> 00:09:48,240
Speaker 3: It's the study of speech and language?

151
00:09:48,600 --> 00:09:51,360
Speaker 1: Okay, the study of speech and language? And why is

152
00:09:51,400 --> 00:09:52,040
Speaker 1: that important?

153
00:09:52,240 --> 00:09:54,880
Speaker 2: And I suppose it's mostly important when you do it

154
00:09:54,920 --> 00:10:00,840
Speaker 2: comparatively correct, when you compare this this culture speech with

155
00:10:01,080 --> 00:10:02,760
Speaker 2: their customs and things like that.

156
00:10:04,440 --> 00:10:08,520
Speaker 3: Well, that's that's fun and interesting and often worthwhile and valuable.

157
00:10:08,640 --> 00:10:13,880
Speaker 3: But uh, there's a full computational side of it, which is,

158
00:10:14,240 --> 00:10:18,240
Speaker 3: you know, everything that Google does and Apple does and

159
00:10:18,400 --> 00:10:23,000
Speaker 3: IBM does and so on with what's now called AI

160
00:10:23,120 --> 00:10:26,920
Speaker 3: and large language models that came out of computational linguistics

161
00:10:27,559 --> 00:10:32,319
Speaker 3: over the last few decades. There's also there's also applications

162
00:10:32,360 --> 00:10:36,120
Speaker 3: in teaching when you think about how to teach kids

163
00:10:36,120 --> 00:10:39,439
Speaker 3: to read, or how to deal with language development. Issues.

164
00:10:40,960 --> 00:10:47,920
Speaker 3: There's applications in political science figuring out what's going on

165
00:10:48,120 --> 00:10:54,240
Speaker 3: when people talk about politics, what they really need, how

166
00:10:54,280 --> 00:10:57,040
Speaker 3: you compare different ways of talking about politics.

167
00:10:57,960 --> 00:11:01,200
Speaker 2: I think there's interesting classes. There's a lot going on.

168
00:11:01,240 --> 00:11:03,280
Speaker 2: There is at the very root of much of what

169
00:11:03,320 --> 00:11:05,559
Speaker 2: we do, uh you know actually.

170
00:11:05,280 --> 00:11:08,240
Speaker 3: Well, and in the law. A lot of what goes

171
00:11:08,280 --> 00:11:11,000
Speaker 3: on in law is trying to figure out what laws mean,

172
00:11:11,080 --> 00:11:17,000
Speaker 3: what contracts mean, what the Constitution means, and so on.

173
00:11:18,160 --> 00:11:21,840
Speaker 1: So would the worrying of the Second Amendment be a linguistics.

174
00:11:21,360 --> 00:11:25,000
Speaker 3: Issue, absolutely, absolutely, for.

175
00:11:25,920 --> 00:11:27,000
Speaker 1: Saying that a.

176
00:11:28,760 --> 00:11:33,520
Speaker 2: Standing militia being necessary, you have the right to bear arms,

177
00:11:33,559 --> 00:11:36,720
Speaker 2: but what if outstandinglious militia is no longer necessary.

178
00:11:36,760 --> 00:11:39,320
Speaker 1: It seems that they put that caveat in there for

179
00:11:39,360 --> 00:11:39,880
Speaker 1: a purpose.

180
00:11:39,960 --> 00:11:44,199
Speaker 2: Well, and it's necessary for you to have weapons is

181
00:11:44,760 --> 00:11:48,760
Speaker 2: because we have we need a militia, which will seem

182
00:11:48,840 --> 00:11:49,319
Speaker 2: to say man.

183
00:11:49,559 --> 00:11:52,280
Speaker 3: Of course that has changed, and there's been a lot

184
00:11:52,320 --> 00:11:56,080
Speaker 3: of controversy over the over the decades and centuries actually

185
00:11:56,679 --> 00:12:01,120
Speaker 3: about what that preliminary phrase in the Second Amendment means.

186
00:12:01,320 --> 00:12:03,760
Speaker 3: But there's also the question of what bare arms means.

187
00:12:05,200 --> 00:12:08,320
Speaker 3: And actually in the eighteenth century at the time that

188
00:12:08,320 --> 00:12:13,600
Speaker 3: that amendment was written. There's a good argument that bare

189
00:12:13,720 --> 00:12:15,760
Speaker 3: arms meant to serve in the military.

190
00:12:17,280 --> 00:12:19,720
Speaker 2: So a lot of linguistics, you have to pay attention

191
00:12:19,760 --> 00:12:24,800
Speaker 2: to the evolution of meaning and make sure that the meaning.

192
00:12:26,520 --> 00:12:29,560
Speaker 3: How words are used, not just look them up in

193
00:12:29,559 --> 00:12:31,439
Speaker 3: the dictionary and look at how they're actually used.

194
00:12:31,920 --> 00:12:35,320
Speaker 2: Well, if if you had taught at my university, I

195
00:12:35,360 --> 00:12:38,960
Speaker 2: definitely would have taken your course more than one. Probably

196
00:12:39,120 --> 00:12:41,120
Speaker 2: we have Neil in Watertown and wants to join in.

197
00:12:41,160 --> 00:12:45,920
Speaker 4: Hi Neil badly, a little professor. The reason I called

198
00:12:46,080 --> 00:12:50,520
Speaker 4: was I've heard that one fell swoop the first time

199
00:12:50,679 --> 00:12:52,760
Speaker 4: was ever used, at least in print, at least the

200
00:12:52,800 --> 00:12:57,160
Speaker 4: first time that is known about it was And Macduff

201
00:12:57,200 --> 00:12:59,679
Speaker 4: said that in Macbeth's when he got the bad news

202
00:12:59,679 --> 00:13:02,920
Speaker 4: from Ross that all his children had been killed. I

203
00:13:02,920 --> 00:13:05,160
Speaker 4: think he said something like all my chickens and one

204
00:13:05,200 --> 00:13:11,680
Speaker 4: fell swoop. And if I may just to other uses earlier,

205
00:13:11,800 --> 00:13:14,840
Speaker 4: Lady Macbeth's trying to psych herself up to be evil,

206
00:13:14,840 --> 00:13:18,839
Speaker 4: and she says, let no compunctious visitings of nature shake

207
00:13:18,960 --> 00:13:24,400
Speaker 4: my fell purpose and son seventy four, it's but be

208
00:13:24,600 --> 00:13:28,400
Speaker 4: contented when that fell arrests without all bails. She'll carry

209
00:13:28,480 --> 00:13:32,720
Speaker 4: me away. So it's like a personified policeman comes and

210
00:13:32,800 --> 00:13:37,160
Speaker 4: takes you away, meaning death. And it's it's a cousin

211
00:13:37,200 --> 00:13:38,760
Speaker 4: of felon. I think it comes from.

212
00:13:38,640 --> 00:13:41,040
Speaker 1: The same place really fell and fell.

213
00:13:42,000 --> 00:13:43,400
Speaker 4: Well, I haven't looked it up in a while, but

214
00:13:43,440 --> 00:13:45,800
Speaker 4: that's in my head. I apologize if I'm wrong.

215
00:13:45,840 --> 00:13:50,760
Speaker 1: So means cruel, terrible, fear, and fierce.

216
00:13:51,440 --> 00:13:55,360
Speaker 2: So probably a felon a person who has created it,

217
00:13:55,640 --> 00:14:01,199
Speaker 2: who has committed to particularly fearsome crime. I guess that, Neil,

218
00:14:01,880 --> 00:14:06,560
Speaker 2: you bring something up. I believe that the Professor will

219
00:14:06,559 --> 00:14:10,280
Speaker 2: concur that so so many idioms and phrases that we

220
00:14:10,400 --> 00:14:12,960
Speaker 2: use come from Shakespeare. That was really the mother load, right?

221
00:14:13,240 --> 00:14:14,480
Speaker 1: Nothing on?

222
00:14:14,679 --> 00:14:18,920
Speaker 4: Ye, the word useful? I think those so, Professor.

223
00:14:18,520 --> 00:14:21,000
Speaker 1: Thanks a lot, Neil appreciate it. How about that? Is

224
00:14:22,120 --> 00:14:24,520
Speaker 1: is it true that Shakespeare.

225
00:14:24,880 --> 00:14:30,400
Speaker 2: Has is responsible for many of the little phrases we

226
00:14:30,480 --> 00:14:31,720
Speaker 2: use and don't even realize it.

227
00:14:33,560 --> 00:14:37,320
Speaker 3: We believe there are more phrases with an origin in

228
00:14:37,400 --> 00:14:41,680
Speaker 3: Shakespeare than for any other English language writer. Of course,

229
00:14:41,680 --> 00:14:45,800
Speaker 3: there are plenty of phrases that were not first used

230
00:14:45,800 --> 00:14:48,280
Speaker 3: by a famous writer, or were first used by some

231
00:14:48,440 --> 00:14:51,760
Speaker 3: other famous writer. But still Shakespeare's definitely up there.

232
00:14:51,880 --> 00:14:56,160
Speaker 1: What are some of the others Shakespeare might have given us?

233
00:14:58,360 --> 00:15:02,960
Speaker 3: Uh? Gee, that is something I can immediately bring the mind.

234
00:15:03,040 --> 00:15:08,360
Speaker 3: But I bet that if I asked asked Google for

235
00:15:11,160 --> 00:15:14,640
Speaker 3: phrases that originate in Shakespeare will get a long list.

236
00:15:15,440 --> 00:15:18,880
Speaker 2: Okay, So let's get back to the phrase egg corn,

237
00:15:19,520 --> 00:15:25,120
Speaker 2: and there are related, for the better lack of a

238
00:15:25,120 --> 00:15:28,640
Speaker 2: better word, things, there are things types of phrases related

239
00:15:28,680 --> 00:15:31,800
Speaker 2: to egg corn. They have different titles. We talked about

240
00:15:31,800 --> 00:15:34,960
Speaker 2: those earlier. Can you mona Green is one of them?

241
00:15:35,080 --> 00:15:37,320
Speaker 2: Can you explain what that is and how it's different.

242
00:15:38,800 --> 00:15:45,520
Speaker 3: Manda Green? Well, that arose from an old Scottish ballad

243
00:15:47,320 --> 00:15:51,840
Speaker 3: that talked about the fact that they killed the Earl

244
00:15:51,880 --> 00:15:55,960
Speaker 3: of Moray and laid him on the green. But a

245
00:15:56,000 --> 00:16:00,920
Speaker 3: woman hearing that song sung interpreted that that they killed

246
00:16:00,920 --> 00:16:03,880
Speaker 3: the Earl of Moray and Lady Manda Green.

247
00:16:05,120 --> 00:16:08,400
Speaker 2: So instead of laid him on the green, it was

248
00:16:08,520 --> 00:16:10,880
Speaker 2: it was heard Lady Manda Green.

249
00:16:11,840 --> 00:16:16,680
Speaker 3: Right, and so she wrote about that, and so that became.

250
00:16:16,960 --> 00:16:21,800
Speaker 3: Since that kind of mishearing of lyrics is extremely common,

251
00:16:22,240 --> 00:16:25,440
Speaker 3: it became the way of talking about examples of that

252
00:16:25,560 --> 00:16:28,640
Speaker 3: kind in general. It's examples of the kind are manda greens.

253
00:16:28,800 --> 00:16:33,280
Speaker 2: What are the policies of dictionaries about taking these misheard

254
00:16:33,280 --> 00:16:36,520
Speaker 2: phrases and making them into official words?

255
00:16:36,560 --> 00:16:37,320
Speaker 1: Does that happen?

256
00:16:39,640 --> 00:16:47,560
Speaker 3: Well, there are many words that everybody uses that started

257
00:16:47,560 --> 00:16:52,640
Speaker 3: out as as a mishearing of this kind. The one

258
00:16:52,720 --> 00:16:57,760
Speaker 3: famous one is Jerusalem artichoke, which is not all that

259
00:16:57,840 --> 00:16:59,720
Speaker 3: well known, but probably a lot of people around the

260
00:16:59,760 --> 00:17:03,680
Speaker 3: boss scenario have eaten Jerusalem artichokes. It's kind of a

261
00:17:03,760 --> 00:17:08,200
Speaker 3: root vegetable and it's easy. It's native to North America.

262
00:17:08,280 --> 00:17:12,320
Speaker 3: It's easy to grow, and it's the it's two words

263
00:17:12,400 --> 00:17:16,480
Speaker 3: or roots of a sunflower like plant. It's kind of

264
00:17:16,480 --> 00:17:20,119
Speaker 3: a pretty planet. But because it's a sunflower like planet,

265
00:17:20,119 --> 00:17:24,960
Speaker 3: it was originally called an Italian giusole, which means turned

266
00:17:25,000 --> 00:17:29,320
Speaker 3: to the sun in Italian, but English speakers misheard that

267
00:17:29,440 --> 00:17:32,320
Speaker 3: as Jerusalem has nothing to do with Jerusalem, it was

268
00:17:32,560 --> 00:17:35,760
Speaker 3: never been seen in Jerusalem and so on, but that

269
00:17:36,080 --> 00:17:37,400
Speaker 3: they thought it was Jerusalem.

270
00:17:38,640 --> 00:17:40,440
Speaker 1: So there's there are a lot.

271
00:17:40,240 --> 00:17:43,760
Speaker 2: Of musical manda greens, and I want to just share

272
00:17:44,240 --> 00:17:51,240
Speaker 2: they're funny. So here are some classic examples from Cretan's

273
00:17:51,280 --> 00:17:55,160
Speaker 2: Clear Clear Water Revival CCR.

274
00:17:55,400 --> 00:17:56,480
Speaker 1: Let's go with CCR.

275
00:17:57,600 --> 00:18:03,480
Speaker 2: There's the song bad Moon Rising, and some folks think

276
00:18:03,680 --> 00:18:06,879
Speaker 2: or thought they were saying, instead of saying there's a

277
00:18:06,920 --> 00:18:11,040
Speaker 2: bad moon on the rise, they were humming along as

278
00:18:11,040 --> 00:18:14,359
Speaker 2: they were vacuuming, there's a bathroom on the right, is

279
00:18:14,400 --> 00:18:15,200
Speaker 2: what they thought.

280
00:18:14,960 --> 00:18:18,520
Speaker 3: This, Yes, there's a bathroom on the right.

281
00:18:19,040 --> 00:18:23,080
Speaker 2: And now anytime anybody hears that CCR song, they will

282
00:18:23,080 --> 00:18:27,520
Speaker 2: be saying that. Okay, we talked about the Jimmy Hendrick's one,

283
00:18:27,520 --> 00:18:31,840
Speaker 2: and that is that's completely understandable because it certainly sounds

284
00:18:31,880 --> 00:18:35,960
Speaker 2: exactly like it. It's not excuse me while I kiss

285
00:18:36,040 --> 00:18:39,520
Speaker 2: this It's not they think it's excuse me while I

286
00:18:39,600 --> 00:18:42,760
Speaker 2: kiss this guy, not excuse me while I kiss the sky.

287
00:18:44,040 --> 00:18:48,680
Speaker 2: This is an interesting one. Elton John song Tiny Dancer.

288
00:18:51,000 --> 00:18:55,720
Speaker 2: Instead of hold Me Closer, Tiny Dancer, some folks, and

289
00:18:55,760 --> 00:18:58,400
Speaker 2: I wish I had some numbers on how many folks.

290
00:18:59,280 --> 00:19:03,080
Speaker 2: Instead of hold Me Closer, Tiny Dancer, it's hold Me

291
00:19:03,240 --> 00:19:10,600
Speaker 2: Closer Tony Danza. There's something really inherently funny about these

292
00:19:10,960 --> 00:19:13,119
Speaker 2: hold Me Closer Tony dance.

293
00:19:14,200 --> 00:19:14,560
Speaker 1: Okay.

294
00:19:15,920 --> 00:19:22,439
Speaker 2: Next we have the Beatles song, and this is almost

295
00:19:22,440 --> 00:19:26,720
Speaker 2: too weird to believe. From the Beatles song Lucy in

296
00:19:26,760 --> 00:19:32,480
Speaker 2: the Sky with Diamonds instead of the girl with Kaleidoscope eyes.

297
00:19:33,440 --> 00:19:35,679
Speaker 1: Some and again, i'd like to see the number.

298
00:19:36,280 --> 00:19:40,200
Speaker 2: Some think it's said the girl with kaliis goes by

299
00:19:41,600 --> 00:19:45,280
Speaker 2: instead of the girl with kaleidoscope eyes, and Lor knows.

300
00:19:45,320 --> 00:19:47,920
Speaker 2: I don't want to laugh at that, but it is interesting.

301
00:19:49,600 --> 00:19:50,320
Speaker 1: So there are.

302
00:19:50,200 --> 00:19:54,240
Speaker 3: Those, so yeah, watches on Yeah, my favorite. My favorite

303
00:19:54,359 --> 00:19:58,439
Speaker 3: is in the Taylor Swift song the Long List of

304
00:19:58,520 --> 00:20:01,440
Speaker 3: ex lovers as all the lonely Starbucks lovers.

305
00:20:02,720 --> 00:20:07,680
Speaker 2: Yes, that's listed in the modern pop examples. Let's see

306
00:20:08,119 --> 00:20:11,959
Speaker 2: these are modern. I hope I get these right. So

307
00:20:12,080 --> 00:20:18,439
Speaker 2: instead of in the Pearl Jam song, Jeremy instead of

308
00:20:18,520 --> 00:20:23,200
Speaker 2: Jeremy spoken class today, it's Jeremy smoking grass today.

309
00:20:23,800 --> 00:20:26,200
Speaker 1: That's all I can say. And you know, there is a.

310
00:20:27,080 --> 00:20:33,119
Speaker 2: There's currently a trend in music production, at least some

311
00:20:33,440 --> 00:20:36,880
Speaker 2: genres of music production, to bury the vocals a little

312
00:20:36,920 --> 00:20:39,800
Speaker 2: bit more than before. It used to be that the

313
00:20:39,880 --> 00:20:43,359
Speaker 2: music was just kind of this little ice skating rank

314
00:20:43,359 --> 00:20:49,040
Speaker 2: for the vocals to skate around on, very clearly distinct,

315
00:20:49,560 --> 00:20:54,320
Speaker 2: but now many times the vocals are treated more as instruments.

316
00:20:54,760 --> 00:21:00,639
Speaker 2: They do have meaning, but a lot of the essence

317
00:21:00,680 --> 00:21:04,520
Speaker 2: of the vocal is more instrument like than the meaning

318
00:21:04,520 --> 00:21:07,760
Speaker 2: of the words, so it gets easier to make mistakes.

319
00:21:08,400 --> 00:21:10,399
Speaker 2: Let's do another one here, and we do have a

320
00:21:10,520 --> 00:21:12,640
Speaker 2: David in San Francisco to get you after the break

321
00:21:13,359 --> 00:21:18,080
Speaker 2: from Queen, we will rock you. Instead of kicking your

322
00:21:18,200 --> 00:21:21,960
Speaker 2: can all over the police, it's kicking your cat all

323
00:21:22,000 --> 00:21:24,680
Speaker 2: over the place. How can you go through life taking that?

324
00:21:24,960 --> 00:21:30,440
Speaker 1: Anyone? Come on? And the fun thing is what happens someday.

325
00:21:30,800 --> 00:21:34,600
Speaker 2: You're in a car with your friend and you're singing

326
00:21:34,600 --> 00:21:38,640
Speaker 2: along to the radio, and you sing along kicking your

327
00:21:38,680 --> 00:21:42,119
Speaker 2: cat all over the place, and the friend looks at you,

328
00:21:42,240 --> 00:21:47,920
Speaker 2: like what what And you say, yeah, kicking your can,

329
00:21:48,280 --> 00:21:51,959
Speaker 2: kicking your cat all over the place. See, you're super embarrassed.

330
00:21:52,119 --> 00:21:55,879
Speaker 2: I can't think of any of these that I have done,

331
00:21:55,880 --> 00:22:00,000
Speaker 2: but I'm sure there is some. Let's see a more another,

332
00:22:00,000 --> 00:22:06,080
Speaker 2: the more recent one instead of she's an easy lover.

333
00:22:07,640 --> 00:22:08,440
Speaker 1: And I have to doubt.

334
00:22:08,480 --> 00:22:13,000
Speaker 2: I don't know if I believe this. She's an easy lama. No,

335
00:22:13,480 --> 00:22:15,840
Speaker 2: I mean, come on, she's an easy lama. I don't

336
00:22:15,880 --> 00:22:16,880
Speaker 2: know if he sings.

337
00:22:16,560 --> 00:22:19,840
Speaker 3: It does not make sense, but you never know.

338
00:22:21,359 --> 00:22:24,000
Speaker 2: And here's I'm reading these now and I don't know

339
00:22:24,000 --> 00:22:25,879
Speaker 2: what's coming up. I did not proofread these. It's kind

340
00:22:25,880 --> 00:22:33,200
Speaker 2: of dangerous America America, okay, for America America. God shed

341
00:22:33,320 --> 00:22:39,360
Speaker 2: his grace on the from America the Beautiful. Some folks

342
00:22:40,359 --> 00:22:50,000
Speaker 2: think it America America. God is chef Boyard. Whether that's

343
00:22:50,040 --> 00:22:55,400
Speaker 2: true or not, whether a significant section of the population

344
00:22:55,520 --> 00:22:59,160
Speaker 2: thought that, it is kind of funny. I won more

345
00:22:59,560 --> 00:23:03,880
Speaker 2: before the right. This is from three doc Knight, Joy

346
00:23:03,920 --> 00:23:08,080
Speaker 2: to the world. The line is joy to the fishes

347
00:23:08,119 --> 00:23:10,919
Speaker 2: in the deep blue sea, Joy to you and me.

348
00:23:11,440 --> 00:23:15,280
Speaker 2: What they heard was joy to the visions.

349
00:23:15,000 --> 00:23:18,640
Speaker 1: That people see. That's not very funny, but there you go.

350
00:23:19,520 --> 00:23:22,440
Speaker 3: So, uh, it's a reasonable misunderstanding.

351
00:23:23,000 --> 00:23:25,880
Speaker 1: I love those. It'd be fun to make a song

352
00:23:26,119 --> 00:23:28,800
Speaker 1: out of only those.

353
00:23:29,080 --> 00:23:31,240
Speaker 2: So we have Actually David and San Francisco wants to

354
00:23:31,280 --> 00:23:35,920
Speaker 2: speak with us and perhaps someone else after this break,

355
00:23:35,920 --> 00:23:37,760
Speaker 2: and thanks for being with us, Professor more in a

356
00:23:37,760 --> 00:23:39,280
Speaker 2: moment on WBZ.

357
00:23:41,920 --> 00:23:42,960
Speaker 5: It's Night Side with.

358
00:23:45,359 --> 00:23:48,320
Speaker 1: Boston's news Radio. That's correct.

359
00:23:48,480 --> 00:23:52,520
Speaker 2: We are talking linguistics. I guess we're having fun with words.

360
00:23:52,520 --> 00:23:55,360
Speaker 2: That's a better way to say it. I'm an entertainer.

361
00:23:55,400 --> 00:23:57,720
Speaker 2: I want to have fun and I like words. And

362
00:23:57,760 --> 00:24:00,960
Speaker 2: when I can meld the two, I like to I

363
00:24:01,119 --> 00:24:05,960
Speaker 2: like to do it. The uh, professor, let's talk to

364
00:24:06,040 --> 00:24:08,760
Speaker 2: David in San Francisco. David in San Francisco, how do

365
00:24:08,800 --> 00:24:09,000
Speaker 2: you do?

366
00:24:09,119 --> 00:24:09,439
Speaker 3: David?

367
00:24:09,880 --> 00:24:12,160
Speaker 1: Say hello to Professor Mark.

368
00:24:13,520 --> 00:24:16,840
Speaker 6: Bradley, our professor are Let me ask you. Do you

369
00:24:16,880 --> 00:24:23,160
Speaker 6: know where the term drag, as in drag queen comes from?

370
00:24:24,000 --> 00:24:27,520
Speaker 3: The word comes from? Yes?

371
00:24:27,560 --> 00:24:31,800
Speaker 2: The question is where does the word drag from drag

372
00:24:31,880 --> 00:24:34,040
Speaker 2: queen come from. That is the question posed by David

373
00:24:34,040 --> 00:24:35,080
Speaker 2: from San Francisco.

374
00:24:35,600 --> 00:24:38,119
Speaker 6: Yeah, I told you, Bradley.

375
00:24:38,640 --> 00:24:38,919
Speaker 3: I know.

376
00:24:39,160 --> 00:24:45,399
Speaker 2: Let's see if the professor can answer. So it's just

377
00:24:45,440 --> 00:24:49,080
Speaker 2: a guess, nothing, nothing fancy. You're just gonna gotta make

378
00:24:49,080 --> 00:24:52,399
Speaker 2: a wild guess. It's not academic and it's not really

379
00:24:52,760 --> 00:24:54,040
Speaker 2: there's no rhyme or reason.

380
00:24:54,040 --> 00:24:55,160
Speaker 1: It's just a guess.

381
00:24:57,640 --> 00:25:05,600
Speaker 3: Well, I can I bet that? Uh, the Internet can

382
00:25:05,640 --> 00:25:10,160
Speaker 3: tell us, But can tell us, David tell the professor.

383
00:25:12,080 --> 00:25:16,719
Speaker 6: Professor comes from Shakespeare. Back when Shakespeare was writing his stories,

384
00:25:17,320 --> 00:25:23,800
Speaker 6: all the actors were men. Character in the story, he

385
00:25:23,840 --> 00:25:27,679
Speaker 6: would write the word drag next to a character and

386
00:25:27,960 --> 00:25:30,159
Speaker 6: the drag was dressed as girl.

387
00:25:32,680 --> 00:25:33,320
Speaker 1: So there you go.

388
00:25:34,800 --> 00:25:38,840
Speaker 3: It's certainly, that's certainly what it means. I don't recall

389
00:25:39,320 --> 00:25:43,719
Speaker 3: having seen it from that long ago, but you might

390
00:25:43,840 --> 00:25:44,640
Speaker 3: very well be right.

391
00:25:50,080 --> 00:25:52,199
Speaker 6: I've heard that studying history. I learned that.

392
00:25:56,480 --> 00:25:59,000
Speaker 2: Well, there you go, you win, you win there, David.

393
00:25:59,080 --> 00:26:02,960
Speaker 2: That's good, Thank you very much, Andrew and Wooster. Hello,

394
00:26:03,000 --> 00:26:05,520
Speaker 2: you're on WBZ with Professor Mark Lieberman.

395
00:26:06,359 --> 00:26:10,960
Speaker 5: Hi, Bradley. I used to talk to you a while back.

396
00:26:12,000 --> 00:26:16,160
Speaker 5: I lived in Spencer and I'm the Trader Joe's guy.

397
00:26:17,000 --> 00:26:18,040
Speaker 1: Oh wow, how are you?

398
00:26:19,119 --> 00:26:19,560
Speaker 5: How are you?

399
00:26:19,680 --> 00:26:21,280
Speaker 1: I'm great Trader Joe's guy.

400
00:26:24,320 --> 00:26:29,399
Speaker 5: They The one that always gets me is keen wah. Okay,

401
00:26:29,480 --> 00:26:34,920
Speaker 5: Customers come in and say keno or keno keno keen wah.

402
00:26:35,880 --> 00:26:39,040
Speaker 1: So so people don't know how to say keene wa.

403
00:26:39,880 --> 00:26:43,320
Speaker 4: Yeah, that's interesting as well.

404
00:26:43,840 --> 00:26:44,520
Speaker 1: And how are you been?

405
00:26:44,720 --> 00:26:45,040
Speaker 3: Andrew?

406
00:26:45,320 --> 00:26:46,800
Speaker 1: You're still with the Trader Joe's.

407
00:26:47,520 --> 00:26:50,160
Speaker 5: Yeah, I'm still Trader Joe is twenty years Wow.

408
00:26:51,040 --> 00:26:53,000
Speaker 1: Good for you and thanks for joining in with us.

409
00:26:53,000 --> 00:26:54,240
Speaker 1: That was a good one. I appreciate it.

410
00:26:55,520 --> 00:27:00,920
Speaker 2: So, professor, let me see, oh now you you I

411
00:27:01,000 --> 00:27:05,399
Speaker 2: don't want to overlook your wheelhouse, professor, and part of

412
00:27:05,440 --> 00:27:08,280
Speaker 2: your wheelhouse is how linguistics.

413
00:27:09,440 --> 00:27:15,960
Speaker 1: It's integral to AI. Can you flesh that out for

414
00:27:16,080 --> 00:27:22,719
Speaker 1: us some more and make us understand that better? Right now,

415
00:27:22,760 --> 00:27:27,440
Speaker 1: it's a pretty amorphous feeling I have about her connection.

416
00:27:27,600 --> 00:27:32,399
Speaker 3: I'd rather amorphous term. It basically means, you know, complicated

417
00:27:32,440 --> 00:27:37,280
Speaker 3: computer program that answers what seemed like hard questions. But

418
00:27:37,840 --> 00:27:42,359
Speaker 3: one one of the most important particulars in that area

419
00:27:42,920 --> 00:27:44,760
Speaker 3: is what are called large language models.

420
00:27:47,520 --> 00:27:48,040
Speaker 1: What does it mean?

421
00:27:49,359 --> 00:27:52,040
Speaker 3: Well, let's leaves a large part out for the moment

422
00:27:52,160 --> 00:27:56,800
Speaker 3: and get onto language model. And the very first language

423
00:27:56,880 --> 00:28:02,320
Speaker 3: model was developed by Alan Turing in the late nineteen

424
00:28:02,440 --> 00:28:06,600
Speaker 3: thirties as part of the effort to do cryptanalysis on

425
00:28:06,920 --> 00:28:13,920
Speaker 3: the Enigma encryption machine that the Germans used. And why

426
00:28:13,960 --> 00:28:17,840
Speaker 3: do you need a language model? Because there were an

427
00:28:18,480 --> 00:28:22,520
Speaker 3: unreasonably large and astronomically large number of possible settings with

428
00:28:22,680 --> 00:28:26,440
Speaker 3: the encryption machine. But they had a clever way of

429
00:28:26,680 --> 00:28:32,359
Speaker 3: using patterns in the ciphertext to cut that number down

430
00:28:32,920 --> 00:28:37,160
Speaker 3: to a few thousands. And then in principle, you could

431
00:28:37,240 --> 00:28:40,920
Speaker 3: try each of the thousand possible settings and see whether

432
00:28:41,520 --> 00:28:48,120
Speaker 3: the result of producing clear text from the ciphertext made

433
00:28:48,160 --> 00:28:50,920
Speaker 3: any sense. But that would take a long time. If

434
00:28:50,960 --> 00:28:53,240
Speaker 3: you had a human being do it looking at each

435
00:28:53,320 --> 00:28:56,560
Speaker 3: of the outputs, most of which would be gibberish. And

436
00:28:56,720 --> 00:29:03,640
Speaker 3: so Touring invented a model of German letter sequences, and

437
00:29:03,840 --> 00:29:09,240
Speaker 3: that model would inclusively assign a probability to a sequence

438
00:29:09,320 --> 00:29:13,120
Speaker 3: of letters, as how likely is it that this is German?

439
00:29:14,240 --> 00:29:17,400
Speaker 3: And so they built a machine, which was actually also

440
00:29:17,520 --> 00:29:21,360
Speaker 3: one of the world's first computers, which would run through

441
00:29:21,520 --> 00:29:27,360
Speaker 3: thousands of possible encryption keys, produce the clear text corresponding

442
00:29:27,440 --> 00:29:30,000
Speaker 3: to that key, and then ask it does this seem

443
00:29:30,120 --> 00:29:32,479
Speaker 3: like German? And if it seemed enough like German, then

444
00:29:32,520 --> 00:29:34,200
Speaker 3: it would spit it out and give it to a

445
00:29:34,280 --> 00:29:38,040
Speaker 3: human being to check. So, you know, that was in

446
00:29:38,640 --> 00:29:41,240
Speaker 3: thirty nine or so, and ever since then people have

447
00:29:41,400 --> 00:29:45,600
Speaker 3: been using techniques of that kind to produce models of language,

448
00:29:45,960 --> 00:29:50,920
Speaker 3: and those were used, for example, in speech recognition. If

449
00:29:50,960 --> 00:29:53,760
Speaker 3: you're trying to figure out you have some noises and

450
00:29:53,800 --> 00:29:56,600
Speaker 3: you're trying to figure out what's the text that corresponds,

451
00:29:57,240 --> 00:30:00,920
Speaker 3: you put together what you can gather from looking at

452
00:30:00,960 --> 00:30:04,320
Speaker 3: the sounds and what you can predict from asking how

453
00:30:04,560 --> 00:30:08,920
Speaker 3: likely is this is? This is this text that the

454
00:30:09,800 --> 00:30:13,920
Speaker 3: species tech system is producing, And that again has been

455
00:30:14,000 --> 00:30:16,960
Speaker 3: going on for many decades. And then at a certain

456
00:30:17,040 --> 00:30:19,760
Speaker 3: point it occurred to somebody, you know, we could take

457
00:30:20,320 --> 00:30:24,840
Speaker 3: these language models and use them in a generative way.

458
00:30:25,280 --> 00:30:28,920
Speaker 3: That is, rather than using them to ask what is

459
00:30:29,000 --> 00:30:31,760
Speaker 3: the right way to interpret this sound? What is the

460
00:30:31,840 --> 00:30:35,120
Speaker 3: right way to interpret this encryption? Instead, let's turn it

461
00:30:35,160 --> 00:30:38,280
Speaker 3: the other way around and say, all right, you know,

462
00:30:38,440 --> 00:30:43,000
Speaker 3: write me a story that and you could you could

463
00:30:43,040 --> 00:30:45,440
Speaker 3: do it by giving a prompt, by saying, all right,

464
00:30:45,560 --> 00:30:48,360
Speaker 3: start this way and then continue. I think that's what

465
00:30:48,480 --> 00:30:51,840
Speaker 3: a language model does. It says, if you have a

466
00:30:51,960 --> 00:30:55,280
Speaker 3: certain pattern of letters, what are the likely ways to

467
00:30:55,840 --> 00:30:58,480
Speaker 3: go to get to there or to go from there?

468
00:31:00,120 --> 00:31:03,080
Speaker 3: And what is large means well, because now we have

469
00:31:03,240 --> 00:31:08,640
Speaker 3: big computers and lots of GPUs, you know, warehouses full

470
00:31:08,680 --> 00:31:13,200
Speaker 3: of computers. These language models, instead of being the small

471
00:31:13,320 --> 00:31:16,640
Speaker 3: to medium sized language models from the thirties and forties

472
00:31:16,720 --> 00:31:20,240
Speaker 3: and fifties and sixties and seventies, now they're large language

473
00:31:20,280 --> 00:31:25,840
Speaker 3: models with hundreds of billions, even trillions of parameters. And

474
00:31:25,960 --> 00:31:30,280
Speaker 3: it turns out that that scale, that that scale of

475
00:31:30,440 --> 00:31:34,800
Speaker 3: parameters for modeling what is a likely sentence of English

476
00:31:34,920 --> 00:31:40,040
Speaker 3: or German or Chinese or branch or whatever produces there

477
00:31:40,320 --> 00:31:44,080
Speaker 3: these amazing emergent properties that come up when you start

478
00:31:44,200 --> 00:31:48,040
Speaker 3: using those very large models in a generative fashion.

479
00:31:48,720 --> 00:31:49,200
Speaker 1: Thank you for that.

480
00:31:49,320 --> 00:31:51,960
Speaker 2: I can come as well, but I kind of get

481
00:31:52,120 --> 00:31:58,360
Speaker 2: thank you and the fact that there are many languages,

482
00:31:58,480 --> 00:32:04,600
Speaker 2: So there's AI that generative AI. There are many languages,

483
00:32:06,560 --> 00:32:09,800
Speaker 2: but there are subtle differences. There are certain words in

484
00:32:09,920 --> 00:32:13,880
Speaker 2: one language that don't exist in another, and a translation

485
00:32:14,120 --> 00:32:15,640
Speaker 2: might not be exactly the same.

486
00:32:16,240 --> 00:32:21,479
Speaker 1: So does this cause a problem for AI that has

487
00:32:21,520 --> 00:32:22,560
Speaker 1: to somehow be dealt.

488
00:32:22,400 --> 00:32:23,240
Speaker 5: With and.

489
00:32:25,400 --> 00:32:29,520
Speaker 1: One of the in the end result one language. Will

490
00:32:29,560 --> 00:32:32,800
Speaker 1: we someday way down the line, to.

491
00:32:34,280 --> 00:32:38,560
Speaker 2: Maximize the use of artificial intelligence, only have one language

492
00:32:38,920 --> 00:32:39,640
Speaker 2: around the world.

493
00:32:42,560 --> 00:32:47,040
Speaker 3: Well, I don't think that's very likely. Although the seven

494
00:32:47,160 --> 00:32:51,800
Speaker 3: or eight thousand existing languages in the world, probably at

495
00:32:51,880 --> 00:32:55,520
Speaker 3: least half of them are what's called endangered in the

496
00:32:55,640 --> 00:32:59,800
Speaker 3: sense that there are relatively few, if any, kids that

497
00:32:59,840 --> 00:33:02,560
Speaker 3: are learning that language and using it at home. But

498
00:33:02,680 --> 00:33:06,520
Speaker 3: there are still thousands of languages and language varieties that

499
00:33:06,840 --> 00:33:10,440
Speaker 3: are being learned and used in the home. And even

500
00:33:10,520 --> 00:33:14,880
Speaker 3: when you have a situation where there's a dominant language

501
00:33:15,480 --> 00:33:20,880
Speaker 3: over a large area that everybody is motivated to learn

502
00:33:20,920 --> 00:33:24,280
Speaker 3: and in many cases forced to learn that doesn't actually

503
00:33:24,480 --> 00:33:28,000
Speaker 3: prevent people from speaking other languages at home, so to speak.

504
00:33:28,440 --> 00:33:30,440
Speaker 3: And so I think it's very unlikely that we're going

505
00:33:30,480 --> 00:33:32,680
Speaker 3: to wind up in one single language.

506
00:33:34,680 --> 00:33:35,600
Speaker 1: But go ahead.

507
00:33:36,600 --> 00:33:38,440
Speaker 3: What I was going to break in and say is

508
00:33:38,560 --> 00:33:42,760
Speaker 3: that one of the first practical applications of language models

509
00:33:42,920 --> 00:33:48,800
Speaker 3: was in machine translation, and that was that was created

510
00:33:48,920 --> 00:33:53,960
Speaker 3: by that was used in systems that had a model

511
00:33:54,000 --> 00:33:55,840
Speaker 3: of English and a model of French, or a model

512
00:33:55,880 --> 00:33:58,240
Speaker 3: of English and a model of German, or a model

513
00:33:58,280 --> 00:34:01,360
Speaker 3: of English and a model of Chinese or whatever, and

514
00:34:02,640 --> 00:34:10,839
Speaker 3: would attempt to construct a system which would figure out

515
00:34:11,160 --> 00:34:16,560
Speaker 3: what pattern of letters in language X corresponds best to

516
00:34:16,680 --> 00:34:20,600
Speaker 3: a given input pattern of letters in language Y. And

517
00:34:21,400 --> 00:34:28,120
Speaker 3: that you know, worked not very well in the nineteen seventies,

518
00:34:28,160 --> 00:34:30,359
Speaker 3: and a little better in the nineteen eighties, a little

519
00:34:30,440 --> 00:34:33,719
Speaker 3: better in the nineteen nineties, and by now for languages

520
00:34:33,760 --> 00:34:36,719
Speaker 3: that have lots and lots of training material, although it's

521
00:34:36,800 --> 00:34:40,719
Speaker 3: far from perfect, it works pretty well. And that you know,

522
00:34:40,800 --> 00:34:44,640
Speaker 3: there's an old saying that goes back to a British

523
00:34:45,880 --> 00:34:52,920
Speaker 3: linguist in the nineteen fifties JR. Firth, which is, you

524
00:34:53,040 --> 00:34:56,040
Speaker 3: shall know a word by the company it keeps and

525
00:34:56,200 --> 00:34:59,160
Speaker 3: so rather than looking, rather than knowing what an English

526
00:34:59,200 --> 00:35:01,920
Speaker 3: word means, or what a Chinese word means, or what

527
00:35:02,000 --> 00:35:06,480
Speaker 3: a Spanish word means by looking it up in a dictionary. Instead,

528
00:35:06,600 --> 00:35:10,279
Speaker 3: what they do is they asked, well, okay, now we

529
00:35:10,440 --> 00:35:16,080
Speaker 3: have a few billion documents in English and this word

530
00:35:16,200 --> 00:35:21,000
Speaker 3: occurs hundreds of thousands or millions of times. What is

531
00:35:21,080 --> 00:35:23,200
Speaker 3: in its neighborhood? What is the commons?

532
00:35:23,280 --> 00:35:23,840
Speaker 1: The context?

533
00:35:23,960 --> 00:35:27,040
Speaker 3: Yeah? So how does it get used?

534
00:35:27,280 --> 00:35:32,480
Speaker 2: Linguistics must be really important in proper in deciding prompting

535
00:35:32,600 --> 00:35:37,520
Speaker 2: for AI, especially again, different languages not quite meaning the

536
00:35:37,600 --> 00:35:40,120
Speaker 2: same thing, but you want to get the exact prompt right.

537
00:35:41,320 --> 00:35:46,560
Speaker 2: Are you or people in your discipline involved with standardization

538
00:35:46,760 --> 00:35:49,800
Speaker 2: of prompts across different languages?

539
00:35:51,320 --> 00:35:54,800
Speaker 3: Well, I do know that. So there is this field

540
00:35:54,880 --> 00:35:57,560
Speaker 3: that was extremely popular a couple of years ago. It

541
00:35:57,680 --> 00:36:01,320
Speaker 3: is kind of declining in popularity now called prompt engineering,

542
00:36:01,960 --> 00:36:06,720
Speaker 3: the idea of being that rather than rewrite AI systems,

543
00:36:06,800 --> 00:36:09,040
Speaker 3: what you do is have to worry about how to

544
00:36:09,160 --> 00:36:13,759
Speaker 3: construct the prompt to a general system in order to

545
00:36:13,800 --> 00:36:15,640
Speaker 3: get it to do whatever you want it to do.

546
00:36:16,280 --> 00:36:19,799
Speaker 3: And it is true that some prompts work a lot

547
00:36:19,920 --> 00:36:22,960
Speaker 3: better than other prompts, and so knowing how to construct

548
00:36:23,040 --> 00:36:28,160
Speaker 3: prompts is you know, is worthwhile. But a lot of

549
00:36:28,239 --> 00:36:31,480
Speaker 3: what's going on now is more along the lines of

550
00:36:32,560 --> 00:36:39,600
Speaker 3: trying to adapt the systems to the task, the relevant

551
00:36:39,680 --> 00:36:42,759
Speaker 3: tasks and to sort of fine tune them to work

552
00:36:42,840 --> 00:36:46,799
Speaker 3: better on one thing or another, rather than to make

553
00:36:46,920 --> 00:36:50,759
Speaker 3: people learn how to praise the prompt exactly right to

554
00:36:50,800 --> 00:36:51,759
Speaker 3: get it to do what they want.

555
00:36:52,480 --> 00:36:54,719
Speaker 2: Thank you so much s for joining us, professor. I

556
00:36:55,120 --> 00:36:58,680
Speaker 2: really I do appreciate it, and that's my pleasure.

557
00:36:58,800 --> 00:37:00,399
Speaker 3: Thanks for having me of fun.

558
00:37:00,480 --> 00:37:02,160
Speaker 1: And we'll talk to you again soon. Thanks.

559
00:37:03,440 --> 00:37:05,240
Speaker 3: Okay, bye bye, good luck, bye.

560
00:37:05,120 --> 00:37:07,279
Speaker 2: Bye, and say to Rob Watts, say hi to Rob

561
00:37:07,360 --> 00:37:09,160
Speaker 2: Wats for me if if you see him again.

562
00:37:11,320 --> 00:37:13,000
Speaker 1: So, uh, there you go, folks.

563
00:37:13,920 --> 00:37:15,880
Speaker 2: We have a little bit more before the top of

564
00:37:15,920 --> 00:37:18,800
Speaker 2: the hour, and then we'll have open lines. I have

565
00:37:18,960 --> 00:37:21,719
Speaker 2: some suggestions. I do like the open lines because that's

566
00:37:21,719 --> 00:37:25,160
Speaker 2: when people feel comfortable and calling in. I noticed that

567
00:37:25,360 --> 00:37:31,279
Speaker 2: Alex and Millis did, did Colin haven't heard from some

568
00:37:31,440 --> 00:37:31,880
Speaker 2: of the folks.

569
00:37:32,239 --> 00:37:33,040
Speaker 1: Glenn. I don't know.

570
00:37:33,239 --> 00:37:34,880
Speaker 2: Glenn A lot of time likes to wait till the

571
00:37:35,000 --> 00:37:38,399
Speaker 2: end on Friday to Colin. Maybe we'll hear from Glenn

572
00:37:38,480 --> 00:37:41,239
Speaker 2: get some find now, what's up with him? Something like

573
00:37:41,360 --> 00:37:44,200
Speaker 2: that that's coming up on w b Z.

574
00:37:45,560 --> 00:37:49,560
Speaker 1: It's Night Side with Dan Ray on w b Boston's

575
00:37:49,640 --> 00:37:50,200
Speaker 1: news radio.

576
00:37:50,320 --> 00:37:51,440
Speaker 2: Before we get to the top of the air, I

577
00:37:51,560 --> 00:37:54,160
Speaker 2: just want to say it's open lines and maybe throw

578
00:37:54,239 --> 00:37:58,520
Speaker 2: some things out there for you that they have been

579
00:37:58,640 --> 00:38:04,280
Speaker 2: ruminating with me. One of them is the following should

580
00:38:04,280 --> 00:38:09,319
Speaker 2: weed Should smoking marijuana in public be banned? You can't

581
00:38:09,440 --> 00:38:13,000
Speaker 2: drink in public, why is it okay to smoke in public?

582
00:38:13,080 --> 00:38:16,080
Speaker 2: I don't know the rule, but and I'm walking trying

583
00:38:16,080 --> 00:38:18,080
Speaker 2: to live my life, and I don't mind marijuana, and

584
00:38:18,080 --> 00:38:20,400
Speaker 2: I don't mind the smell, but I kind of do

585
00:38:21,239 --> 00:38:24,800
Speaker 2: mind the intrusiveness of the smell. Say I'm walking in

586
00:38:24,880 --> 00:38:27,839
Speaker 2: the park walking my doggie and you get this big

587
00:38:27,960 --> 00:38:30,160
Speaker 2: blast of skunk smelling smell.

588
00:38:31,160 --> 00:38:31,839
Speaker 3: That is uh.

589
00:38:31,960 --> 00:38:35,240
Speaker 2: You know, we have freedom of We have certain freedoms.

590
00:38:35,280 --> 00:38:36,800
Speaker 2: You have the freedom to move your arm wherever you

591
00:38:36,840 --> 00:38:38,200
Speaker 2: want until it did someone's face.

592
00:38:38,920 --> 00:38:40,719
Speaker 1: You have the you have the freedom to.

593
00:38:42,640 --> 00:38:45,200
Speaker 2: Blow your smoke where you want until it's blown in

594
00:38:45,280 --> 00:38:51,640
Speaker 2: my face. So I'm gonna say, yeah, let's let's just

595
00:38:51,719 --> 00:38:55,719
Speaker 2: make a rule no smoking weed in public. I know

596
00:38:55,880 --> 00:38:58,040
Speaker 2: nobody's going to enforce it, but at least let's let

597
00:38:58,120 --> 00:39:00,120
Speaker 2: us get it down there so so we know, oh,

598
00:39:00,200 --> 00:39:02,960
Speaker 2: that's kind of what the community.

599
00:39:02,640 --> 00:39:05,840
Speaker 1: Expects and what the law expects. And if you really.

600
00:39:05,640 --> 00:39:07,480
Speaker 2: Flaut it, maybe you get a ticket or something. So

601
00:39:07,600 --> 00:39:11,200
Speaker 2: that's one thing, but it's open line six months, seven, two, five, four, ten.

602
00:39:11,160 --> 00:39:13,360
Speaker 1: Thirty, easy go on, good fun. I'm WBZ