WEBVTT 00:00:00.000 --> 00:00:01.895 align:middle line:90% [SQUEAKING] 00:00:01.895 --> 00:00:02.395 align:middle line:90% 00:00:02.395 --> 00:00:03.811 align:middle line:90% [RUSTLING] 00:00:03.811 --> 00:00:04.311 align:middle line:90% 00:00:04.311 --> 00:00:05.748 align:middle line:90% [CLICKING] 00:00:05.748 --> 00:00:11.030 align:middle line:90% 00:00:11.030 --> 00:00:15.082 align:middle line:84% NORVIN RICHARDS: All right, so one 00:00:15.082 --> 00:00:16.790 align:middle line:84% of the things we've learned in this class 00:00:16.790 --> 00:00:18.590 align:middle line:90% is that language is complicated. 00:00:18.590 --> 00:00:20.870 align:middle line:84% At least, maybe you already knew that, 00:00:20.870 --> 00:00:23.060 align:middle line:84% but we've spent a semester looking 00:00:23.060 --> 00:00:26.308 align:middle line:84% at various ways in which language is more complicated 00:00:26.308 --> 00:00:28.850 align:middle line:84% than you might have-- than you might have realized before you 00:00:28.850 --> 00:00:30.475 align:middle line:84% came in here and in ways that maybe you 00:00:30.475 --> 00:00:33.032 align:middle line:90% hadn't thought about much. 00:00:33.032 --> 00:00:34.490 align:middle line:84% These are a few of the things we've 00:00:34.490 --> 00:00:37.790 align:middle line:84% talked about, various special properties of language, 00:00:37.790 --> 00:00:41.300 align:middle line:84% fancy things that language does that we've spent the semester 00:00:41.300 --> 00:00:43.290 align:middle line:90% puzzling over. 00:00:43.290 --> 00:00:47.940 align:middle line:84% And if language has all of these complicated properties, 00:00:47.940 --> 00:00:51.180 align:middle line:84% the recurring question, which I've raised a couple of times, 00:00:51.180 --> 00:00:54.270 align:middle line:90% is how do we acquire this? 00:00:54.270 --> 00:00:56.850 align:middle line:84% We're going to talk seriously about language acquisition 00:00:56.850 --> 00:00:59.200 align:middle line:90% sometime next week. 00:00:59.200 --> 00:01:03.510 align:middle line:84% But a provisional answer to this that people have offered 00:01:03.510 --> 00:01:07.320 align:middle line:84% and taken seriously is that at least some of this stuff 00:01:07.320 --> 00:01:09.510 align:middle line:90% is innate, in the sense that-- 00:01:09.510 --> 00:01:11.220 align:middle line:84% I think I've said it this way before-- 00:01:11.220 --> 00:01:13.470 align:middle line:84% part of being a human being is having 00:01:13.470 --> 00:01:16.200 align:middle line:84% the kind of mind that constructs language in certain ways, 00:01:16.200 --> 00:01:17.020 align:middle line:90% but not others. 00:01:17.020 --> 00:01:20.640 align:middle line:84% So if we ask questions like, why do languages 00:01:20.640 --> 00:01:22.110 align:middle line:90% have binary branching? 00:01:22.110 --> 00:01:24.120 align:middle line:90% Why not ternary branching? 00:01:24.120 --> 00:01:25.800 align:middle line:84% The answer is, well, we're not equipped 00:01:25.800 --> 00:01:27.120 align:middle line:90% with ternary branching. 00:01:27.120 --> 00:01:29.850 align:middle line:84% The human brain does it this way and not some other way. 00:01:29.850 --> 00:01:33.150 align:middle line:84% And that leads to lots of other interesting questions, 00:01:33.150 --> 00:01:33.840 align:middle line:90% like why? 00:01:33.840 --> 00:01:38.910 align:middle line:84% What is it about human minds that are set up this way? 00:01:38.910 --> 00:01:43.110 align:middle line:84% Can we get that to follow from anything else, any deeper 00:01:43.110 --> 00:01:44.010 align:middle line:90% conditions? 00:01:44.010 --> 00:01:45.910 align:middle line:84% But that's a claim that's out there, 00:01:45.910 --> 00:01:49.890 align:middle line:84% which we have referred to a couple of times. 00:01:49.890 --> 00:01:51.407 align:middle line:84% We don't start with a blank slate. 00:01:51.407 --> 00:01:53.490 align:middle line:84% We start with a rich body of linguistic knowledge, 00:01:53.490 --> 00:01:55.323 align:middle line:84% and learning our actual language is a matter 00:01:55.323 --> 00:01:57.090 align:middle line:90% of filling in some details. 00:01:57.090 --> 00:01:59.880 align:middle line:90% 00:01:59.880 --> 00:02:04.410 align:middle line:84% But it's very clear that not everything is innate. 00:02:04.410 --> 00:02:06.810 align:middle line:90% For example, here's a cartoon. 00:02:06.810 --> 00:02:09.000 align:middle line:90% It has a cat in it. 00:02:09.000 --> 00:02:12.990 align:middle line:84% And the fact that animal is pronounced "cat" 00:02:12.990 --> 00:02:14.850 align:middle line:90% is an accident. 00:02:14.850 --> 00:02:17.610 align:middle line:84% It could have been something else. 00:02:17.610 --> 00:02:20.520 align:middle line:84% So Saussure famously pointed this out and called it 00:02:20.520 --> 00:02:23.310 align:middle line:84% the arbitrariness of the sign, that, 00:02:23.310 --> 00:02:27.780 align:middle line:84% for the most part, what word the language has 00:02:27.780 --> 00:02:31.302 align:middle line:84% for a given thing has to do with the phonetic inventory 00:02:31.302 --> 00:02:33.510 align:middle line:84% of the language and other properties of the language, 00:02:33.510 --> 00:02:37.062 align:middle line:84% but the English word "cat" refers to cats, 00:02:37.062 --> 00:02:38.770 align:middle line:84% but that's kind of a historical accident. 00:02:38.770 --> 00:02:41.640 align:middle line:84% It could have referred to dogs or mice or anything else. 00:02:41.640 --> 00:02:43.620 align:middle line:90% Just happens to refer to cats. 00:02:43.620 --> 00:02:46.250 align:middle line:90% 00:02:46.250 --> 00:02:49.100 align:middle line:84% And the fact is that kids, if you're ever around small kids, 00:02:49.100 --> 00:02:55.250 align:middle line:84% you'll hear them trying out theories about what words mean 00:02:55.250 --> 00:02:57.740 align:middle line:90% and being wrong. 00:02:57.740 --> 00:03:00.170 align:middle line:84% I was once around a small child who used the word 00:03:00.170 --> 00:03:02.780 align:middle line:90% "moon" for every light. 00:03:02.780 --> 00:03:06.530 align:middle line:84% So the hall light was a moon, and the moon was a moon, 00:03:06.530 --> 00:03:07.820 align:middle line:90% and the sun was a moon. 00:03:07.820 --> 00:03:10.910 align:middle line:90% Everything was a moon. 00:03:10.910 --> 00:03:16.680 align:middle line:84% And sometimes these mistakes survive to adulthood. 00:03:16.680 --> 00:03:19.508 align:middle line:90% So take the word "livid." 00:03:19.508 --> 00:03:21.050 align:middle line:84% What do you think of the word "livid" 00:03:21.050 --> 00:03:23.930 align:middle line:84% as meaning, if you're a native speaker of English 00:03:23.930 --> 00:03:26.010 align:middle line:90% or even if you're not? 00:03:26.010 --> 00:03:27.150 align:middle line:90% Very angry. 00:03:27.150 --> 00:03:29.400 align:middle line:90% That's what it often means. 00:03:29.400 --> 00:03:32.910 align:middle line:90% It's originally a color term. 00:03:32.910 --> 00:03:34.800 align:middle line:90% It meant white. 00:03:34.800 --> 00:03:37.470 align:middle line:84% Because there was an idea that when people got really angry, 00:03:37.470 --> 00:03:38.730 align:middle line:90% their faces turned white. 00:03:38.730 --> 00:03:41.400 align:middle line:90% They were white with rage. 00:03:41.400 --> 00:03:43.620 align:middle line:84% You sometimes see the expression, 00:03:43.620 --> 00:03:44.490 align:middle line:90% his face went livid. 00:03:44.490 --> 00:03:46.073 align:middle line:84% When I was a kid, I thought that meant 00:03:46.073 --> 00:03:48.210 align:middle line:84% his face was red because I think of people 00:03:48.210 --> 00:03:50.940 align:middle line:90% who are angry as being red. 00:03:50.940 --> 00:03:52.955 align:middle line:84% But originally, that's what the word meant. 00:03:52.955 --> 00:03:53.580 align:middle line:90% It meant white. 00:03:53.580 --> 00:03:58.230 align:middle line:84% And then yes, I think for most of us, now it means angry. 00:03:58.230 --> 00:04:01.532 align:middle line:84% Or I used to believe in the verb "misle" until-- 00:04:01.532 --> 00:04:03.990 align:middle line:84% I'm embarrassed to say when I stopped believing in the word 00:04:03.990 --> 00:04:04.620 align:middle line:90% "misle." 00:04:04.620 --> 00:04:08.438 align:middle line:84% I was an adult. Did anybody else have the verb "misle"? 00:04:08.438 --> 00:04:10.230 align:middle line:84% Is there anybody who right now is thinking, 00:04:10.230 --> 00:04:11.355 align:middle line:90% what are you talking about? 00:04:11.355 --> 00:04:13.200 align:middle line:84% Of course, there's a verb "misle." 00:04:13.200 --> 00:04:15.510 align:middle line:84% I believed in the verb "misle" because I had read it 00:04:15.510 --> 00:04:18.120 align:middle line:90% in expressions like this. 00:04:18.120 --> 00:04:23.570 align:middle line:84% I had been "misle-d" was what I believed. 00:04:23.570 --> 00:04:25.680 align:middle line:90% It wasn't-- I think-- 00:04:25.680 --> 00:04:29.100 align:middle line:84% I don't believe I actually used the word in conversation 00:04:29.100 --> 00:04:30.270 align:middle line:90% and had somebody correct me. 00:04:30.270 --> 00:04:33.390 align:middle line:84% I think this was, like, I was in my 20s or something 00:04:33.390 --> 00:04:37.570 align:middle line:84% when I sat down and was like, wait a minute. 00:04:37.570 --> 00:04:38.850 align:middle line:90% So stuff like this happens. 00:04:38.850 --> 00:04:41.730 align:middle line:90% 00:04:41.730 --> 00:04:44.850 align:middle line:84% And sometimes these mistakes catch on, and people-- 00:04:44.850 --> 00:04:46.330 align:middle line:84% and sort of survive into adulthood. 00:04:46.330 --> 00:04:48.060 align:middle line:84% As I say, I was in my 20s, and I still 00:04:48.060 --> 00:04:50.280 align:middle line:84% believed that there was a verb "misle." 00:04:50.280 --> 00:04:54.750 align:middle line:84% If I were a king or something, then 00:04:54.750 --> 00:04:56.225 align:middle line:84% if my subjects tried to convince me 00:04:56.225 --> 00:04:57.600 align:middle line:84% that there wasn't a verb "misle," 00:04:57.600 --> 00:04:58.933 align:middle line:90% I would just have them beheaded. 00:04:58.933 --> 00:05:00.360 align:middle line:84% And before we knew it, there would 00:05:00.360 --> 00:05:05.190 align:middle line:90% be a verb "misle" in English. 00:05:05.190 --> 00:05:09.030 align:middle line:84% Or if I were somebody else who was hugely, hugely 00:05:09.030 --> 00:05:13.230 align:middle line:84% influential, like a linguistics professor or something. 00:05:13.230 --> 00:05:15.180 align:middle line:90% Wait a minute. 00:05:15.180 --> 00:05:17.250 align:middle line:84% There are also various forms of semantic drift 00:05:17.250 --> 00:05:19.050 align:middle line:84% that languages undergo, where words 00:05:19.050 --> 00:05:22.950 align:middle line:90% change what they refer to. 00:05:22.950 --> 00:05:24.520 align:middle line:84% "Livid" is maybe an example of that. 00:05:24.520 --> 00:05:26.550 align:middle line:84% It used to mean white, and now it means angry. 00:05:26.550 --> 00:05:27.760 align:middle line:90% Here's another example. 00:05:27.760 --> 00:05:32.700 align:middle line:84% The modern English word "bead" originally meant a prayer. 00:05:32.700 --> 00:05:35.100 align:middle line:84% So there's still-- there's a German verb "beten." 00:05:35.100 --> 00:05:36.210 align:middle line:90% That's the verb "to pray." 00:05:36.210 --> 00:05:38.970 align:middle line:84% It's related to the English word "bead"-- 00:05:38.970 --> 00:05:43.470 align:middle line:84% in Old English, the ancestor of the word "bead" meant a prayer. 00:05:43.470 --> 00:05:48.660 align:middle line:84% The switch from prayer to bead, you know, small, round object, 00:05:48.660 --> 00:05:51.390 align:middle line:84% came via rosaries, which are devices 00:05:51.390 --> 00:05:55.090 align:middle line:84% for counting prayers in certain religious traditions, 00:05:55.090 --> 00:05:57.558 align:middle line:84% including Catholicism and others. 00:05:57.558 --> 00:05:59.100 align:middle line:84% So if you're in a religious tradition 00:05:59.100 --> 00:06:01.260 align:middle line:84% where it's important to say a certain number of prayers, 00:06:01.260 --> 00:06:03.302 align:middle line:84% you have this little device that has beads on it. 00:06:03.302 --> 00:06:06.505 align:middle line:84% And you're counting the prayers using this device. 00:06:06.505 --> 00:06:08.130 align:middle line:84% So if you're asked, what are you doing? 00:06:08.130 --> 00:06:09.270 align:middle line:90% Well, you're counting prayers. 00:06:09.270 --> 00:06:10.200 align:middle line:84% But there's another sense in which 00:06:10.200 --> 00:06:12.930 align:middle line:84% you're counting, well, beads, like small, round objects 00:06:12.930 --> 00:06:13.990 align:middle line:90% on a wire. 00:06:13.990 --> 00:06:19.940 align:middle line:84% And this is thought to be how the word shifted its meaning. 00:06:19.940 --> 00:06:25.130 align:middle line:84% Or there are other things like this, words widen or narrow 00:06:25.130 --> 00:06:26.120 align:middle line:90% their meanings. 00:06:26.120 --> 00:06:29.150 align:middle line:84% In Old English, "steorfan" meant "to die." 00:06:29.150 --> 00:06:31.310 align:middle line:84% It's cognate with German "sterben." 00:06:31.310 --> 00:06:34.100 align:middle line:84% It's the ancestor of modern English "starve," 00:06:34.100 --> 00:06:38.450 align:middle line:90% which went through a stage of-- 00:06:38.450 --> 00:06:39.800 align:middle line:90% sort of originally meant "die." 00:06:39.800 --> 00:06:41.758 align:middle line:84% It went through a stage of meaning specifically 00:06:41.758 --> 00:06:43.100 align:middle line:90% "die from hunger." 00:06:43.100 --> 00:06:46.250 align:middle line:84% And today, I think of it as basically just meaning, 00:06:46.250 --> 00:06:48.110 align:middle line:90% be very hungry. 00:06:48.110 --> 00:06:50.060 align:middle line:84% So you can still say "He starved," 00:06:50.060 --> 00:06:53.210 align:middle line:84% and have that mean he died, I guess. 00:06:53.210 --> 00:06:56.600 align:middle line:84% But I'm a lot more likely to use it in an expression 00:06:56.600 --> 00:06:57.450 align:middle line:90% like "I'm starving. 00:06:57.450 --> 00:06:59.780 align:middle line:90% Let's have lunch." 00:06:59.780 --> 00:07:01.430 align:middle line:84% Maybe when I say that, I literally 00:07:01.430 --> 00:07:03.410 align:middle line:84% mean I'm dying of hunger, or figuratively 00:07:03.410 --> 00:07:06.440 align:middle line:90% I mean I'm dying of hunger. 00:07:06.440 --> 00:07:11.000 align:middle line:84% Or there's a French word "negre" which means a Black man. 00:07:11.000 --> 00:07:13.700 align:middle line:84% It has a descendant in modern Haitian Creole, 00:07:13.700 --> 00:07:16.700 align:middle line:90% "neg," which just means man. 00:07:16.700 --> 00:07:19.550 align:middle line:84% So the meaning of that has broadened 00:07:19.550 --> 00:07:21.990 align:middle line:84% for reasons that are not all that hard to understand. 00:07:21.990 --> 00:07:24.788 align:middle line:84% So Haitian Creole speakers, when they refer to men, 00:07:24.788 --> 00:07:26.330 align:middle line:84% they're often referring to Black men, 00:07:26.330 --> 00:07:31.460 align:middle line:84% since Haitian Creole speakers tend to be Black. 00:07:31.460 --> 00:07:35.060 align:middle line:84% So yeah, in Haitian Creole, I am a "neg," although in French, I 00:07:35.060 --> 00:07:37.760 align:middle line:90% am not a "negre." 00:07:37.760 --> 00:07:41.870 align:middle line:84% Or similar specialization of a term, 00:07:41.870 --> 00:07:45.050 align:middle line:84% there's the old English word "chniht," which 00:07:45.050 --> 00:07:48.620 align:middle line:84% meant a servant, and has a cognate in modern German, 00:07:48.620 --> 00:07:53.840 align:middle line:84% "knecht," which means something like a boy or a servant. 00:07:53.840 --> 00:07:56.630 align:middle line:84% It's the ancestor of the modern English word "knight," 00:07:56.630 --> 00:08:00.532 align:middle line:84% who was originally specifically a servant of the king. 00:08:00.532 --> 00:08:02.240 align:middle line:84% The reason that the English word "knight" 00:08:02.240 --> 00:08:05.750 align:middle line:84% has all of those unpronounced letters in it 00:08:05.750 --> 00:08:08.360 align:middle line:84% is that those are there in an attempt to pronounce the word 00:08:08.360 --> 00:08:11.835 align:middle line:84% or spell the word as it was once pronounced. 00:08:11.835 --> 00:08:13.460 align:middle line:84% We spell that with a K at the beginning 00:08:13.460 --> 00:08:15.200 align:middle line:84% because it used to be pronounced "cniht." 00:08:15.200 --> 00:08:17.090 align:middle line:90% There used to be a K there. 00:08:17.090 --> 00:08:20.780 align:middle line:84% And then English got out of the habit of pronouncing K 00:08:20.780 --> 00:08:23.270 align:middle line:84% before N, and so we have words like "knight" and "knee" 00:08:23.270 --> 00:08:29.240 align:middle line:84% where the K is still there as a memorial to the times 00:08:29.240 --> 00:08:31.220 align:middle line:84% when there was a stop there at the beginning. 00:08:31.220 --> 00:08:32.960 align:middle line:84% Same deal with the G-H, which spelled 00:08:32.960 --> 00:08:36.289 align:middle line:84% a [HISSES IN BACK OF THROAT] sound in the middle. 00:08:36.289 --> 00:08:39.080 align:middle line:90% Or this is one of my favorites. 00:08:39.080 --> 00:08:40.880 align:middle line:84% The old English word "housewife," 00:08:40.880 --> 00:08:43.700 align:middle line:84% "huswif," which is the ancestor of the modern English word 00:08:43.700 --> 00:08:44.870 align:middle line:90% "hussy." 00:08:44.870 --> 00:08:48.650 align:middle line:84% That's undergone some semantic change over the years, 00:08:48.650 --> 00:08:51.650 align:middle line:84% as it's gone down through the years. 00:08:51.650 --> 00:08:56.210 align:middle line:84% There's a disturbing tendency for words 00:08:56.210 --> 00:09:01.610 align:middle line:84% that refer to women to undergo pejoration, that is, to become 00:09:01.610 --> 00:09:03.065 align:middle line:90% worse in their meaning. 00:09:03.065 --> 00:09:05.630 align:middle line:90% 00:09:05.630 --> 00:09:09.800 align:middle line:84% There's possibly a corresponding tendency. 00:09:09.800 --> 00:09:12.290 align:middle line:84% If you look at lots of languages, 00:09:12.290 --> 00:09:16.160 align:middle line:84% there are lots of languages in which the modern word for "man" 00:09:16.160 --> 00:09:19.400 align:middle line:84% is straightforwardly related to earlier words for "man," 00:09:19.400 --> 00:09:21.050 align:middle line:84% but in which the word for "woman" 00:09:21.050 --> 00:09:25.130 align:middle line:84% has been replaced by some other word for "woman." 00:09:25.130 --> 00:09:28.070 align:middle line:84% And I assume it's related to this tendency, that words 00:09:28.070 --> 00:09:31.343 align:middle line:84% for women tend to become insulting over time 00:09:31.343 --> 00:09:33.260 align:middle line:84% in lots of different cultures, and so you need 00:09:33.260 --> 00:09:35.300 align:middle line:90% to replace them with something. 00:09:35.300 --> 00:09:37.820 align:middle line:84% Maybe relatedly, it's not uncommon for modern words 00:09:37.820 --> 00:09:41.740 align:middle line:84% for women to originally refer to queens or noblewomen. 00:09:41.740 --> 00:09:45.815 align:middle line:84% That's true, for example, both in German and in Zulu. 00:09:45.815 --> 00:09:48.065 align:middle line:84% The modern Zulu word for "woman" used to mean "queen." 00:09:48.065 --> 00:09:53.270 align:middle line:84% It replaced an older word for a woman that became insulting. 00:09:53.270 --> 00:09:58.280 align:middle line:84% Disturbing cross-linguistically fairly common tendency. 00:09:58.280 --> 00:10:00.980 align:middle line:84% Or-- this is my actual favorite in this list 00:10:00.980 --> 00:10:04.280 align:middle line:84% because I wonder about the confusion 00:10:04.280 --> 00:10:05.840 align:middle line:84% that it must have caused-- so there's 00:10:05.840 --> 00:10:08.420 align:middle line:84% a Proto-Austronesian word reconstructed 00:10:08.420 --> 00:10:11.370 align:middle line:84% as "wada," which was an existential construction. 00:10:11.370 --> 00:10:12.770 align:middle line:90% It meant "there is." 00:10:12.770 --> 00:10:15.170 align:middle line:84% It has a nice, straightforward descendant 00:10:15.170 --> 00:10:18.170 align:middle line:84% in Tagalog, "wala," which means "there is not." 00:10:18.170 --> 00:10:19.850 align:middle line:90% I don't know how that happened. 00:10:19.850 --> 00:10:23.300 align:middle line:84% There's an Arabic word, "wala," which 00:10:23.300 --> 00:10:25.280 align:middle line:90% means something like "no." 00:10:25.280 --> 00:10:26.960 align:middle line:84% If you want to say there are no books, 00:10:26.960 --> 00:10:28.490 align:middle line:90% you use "wala" to say that. 00:10:28.490 --> 00:10:31.350 align:middle line:84% And it's possible that Tagalog got influenced by Arabic. 00:10:31.350 --> 00:10:32.600 align:middle line:90% It was in contact with Arabic. 00:10:32.600 --> 00:10:34.457 align:middle line:90% It has Arabic borrowings. 00:10:34.457 --> 00:10:36.290 align:middle line:84% So maybe that's got something to do with it. 00:10:36.290 --> 00:10:39.710 align:middle line:84% But yes, there's presumably a fairly weird period 00:10:39.710 --> 00:10:41.870 align:middle line:84% in the middle there, where people 00:10:41.870 --> 00:10:45.580 align:middle line:90% were having existential crises. 00:10:45.580 --> 00:10:47.620 align:middle line:90% I just used a star. 00:10:47.620 --> 00:10:51.490 align:middle line:84% You guys are used, by now, to seeing stars 00:10:51.490 --> 00:10:53.740 align:middle line:84% on sentences that are ungrammatical, 00:10:53.740 --> 00:10:57.280 align:middle line:84% and that is how stars are used in syntax. 00:10:57.280 --> 00:10:59.620 align:middle line:84% But when people are doing historical linguistics, 00:10:59.620 --> 00:11:01.570 align:middle line:90% stars mean something else. 00:11:01.570 --> 00:11:04.810 align:middle line:84% They mean, this is a word that we don't actually 00:11:04.810 --> 00:11:09.020 align:middle line:84% see in any documents, or you can't hear it. 00:11:09.020 --> 00:11:12.310 align:middle line:84% It's our theory about an ancestral form. 00:11:12.310 --> 00:11:15.610 align:middle line:84% So Proto-Austronesian is the ancestor 00:11:15.610 --> 00:11:20.050 align:middle line:84% of a bunch of languages of the Pacific, mostly. 00:11:20.050 --> 00:11:22.450 align:middle line:84% So the languages of the Philippines and Indonesia 00:11:22.450 --> 00:11:24.940 align:middle line:90% and New Zealand and Hawaii. 00:11:24.940 --> 00:11:28.180 align:middle line:84% It's this gigantic geographic area that they went over. 00:11:28.180 --> 00:11:31.630 align:middle line:84% Plus some who got lost and ended up in Madagascar. 00:11:31.630 --> 00:11:36.280 align:middle line:84% There's an Austronesian language over there, too, Malagasy. 00:11:36.280 --> 00:11:38.440 align:middle line:84% The ancestor of all of those languages, 00:11:38.440 --> 00:11:40.000 align:middle line:90% it's called Proto-Austronesian. 00:11:40.000 --> 00:11:42.880 align:middle line:84% And that form that I just showed you for "there is" 00:11:42.880 --> 00:11:45.490 align:middle line:84% is a reconstruction of what they think that word would have 00:11:45.490 --> 00:11:46.630 align:middle line:90% been in Proto-Austronesian. 00:11:46.630 --> 00:11:48.400 align:middle line:84% We haven't met any Proto-Austronesians. 00:11:48.400 --> 00:11:50.200 align:middle line:84% They didn't leave us any documents. 00:11:50.200 --> 00:11:53.470 align:middle line:84% It's a hypothesized underlying form, given all the forms 00:11:53.470 --> 00:11:54.670 align:middle line:90% that we can see. 00:11:54.670 --> 00:11:56.860 align:middle line:84% That's what star means in historical linguistics. 00:11:56.860 --> 00:11:59.420 align:middle line:90% 00:11:59.420 --> 00:12:02.130 align:middle line:84% So various kinds of semantic drift. 00:12:02.130 --> 00:12:05.990 align:middle line:84% Recuttings-- so the example of "misle" is an example of that. 00:12:05.990 --> 00:12:08.390 align:middle line:84% I saw the word "misled," and I thought 00:12:08.390 --> 00:12:12.080 align:middle line:84% that it was the past tense or the participle of a verb, 00:12:12.080 --> 00:12:14.330 align:middle line:90% "to misle." 00:12:14.330 --> 00:12:16.460 align:middle line:84% There are other lots of examples of this 00:12:16.460 --> 00:12:19.400 align:middle line:84% with the indefinite article in English. 00:12:19.400 --> 00:12:22.670 align:middle line:84% So the indefinite article in English has two forms. 00:12:22.670 --> 00:12:27.990 align:middle line:84% It's "an" before a vowel, and it's "a" before a consonant. 00:12:27.990 --> 00:12:33.275 align:middle line:84% So we say "an apple," but we say "a banana." 00:12:33.275 --> 00:12:36.740 align:middle line:90% 00:12:36.740 --> 00:12:37.460 align:middle line:90% Yes, good. 00:12:37.460 --> 00:12:39.780 align:middle line:84% I always worry about when to stop spelling banana. 00:12:39.780 --> 00:12:42.620 align:middle line:90% It's not easy. 00:12:42.620 --> 00:12:45.440 align:middle line:84% A consequence of that is that if you hear a sequence like-- 00:12:45.440 --> 00:12:48.650 align:middle line:84% this is one of the examples for which this happened-- 00:12:48.650 --> 00:12:51.830 align:middle line:84% you hear a sequence like "a nickname," 00:12:51.830 --> 00:12:53.960 align:middle line:84% it's sort of hard to know whether you're hearing 00:12:53.960 --> 00:12:56.990 align:middle line:90% "a nickname" or "an ickname." 00:12:56.990 --> 00:13:00.740 align:middle line:84% So "nickname" was originally "ickname," "ekename." 00:13:00.740 --> 00:13:05.120 align:middle line:90% "Eke" is an old word for "also." 00:13:05.120 --> 00:13:08.930 align:middle line:84% It has a cognate in German, which is "auch." 00:13:08.930 --> 00:13:13.773 align:middle line:84% It shows up in The Canterbury Tales, 00:13:13.773 --> 00:13:15.440 align:middle line:84% in the prologue to The Canterbury Tales. 00:13:15.440 --> 00:13:18.248 align:middle line:84% If you ever read the prologue to The Canterbury Tales, 00:13:18.248 --> 00:13:19.956 align:middle line:84% Chaucer says something like "and Zephyrus 00:13:19.956 --> 00:13:21.500 align:middle line:90% eke, with his swete brethe"-- 00:13:21.500 --> 00:13:23.540 align:middle line:84% "And Zephyr also with his sweet breath." 00:13:23.540 --> 00:13:25.130 align:middle line:90% So that's what that "eke" is. 00:13:25.130 --> 00:13:28.160 align:middle line:84% It's an "also name," that is, a name that you have in addition 00:13:28.160 --> 00:13:30.740 align:middle line:90% to your other name. 00:13:30.740 --> 00:13:34.430 align:middle line:84% If ordinary sound changes had taken place, 00:13:34.430 --> 00:13:37.070 align:middle line:84% the word for "nickname" would be "ickname," 00:13:37.070 --> 00:13:39.440 align:middle line:84% but it's not, because of confusion about where 00:13:39.440 --> 00:13:41.050 align:middle line:90% the N went, basically. 00:13:41.050 --> 00:13:44.150 align:middle line:84% There are a bunch of words like this in English. 00:13:44.150 --> 00:13:47.460 align:middle line:90% This is another example. 00:13:47.460 --> 00:13:51.200 align:middle line:84% Middle English had a word "pease" which was a mass noun. 00:13:51.200 --> 00:13:55.640 align:middle line:84% By mass noun, I mean it referred to a certain kind of stuff that 00:13:55.640 --> 00:13:56.600 align:middle line:90% didn't have a plural. 00:13:56.600 --> 00:14:01.250 align:middle line:84% We have a bunch of words like this, words that 00:14:01.250 --> 00:14:03.260 align:middle line:90% don't ordinarily have plurals. 00:14:03.260 --> 00:14:09.770 align:middle line:84% Words like "water" or "ketchup" or "flour," 00:14:09.770 --> 00:14:14.750 align:middle line:84% where if you have a barrel of flour, 00:14:14.750 --> 00:14:18.278 align:middle line:84% you don't talk about it as a barrel of 16 million "flours." 00:14:18.278 --> 00:14:19.820 align:middle line:84% That's not what you're talking about. 00:14:19.820 --> 00:14:21.487 align:middle line:84% You're talking about a substance, right? 00:14:21.487 --> 00:14:23.120 align:middle line:90% A stuff. 00:14:23.120 --> 00:14:25.040 align:middle line:90% Same deal with ketchup or water. 00:14:25.040 --> 00:14:28.130 align:middle line:84% You can talk about different kinds of flours 00:14:28.130 --> 00:14:29.750 align:middle line:90% being different flours. 00:14:29.750 --> 00:14:32.030 align:middle line:84% I guess if you were really, really into ketchup, 00:14:32.030 --> 00:14:34.030 align:middle line:84% you could do the same thing with ketchup, right? 00:14:34.030 --> 00:14:37.940 align:middle line:84% I'm fond of the ketchups that you get in Southeast Asia, 00:14:37.940 --> 00:14:40.640 align:middle line:90% or whatever. 00:14:40.640 --> 00:14:43.130 align:middle line:84% "Pease" used to be a mass noun like that. 00:14:43.130 --> 00:14:46.550 align:middle line:84% It was just a name for this green stuff that consists 00:14:46.550 --> 00:14:49.130 align:middle line:90% of lots of little round things. 00:14:49.130 --> 00:14:52.790 align:middle line:84% But because this particular mass, well, 00:14:52.790 --> 00:14:55.370 align:middle line:84% can be subdivided into these little individual, round, 00:14:55.370 --> 00:14:58.160 align:middle line:84% green things, and because the word "pease" happened 00:14:58.160 --> 00:15:00.830 align:middle line:84% to end in a Z sound so it looked like a plural, 00:15:00.830 --> 00:15:02.600 align:middle line:90% it got reanalyzed as a plural. 00:15:02.600 --> 00:15:05.300 align:middle line:84% And so they made up a singular "pea," 00:15:05.300 --> 00:15:07.168 align:middle line:90% which didn't exist before. 00:15:07.168 --> 00:15:08.960 align:middle line:84% English has done a lot of things like this, 00:15:08.960 --> 00:15:10.820 align:middle line:84% sometimes called back- formation, 00:15:10.820 --> 00:15:14.180 align:middle line:84% where you take a word that looks polymorphemic, although it's 00:15:14.180 --> 00:15:16.395 align:middle line:90% not, and pretend that it is. 00:15:16.395 --> 00:15:18.770 align:middle line:84% Another one, which I don't think I have anywhere in here, 00:15:18.770 --> 00:15:21.230 align:middle line:84% is we had for a long time a word, "beggar." 00:15:21.230 --> 00:15:23.750 align:middle line:90% 00:15:23.750 --> 00:15:26.480 align:middle line:84% That noun was in the language for a long time. 00:15:26.480 --> 00:15:31.550 align:middle line:84% It ends in a sequence that's pronounced schwa R-- 00:15:31.550 --> 00:15:35.570 align:middle line:84% I've lost my ability to write the IPA symbol for R-- 00:15:35.570 --> 00:15:38.208 align:middle line:84% which sounds like our agentive suffix, the "-er" 00:15:38.208 --> 00:15:39.500 align:middle line:90% that's at the end of "teacher." 00:15:39.500 --> 00:15:41.930 align:middle line:84% And so people made up a verb, "beg." 00:15:41.930 --> 00:15:45.950 align:middle line:84% But it wasn't originally a polymorphemic word. 00:15:45.950 --> 00:15:52.337 align:middle line:84% That's one reason it's not spelled "begger," which 00:15:52.337 --> 00:15:54.920 align:middle line:84% is what you would have expected if you were adding the teacher 00:15:54.920 --> 00:15:56.962 align:middle line:90% suffix to "beg." 00:15:56.962 --> 00:15:59.420 align:middle line:84% It's spelled like this because, well, this is not the "-er" 00:15:59.420 --> 00:15:59.920 align:middle line:90% suffix. 00:15:59.920 --> 00:16:01.580 align:middle line:90% It's just pronounced like it. 00:16:01.580 --> 00:16:07.700 align:middle line:84% So the verb "to beg" got back-formed. 00:16:07.700 --> 00:16:09.500 align:middle line:84% Another similar kind of recutting. 00:16:09.500 --> 00:16:12.770 align:middle line:84% Here are three Old English words. 00:16:12.770 --> 00:16:17.040 align:middle line:84% The positive form, "near," which is "neah." 00:16:17.040 --> 00:16:18.530 align:middle line:84% Then that has a comparative form, 00:16:18.530 --> 00:16:20.150 align:middle line:90% "nearer," which was "nearra." 00:16:20.150 --> 00:16:22.190 align:middle line:84% And then there was a superlative form, 00:16:22.190 --> 00:16:24.320 align:middle line:90% "nearest," which was "neahsta." 00:16:24.320 --> 00:16:27.800 align:middle line:84% All three of those words have descendants in modern English. 00:16:27.800 --> 00:16:31.550 align:middle line:84% "neah" is the ancestor of "nigh," which we sort of still 00:16:31.550 --> 00:16:34.160 align:middle line:84% have in modern English, as a sort of archaic-sounding word 00:16:34.160 --> 00:16:35.630 align:middle line:90% for "near." 00:16:35.630 --> 00:16:37.640 align:middle line:90% Don't use it so much anymore. 00:16:37.640 --> 00:16:40.340 align:middle line:84% The comparative, the word that meant "nearer," turned 00:16:40.340 --> 00:16:43.070 align:middle line:90% into "near" in modern English. 00:16:43.070 --> 00:16:46.190 align:middle line:84% And "neahsta," the superlative, the nearest, 00:16:46.190 --> 00:16:49.610 align:middle line:84% is the ancestor of the modern English word "next." 00:16:49.610 --> 00:16:53.270 align:middle line:84% Regular sound change applied to "nearra" to give you "near" 00:16:53.270 --> 00:16:56.990 align:middle line:84% and obscured the fact that it was a comparative. 00:16:56.990 --> 00:17:00.350 align:middle line:84% So comparatives normally end in a schwa-R sequence. 00:17:00.350 --> 00:17:02.120 align:middle line:84% And because of regular sound change, 00:17:02.120 --> 00:17:04.993 align:middle line:84% basically you were adding the comparative suffix 00:17:04.993 --> 00:17:06.410 align:middle line:84% to something that ended in a vowel 00:17:06.410 --> 00:17:08.385 align:middle line:84% and then an H. The consequence was 00:17:08.385 --> 00:17:10.760 align:middle line:84% that you couldn't hear anymore that it was a comparative, 00:17:10.760 --> 00:17:12.365 align:middle line:84% so it got reanalyzed as something 00:17:12.365 --> 00:17:13.490 align:middle line:90% that was not a comparative. 00:17:13.490 --> 00:17:15.230 align:middle line:90% So it's now our word for "near." 00:17:15.230 --> 00:17:18.410 align:middle line:84% It's more or less replaced "nigh" as our word for "near." 00:17:18.410 --> 00:17:22.700 align:middle line:84% And then we built new comparatives and superlatives 00:17:22.700 --> 00:17:23.240 align:middle line:90% on that. 00:17:23.240 --> 00:17:25.047 align:middle line:90% It was originally comparative. 00:17:25.047 --> 00:17:27.680 align:middle line:90% 00:17:27.680 --> 00:17:31.730 align:middle line:84% All kinds of entertaining things happen to languages, 00:17:31.730 --> 00:17:35.150 align:middle line:84% sometimes as a matter of somebody 00:17:35.150 --> 00:17:39.020 align:middle line:84% like me goes through life a fairly long time believing 00:17:39.020 --> 00:17:39.980 align:middle line:90% in a verb, "to misle." 00:17:39.980 --> 00:17:42.620 align:middle line:84% And if I had a more forceful personality and more 00:17:42.620 --> 00:17:45.140 align:middle line:84% political clout, then I might have 00:17:45.140 --> 00:17:46.940 align:middle line:84% been able to force other people to believe 00:17:46.940 --> 00:17:50.600 align:middle line:90% in the verb "to misle" as well. 00:17:50.600 --> 00:17:53.420 align:middle line:84% Some of this presumably also is a matter 00:17:53.420 --> 00:17:56.690 align:middle line:84% of not misunderstandings, but just people deliberately 00:17:56.690 --> 00:17:57.420 align:middle line:90% being weird. 00:17:57.420 --> 00:17:59.930 align:middle line:84% I mean, if you think of things that people do with language 00:17:59.930 --> 00:18:02.990 align:middle line:84% today, they have done this all their lives, pretending 00:18:02.990 --> 00:18:05.540 align:middle line:84% that words mean things other than what they mean just 00:18:05.540 --> 00:18:08.180 align:middle line:84% for fun, as a way of having a good time 00:18:08.180 --> 00:18:11.450 align:middle line:84% or confusing your parents or whatever else. 00:18:11.450 --> 00:18:13.550 align:middle line:84% The other big kind of sound change, though, 00:18:13.550 --> 00:18:15.300 align:middle line:84% that I mainly want to talk about today are 00:18:15.300 --> 00:18:17.840 align:middle line:90% what are called sound changes. 00:18:17.840 --> 00:18:23.157 align:middle line:84% Here are some numbers from various languages of Europe. 00:18:23.157 --> 00:18:23.990 align:middle line:90% Are they all Europe? 00:18:23.990 --> 00:18:24.860 align:middle line:90% Yeah, mostly. 00:18:24.860 --> 00:18:25.480 align:middle line:90% Yeah. 00:18:25.480 --> 00:18:26.240 align:middle line:90% Yeah. 00:18:26.240 --> 00:18:29.420 align:middle line:90% Europe, widely-- well, Sanskrit. 00:18:29.420 --> 00:18:30.560 align:middle line:90% That's not in Europe. 00:18:30.560 --> 00:18:32.240 align:middle line:84% So Sanskrit, Greek, Latin, Gothic, 00:18:32.240 --> 00:18:35.748 align:middle line:84% which is an ancestor of the Germanic. 00:18:35.748 --> 00:18:36.290 align:middle line:90% No, it's not. 00:18:36.290 --> 00:18:37.970 align:middle line:90% It's an old Germanic language. 00:18:37.970 --> 00:18:41.420 align:middle line:90% It was spoken around the Crimea. 00:18:41.420 --> 00:18:43.970 align:middle line:84% Old Irish, Lithuanian, Old Church Slavonic, 00:18:43.970 --> 00:18:47.330 align:middle line:84% which is an ancestor of the Slavic languages. 00:18:47.330 --> 00:18:49.520 align:middle line:90% Basque, and Turkish. 00:18:49.520 --> 00:18:51.470 align:middle line:84% If you look at these languages, I've 00:18:51.470 --> 00:18:53.960 align:middle line:84% highlighted the second line because it's 00:18:53.960 --> 00:18:56.210 align:middle line:84% the line in which this is clearest. 00:18:56.210 --> 00:19:01.190 align:middle line:84% You can see a lot of these languages have words 00:19:01.190 --> 00:19:04.910 align:middle line:84% for the number two that look kind of similar to each other 00:19:04.910 --> 00:19:07.880 align:middle line:84% except for the last two, which really, really don't. 00:19:07.880 --> 00:19:11.660 align:middle line:84% So you've got words that have a "d" 00:19:11.660 --> 00:19:14.180 align:middle line:84% and then a "u," or OK, Gothic has a "t," 00:19:14.180 --> 00:19:16.580 align:middle line:90% the voiceless version of a "d." 00:19:16.580 --> 00:19:18.800 align:middle line:84% And then sometimes there's a "w." 00:19:18.800 --> 00:19:21.620 align:middle line:90% Old Irish lost out on the "w." 00:19:21.620 --> 00:19:23.420 align:middle line:84% And then you get to Basque and Turkish, 00:19:23.420 --> 00:19:26.750 align:middle line:84% where there's no "d" and no "t" and nothing rounded at all. 00:19:26.750 --> 00:19:27.660 align:middle line:90% Yeah. 00:19:27.660 --> 00:19:30.240 align:middle line:84% AUDIENCE: What does the colon mean? 00:19:30.240 --> 00:19:32.790 align:middle line:84% NORVIN RICHARDS: Oh, those indicate long vowels. 00:19:32.790 --> 00:19:34.170 align:middle line:90% Sorry. 00:19:34.170 --> 00:19:34.840 align:middle line:90% Good question. 00:19:34.840 --> 00:19:36.460 align:middle line:90% I should have said that. 00:19:36.460 --> 00:19:38.900 align:middle line:90% Yeah. 00:19:38.900 --> 00:19:40.900 align:middle line:84% Actually for Latin, I should have used a macron. 00:19:40.900 --> 00:19:42.317 align:middle line:84% I should go in there and fix that. 00:19:42.317 --> 00:19:45.020 align:middle line:90% 00:19:45.020 --> 00:19:49.010 align:middle line:84% So a hypothesis that people could entertain, then-- 00:19:49.010 --> 00:19:53.810 align:middle line:84% and this was famously entertained for the first time 00:19:53.810 --> 00:19:55.250 align:middle line:90% a few centuries ago-- 00:19:55.250 --> 00:20:00.230 align:middle line:84% was that it's not just the case that these various European 00:20:00.230 --> 00:20:03.740 align:middle line:84% languages have similar sounding words for one, two, and three, 00:20:03.740 --> 00:20:06.350 align:middle line:84% but that these words are similar sounding because they're 00:20:06.350 --> 00:20:07.460 align:middle line:90% related to each other. 00:20:07.460 --> 00:20:09.260 align:middle line:90% The word for this is cognates. 00:20:09.260 --> 00:20:12.710 align:middle line:84% These are all descendents of a common ancestor. 00:20:12.710 --> 00:20:16.790 align:middle line:84% And that Basque and Turkish are not part of this group. 00:20:16.790 --> 00:20:19.290 align:middle line:84% They don't have words that are cognate with those. 00:20:19.290 --> 00:20:22.760 align:middle line:84% I don't know why I specifically boxed the words for three. 00:20:22.760 --> 00:20:25.130 align:middle line:84% The claim is that the words for one and two 00:20:25.130 --> 00:20:28.310 align:middle line:84% and three in most of those languages in that table 00:20:28.310 --> 00:20:31.220 align:middle line:84% are cognates with each other, and that Basque and Turkish 00:20:31.220 --> 00:20:34.330 align:middle line:90% are the ones that are left out. 00:20:34.330 --> 00:20:37.170 align:middle line:84% So I just said, hey, look, these words look 00:20:37.170 --> 00:20:38.430 align:middle line:90% kind of similar to each other. 00:20:38.430 --> 00:20:41.358 align:middle line:84% And that is indeed often how this kind of thing starts. 00:20:41.358 --> 00:20:43.650 align:middle line:84% You look at two languages and say, hmm, these languages 00:20:43.650 --> 00:20:44.525 align:middle line:90% look kind of similar. 00:20:44.525 --> 00:20:46.470 align:middle line:90% I wonder if they're related. 00:20:46.470 --> 00:20:52.980 align:middle line:84% But big discovery from the early days of systematic linguistics 00:20:52.980 --> 00:20:55.230 align:middle line:84% was the observation that it's possible to be 00:20:55.230 --> 00:20:57.580 align:middle line:90% more systematic than this. 00:20:57.580 --> 00:21:01.200 align:middle line:84% So it isn't just a matter of saying, oh, look. 00:21:01.200 --> 00:21:03.790 align:middle line:84% These languages look similar to each other. 00:21:03.790 --> 00:21:06.810 align:middle line:84% We can actually state laws, generalizations, 00:21:06.810 --> 00:21:10.518 align:middle line:84% about which sounds correspond to which other sounds. 00:21:10.518 --> 00:21:12.060 align:middle line:84% So we can do better than just saying, 00:21:12.060 --> 00:21:13.380 align:middle line:90% these languages look similar. 00:21:13.380 --> 00:21:16.290 align:middle line:84% We get to say things like, for example, where 00:21:16.290 --> 00:21:19.320 align:middle line:84% Latin and Greek have a "d," English 00:21:19.320 --> 00:21:23.050 align:middle line:84% has a "t," for a certain set of words. 00:21:23.050 --> 00:21:29.460 align:middle line:84% So the Latin and Greek words for "two" and "eat" and "ten" 00:21:29.460 --> 00:21:33.480 align:middle line:84% have "d"s in them, which correspond in English to "t"s. 00:21:33.480 --> 00:21:37.560 align:middle line:84% Or the words in Latin and Greek that 00:21:37.560 --> 00:21:42.180 align:middle line:84% are related to "kin" and that second word means field. 00:21:42.180 --> 00:21:46.740 align:middle line:84% They have a "k" as corresponding "k" sound in English. 00:21:46.740 --> 00:21:49.350 align:middle line:84% These words that have a "b" in them-- 00:21:49.350 --> 00:21:52.380 align:middle line:84% words that have a "b" in them are weirdly rare 00:21:52.380 --> 00:21:54.597 align:middle line:84% in Proto-Indo-European, the ancestor 00:21:54.597 --> 00:21:55.680 align:middle line:90% of all of these languages. 00:21:55.680 --> 00:21:57.510 align:middle line:84% There are a couple of examples here. 00:21:57.510 --> 00:22:00.840 align:middle line:84% "Cannabis," the second line there, 00:22:00.840 --> 00:22:03.270 align:middle line:84% is thought to probably be a borrowing 00:22:03.270 --> 00:22:05.820 align:middle line:84% into Proto-Indo-European from some other language 00:22:05.820 --> 00:22:06.790 align:middle line:90% for several reasons. 00:22:06.790 --> 00:22:08.540 align:middle line:84% One is that it has a "b" in it, and that's 00:22:08.540 --> 00:22:09.810 align:middle line:90% rare in Proto-Indo-European. 00:22:09.810 --> 00:22:11.880 align:middle line:84% Another is that it's weirdly long. 00:22:11.880 --> 00:22:13.650 align:middle line:84% Proto-Indo-European words are mostly not 00:22:13.650 --> 00:22:15.270 align:middle line:90% three syllables long. 00:22:15.270 --> 00:22:20.160 align:middle line:84% So a hypothesis is that the ancestors of the Latins 00:22:20.160 --> 00:22:23.310 align:middle line:84% and the Romans and the Greeks and the Sanskrit speakers 00:22:23.310 --> 00:22:26.670 align:middle line:84% and the Gothic speakers, the Proto-Indo-Europeans, 00:22:26.670 --> 00:22:29.040 align:middle line:84% as they were roaming around Europe conquering 00:22:29.040 --> 00:22:32.010 align:middle line:84% various people, encountered some people who had discovered 00:22:32.010 --> 00:22:33.390 align:middle line:90% cannabis and were like, cool. 00:22:33.390 --> 00:22:34.420 align:middle line:90% We will take that. 00:22:34.420 --> 00:22:38.610 align:middle line:84% And while we're at it, we will borrow your word for it. 00:22:38.610 --> 00:22:40.980 align:middle line:84% Apparently, something like that happened. 00:22:40.980 --> 00:22:43.590 align:middle line:84% So there's this generalization-- it's 00:22:43.590 --> 00:22:46.110 align:middle line:84% called Grimm's Law because it was discovered 00:22:46.110 --> 00:22:50.280 align:middle line:84% by a Dane named Rasmus Rask and then discovered again 00:22:50.280 --> 00:22:54.930 align:middle line:84% by Jacob Grimm, who was one of the Grimm brothers of the Grimm 00:22:54.930 --> 00:22:55.770 align:middle line:90% Fairy Tales. 00:22:55.770 --> 00:22:58.740 align:middle line:84% This observation about systematic sound 00:22:58.740 --> 00:23:01.770 align:middle line:84% correspondences between the oldest versions 00:23:01.770 --> 00:23:05.310 align:middle line:90% of these related languages. 00:23:05.310 --> 00:23:08.640 align:middle line:84% You can see with the benefit of all of the phonology 00:23:08.640 --> 00:23:11.970 align:middle line:84% that we did at the beginning of this class that what's going on 00:23:11.970 --> 00:23:15.640 align:middle line:84% is that where Latin and Greek have voiced stops, 00:23:15.640 --> 00:23:18.270 align:middle line:84% English and other Germanic languages tend 00:23:18.270 --> 00:23:19.860 align:middle line:90% to have voiceless stops. 00:23:19.860 --> 00:23:22.790 align:middle line:84% So there was a general sound change. 00:23:22.790 --> 00:23:24.540 align:middle line:84% And there's more to Grimm's Law than that, 00:23:24.540 --> 00:23:26.340 align:middle line:84% but that's one of the observations 00:23:26.340 --> 00:23:29.850 align:middle line:90% that the Grimm made. 00:23:29.850 --> 00:23:32.310 align:middle line:84% Once we've looked at a bunch of languages 00:23:32.310 --> 00:23:35.310 align:middle line:84% and figured out all the sound laws that we need, 00:23:35.310 --> 00:23:37.980 align:middle line:84% we sort of get to posit these underlying forms. 00:23:37.980 --> 00:23:40.620 align:middle line:84% I was talking about this for Proto-Austronesian. 00:23:40.620 --> 00:23:44.190 align:middle line:84% These, you can think of them as being like the underlying forms 00:23:44.190 --> 00:23:46.680 align:middle line:84% we talked about when we did phonology, right? 00:23:46.680 --> 00:23:49.530 align:middle line:84% We posit these sound changes, and we 00:23:49.530 --> 00:23:52.720 align:middle line:84% have a sort of point of origin for all of these words. 00:23:52.720 --> 00:23:55.238 align:middle line:84% It's the form that undergoes different sound changes 00:23:55.238 --> 00:23:57.030 align:middle line:84% in different languages with the consequence 00:23:57.030 --> 00:23:58.770 align:middle line:84% that you have different languages. 00:23:58.770 --> 00:24:01.500 align:middle line:84% That underlying form, the language 00:24:01.500 --> 00:24:03.420 align:middle line:84% that has those underlying forms, is sometimes 00:24:03.420 --> 00:24:04.780 align:middle line:90% called the proto language. 00:24:04.780 --> 00:24:06.720 align:middle line:84% So I've been talking about Proto-Indo-European 00:24:06.720 --> 00:24:10.260 align:middle line:90% and Proto-Austronesian. 00:24:10.260 --> 00:24:13.320 align:middle line:84% Here's one of the examples of Grimm's Law at work. 00:24:13.320 --> 00:24:15.930 align:middle line:84% We look at Sanskrit and English and Latin. 00:24:15.930 --> 00:24:18.700 align:middle line:84% I don't know why I said them in that order. 00:24:18.700 --> 00:24:22.290 align:middle line:84% And we convince ourselves that where Sanskrit and Latin have 00:24:22.290 --> 00:24:25.170 align:middle line:90% a "d," English has a "t." 00:24:25.170 --> 00:24:29.250 align:middle line:84% That's part of Grimm's Law, so it 00:24:29.250 --> 00:24:34.170 align:middle line:84% has the consequence that voiced sounds in Sanskrit and Latin 00:24:34.170 --> 00:24:39.030 align:middle line:84% correlate with voiceless stops in English. 00:24:39.030 --> 00:24:45.240 align:middle line:84% If we look at the vowels, so we have an "a" in Sanskrit 00:24:45.240 --> 00:24:47.790 align:middle line:90% and an "e" in Latin. 00:24:47.790 --> 00:24:52.530 align:middle line:84% And if we look more widely at correlating words, 00:24:52.530 --> 00:24:53.970 align:middle line:84% words that look like they might be 00:24:53.970 --> 00:24:55.980 align:middle line:90% related in Sanskrit and Latin-- 00:24:55.980 --> 00:24:59.130 align:middle line:90% here are a bunch of examples. 00:24:59.130 --> 00:25:01.320 align:middle line:84% What kind of a sound do you think 00:25:01.320 --> 00:25:07.350 align:middle line:84% we should posit as the proto sound, a proto vowel for "eat" 00:25:07.350 --> 00:25:10.090 align:middle line:90% in both Sanskrit and Latin? 00:25:10.090 --> 00:25:13.810 align:middle line:84% Have a look at those data and tell me what you think. 00:25:13.810 --> 00:25:14.650 align:middle line:90% AUDIENCE: [SNEEZE] 00:25:14.650 --> 00:25:15.775 align:middle line:90% NORVIN RICHARDS: Bless you. 00:25:15.775 --> 00:25:19.220 align:middle line:90% 00:25:19.220 --> 00:25:21.770 align:middle line:84% Is there a generalization that we can draw of the form 00:25:21.770 --> 00:25:25.250 align:middle line:84% wherever Sanskrit has an "a," Latin has an "e"? 00:25:25.250 --> 00:25:27.940 align:middle line:90% 00:25:27.940 --> 00:25:29.890 align:middle line:84% That's true for "eat" and "tooth." 00:25:29.890 --> 00:25:33.110 align:middle line:90% 00:25:33.110 --> 00:25:37.660 align:middle line:84% And for the first one in "field." 00:25:37.660 --> 00:25:40.340 align:middle line:84% But we're seeing that for "sheep" and "two" 00:25:40.340 --> 00:25:42.310 align:middle line:84% and for the second vowel in "field," 00:25:42.310 --> 00:25:44.170 align:middle line:84% there are other Latin vowels that correspond 00:25:44.170 --> 00:25:45.790 align:middle line:90% with Sanskrit vowels. 00:25:45.790 --> 00:25:49.000 align:middle line:84% So what type of a vowel should we 00:25:49.000 --> 00:25:51.145 align:middle line:84% posited as the original vowel in "eat"? 00:25:51.145 --> 00:25:55.750 align:middle line:90% 00:25:55.750 --> 00:25:57.430 align:middle line:84% Can we predict the Sanskrit vowels-- 00:25:57.430 --> 00:25:59.740 align:middle line:84% or sorry, the Latin vowels from the Sanskrit vowels? 00:25:59.740 --> 00:26:02.600 align:middle line:90% 00:26:02.600 --> 00:26:03.100 align:middle line:90% No. 00:26:03.100 --> 00:26:03.400 align:middle line:90% Right? 00:26:03.400 --> 00:26:04.817 align:middle line:84% There aren't generalizations like, 00:26:04.817 --> 00:26:08.240 align:middle line:84% wherever Sanskrit has an "a," Latin as an "e." 00:26:08.240 --> 00:26:08.990 align:middle line:90% What should we do? 00:26:08.990 --> 00:26:10.150 align:middle line:90% Yeah. 00:26:10.150 --> 00:26:12.593 align:middle line:90% AUDIENCE: So we [INAUDIBLE]. 00:26:12.593 --> 00:26:13.510 align:middle line:90% NORVIN RICHARDS: Yeah. 00:26:13.510 --> 00:26:16.990 align:middle line:84% So if we do it the other way around, 00:26:16.990 --> 00:26:21.070 align:middle line:84% if we say the original vowels were more 00:26:21.070 --> 00:26:22.750 align:middle line:84% like the Latin vowels, and Sanskrit 00:26:22.750 --> 00:26:25.420 align:middle line:84% underwent a bunch of sound changes that squashed many 00:26:25.420 --> 00:26:26.380 align:middle line:90% of the vowels together. 00:26:26.380 --> 00:26:32.710 align:middle line:84% So the original vowels "e" and "o" and also "a," 00:26:32.710 --> 00:26:36.100 align:middle line:84% like the first vowel in "field," all three 00:26:36.100 --> 00:26:40.750 align:middle line:84% of those vowels became "a," "ah," in Sanskrit. 00:26:40.750 --> 00:26:42.310 align:middle line:84% This is the same kind of reasoning 00:26:42.310 --> 00:26:44.590 align:middle line:84% we were using when we were originally doing phonology. 00:26:44.590 --> 00:26:48.043 align:middle line:84% We can't predict the Latin vowels from the Sanskrit 00:26:48.043 --> 00:26:49.960 align:middle line:84% vowels, but we can predict the Sanskrit vowels 00:26:49.960 --> 00:26:50.770 align:middle line:90% from the Latin vowels. 00:26:50.770 --> 00:26:52.895 align:middle line:84% Namely, the Latin vowels all got squashed together. 00:26:52.895 --> 00:26:56.530 align:middle line:84% So Latin has kept an older vowel system 00:26:56.530 --> 00:27:00.940 align:middle line:84% that Sanskrit has simplified to a certain extent. 00:27:00.940 --> 00:27:04.560 align:middle line:84% So we'll posit a Proto-Indo-European stem, "ed-" 00:27:04.560 --> 00:27:06.000 align:middle line:90% for "eat." 00:27:06.000 --> 00:27:10.380 align:middle line:84% Sanskrit undergoes a change, where the "eh" becomes "ah." 00:27:10.380 --> 00:27:12.390 align:middle line:84% English undergoes Grimm's Law and also 00:27:12.390 --> 00:27:14.640 align:middle line:84% some changes with the vowels, which we'll get a chance 00:27:14.640 --> 00:27:16.200 align:middle line:90% to talk about in the second. 00:27:16.200 --> 00:27:19.170 align:middle line:84% And Latin preserves at least that much 00:27:19.170 --> 00:27:21.540 align:middle line:90% of the Proto-Indo-European form. 00:27:21.540 --> 00:27:22.920 align:middle line:90% It's worth being careful. 00:27:22.920 --> 00:27:26.010 align:middle line:84% Just like at the beginning of this class when 00:27:26.010 --> 00:27:30.930 align:middle line:84% we were doing phonology, we started off with sound changes, 00:27:30.930 --> 00:27:33.720 align:middle line:84% looking at data sets, where it was often the case 00:27:33.720 --> 00:27:37.620 align:middle line:84% that you could posit an underlying form that 00:27:37.620 --> 00:27:42.690 align:middle line:84% was the same as some form that you could see on the surface. 00:27:42.690 --> 00:27:45.510 align:middle line:84% But I think we saw, back when we were doing phonology, 00:27:45.510 --> 00:27:47.200 align:middle line:90% that wasn't necessarily true. 00:27:47.200 --> 00:27:49.890 align:middle line:84% There were cases where it was useful to posit an underlying 00:27:49.890 --> 00:27:54.000 align:middle line:84% form that always underwent some kind of change. 00:27:54.000 --> 00:27:56.910 align:middle line:84% It never just surfaced in its actual form. 00:27:56.910 --> 00:28:00.000 align:middle line:84% And this also happens when we're positing proto forms. 00:28:00.000 --> 00:28:02.910 align:middle line:84% It hasn't happened in this particular case, so this-- 00:28:02.910 --> 00:28:05.580 align:middle line:84% I carefully picked a case where life was simple, 00:28:05.580 --> 00:28:08.640 align:middle line:84% and Latin has just kept the Proto-Indo-European form. 00:28:08.640 --> 00:28:10.660 align:middle line:84% But we can't rely on that happening every time. 00:28:10.660 --> 00:28:13.118 align:middle line:84% Sometimes there are cases where we'll want a positive proto 00:28:13.118 --> 00:28:16.320 align:middle line:84% form that just doesn't survive in its unchanged 00:28:16.320 --> 00:28:17.940 align:middle line:90% form in any case. 00:28:17.940 --> 00:28:20.540 align:middle line:90% 00:28:20.540 --> 00:28:22.910 align:middle line:90% Sound changes. 00:28:22.910 --> 00:28:24.830 align:middle line:90% The study of sound changes-- 00:28:24.830 --> 00:28:28.340 align:middle line:84% first of all, the discovery that sound changes were regular, 00:28:28.340 --> 00:28:30.170 align:middle line:84% that you could make statements of the form 00:28:30.170 --> 00:28:32.090 align:middle line:84% wherever this language has sound X, 00:28:32.090 --> 00:28:34.820 align:middle line:84% this other language has sound Y, that 00:28:34.820 --> 00:28:39.200 align:middle line:84% was a big early discovery in linguistics and the starting 00:28:39.200 --> 00:28:42.830 align:middle line:84% point for a lot of the work that started getting done 00:28:42.830 --> 00:28:45.890 align:middle line:84% in phonology as people started getting 00:28:45.890 --> 00:28:48.440 align:middle line:84% interested in the questions of what kinds of sound changes 00:28:48.440 --> 00:28:51.050 align:middle line:84% there were and why some kinds of sound changes 00:28:51.050 --> 00:28:54.200 align:middle line:84% were more common than others and what motivated them 00:28:54.200 --> 00:28:56.400 align:middle line:90% and why they happened and so on. 00:28:56.400 --> 00:28:58.550 align:middle line:84% Similarly, the discussion that we 00:28:58.550 --> 00:29:00.920 align:middle line:84% had when we were doing phonology rule ordering, 00:29:00.920 --> 00:29:05.183 align:middle line:84% where we decided that it was useful sometimes to posit-- 00:29:05.183 --> 00:29:07.100 align:middle line:84% we talked about this for Lardil, for example-- 00:29:07.100 --> 00:29:08.630 align:middle line:84% that it's useful sometimes to posit 00:29:08.630 --> 00:29:10.790 align:middle line:84% an original form, an underlying form, 00:29:10.790 --> 00:29:14.150 align:middle line:84% and have it undergo sound changes in a particular order. 00:29:14.150 --> 00:29:16.860 align:middle line:84% I think we talked about it as being like an assembly line, 00:29:16.860 --> 00:29:17.360 align:middle line:90% right? 00:29:17.360 --> 00:29:20.340 align:middle line:84% It first goes to the place where it undergoes this sound change, 00:29:20.340 --> 00:29:22.700 align:middle line:84% then it undergoes the sound change, with the consequence 00:29:22.700 --> 00:29:26.090 align:middle line:84% sometimes that the second sound change creates something 00:29:26.090 --> 00:29:27.920 align:middle line:84% that would undergo the first sound change, 00:29:27.920 --> 00:29:28.910 align:middle line:90% but it's too late. 00:29:28.910 --> 00:29:30.470 align:middle line:90% We looked at examples like that. 00:29:30.470 --> 00:29:34.550 align:middle line:90% 00:29:34.550 --> 00:29:37.490 align:middle line:84% Way back when we were talking about Lardil, 00:29:37.490 --> 00:29:41.780 align:middle line:84% we had a rule that changed "u" to "a" 00:29:41.780 --> 00:29:44.510 align:middle line:84% at the ends of words that changed 00:29:44.510 --> 00:29:47.270 align:middle line:84% underlying "muwu," which is the underlying word for "water," 00:29:47.270 --> 00:29:51.350 align:middle line:90% to "muga." 00:29:51.350 --> 00:29:55.160 align:middle line:84% And we had another sound change that got rid of "k' 00:29:55.160 --> 00:29:57.303 align:middle line:90% at the ends of words. 00:29:57.303 --> 00:29:59.720 align:middle line:84% So if you had a "k" at the end of the word, it would drop, 00:29:59.720 --> 00:30:02.000 align:middle line:84% which changed underlying "naluk," 00:30:02.000 --> 00:30:07.492 align:middle line:84% which is the word for "story," to "nalu," 00:30:07.492 --> 00:30:09.950 align:middle line:84% But it was important that these rules applied in this order 00:30:09.950 --> 00:30:13.000 align:middle line:84% because "nalu" doesn't then change to "nala." 00:30:13.000 --> 00:30:14.542 align:middle line:84% So that was one of the kinds of cases 00:30:14.542 --> 00:30:16.917 align:middle line:84% we were talking about when we were talking about ordering 00:30:16.917 --> 00:30:18.140 align:middle line:90% rules in a particular order. 00:30:18.140 --> 00:30:22.310 align:middle line:84% All of that talk started in work on historical linguistics, 00:30:22.310 --> 00:30:25.310 align:middle line:84% where it's easy to imagine, in fact, 00:30:25.310 --> 00:30:27.440 align:middle line:84% that a bunch of sound changes have taken place, 00:30:27.440 --> 00:30:29.360 align:middle line:84% and they've taken place in a particular order. 00:30:29.360 --> 00:30:32.630 align:middle line:84% And people got interested in possible relations 00:30:32.630 --> 00:30:35.048 align:middle line:90% between sound changes. 00:30:35.048 --> 00:30:36.590 align:middle line:84% Some kinds of sound changes are cross 00:30:36.590 --> 00:30:38.010 align:middle line:90% linguistically fairly common. 00:30:38.010 --> 00:30:39.470 align:middle line:90% So here's one. 00:30:39.470 --> 00:30:43.470 align:middle line:84% "w" changed to "gwa," for example, in Chamorro. 00:30:43.470 --> 00:30:49.620 align:middle line:84% Chamorro is a language of Saipan in the Pacific. 00:30:49.620 --> 00:30:53.025 align:middle line:84% It's closely related to Tagalog, which is 00:30:53.025 --> 00:30:54.275 align:middle line:90% a language of the Philippines. 00:30:54.275 --> 00:30:56.533 align:middle line:84% They're both Austronesian languages. 00:30:56.533 --> 00:30:57.950 align:middle line:84% So for example, Tagalog has a word 00:30:57.950 --> 00:30:59.790 align:middle line:90% "asawa," which means "spouse." 00:30:59.790 --> 00:31:03.320 align:middle line:84% So Tagalog kinship terminology mostly 00:31:03.320 --> 00:31:05.060 align:middle line:84% doesn't indicate the gender of the person 00:31:05.060 --> 00:31:05.940 align:middle line:90% you're talking about. 00:31:05.940 --> 00:31:08.540 align:middle line:84% So they don't have words for husband or wife. 00:31:08.540 --> 00:31:11.660 align:middle line:84% They just have this word "asawa," which means spouse. 00:31:11.660 --> 00:31:14.810 align:middle line:84% You can say male spouse or female spouse, 00:31:14.810 --> 00:31:18.530 align:middle line:90% but you don't, usually. 00:31:18.530 --> 00:31:20.210 align:middle line:90% So Tagalog has a word, "asawa." 00:31:20.210 --> 00:31:22.070 align:middle line:90% In Chamorro, this is "asagwa." 00:31:22.070 --> 00:31:23.630 align:middle line:90% So the "w" has become a "gwa." 00:31:23.630 --> 00:31:26.870 align:middle line:84% Tagalog has a number, "dalawa," which is the number two. 00:31:26.870 --> 00:31:30.080 align:middle line:84% That corresponds to "hugwa" in Chamorro, 00:31:30.080 --> 00:31:32.430 align:middle line:84% and this "wala," which we were talking about before, 00:31:32.430 --> 00:31:34.610 align:middle line:84% which means "there isn't" just in Tagalog. 00:31:34.610 --> 00:31:37.790 align:middle line:84% All over Austronesian means "there is," and that's 00:31:37.790 --> 00:31:38.930 align:middle line:90% what it means in Chamorro. 00:31:38.930 --> 00:31:40.847 align:middle line:84% But it doesn't start with the "w" in Chamorro, 00:31:40.847 --> 00:31:42.740 align:middle line:90% it starts with "gwa." 00:31:42.740 --> 00:31:45.110 align:middle line:84% If you think about what a "w" is, 00:31:45.110 --> 00:31:48.080 align:middle line:84% the fact that a "w" becomes a "gwa" in Chamorro 00:31:48.080 --> 00:31:51.740 align:middle line:90% is maybe not such a weird fact. 00:31:51.740 --> 00:31:53.030 align:middle line:90% So think about what a "w" is. 00:31:53.030 --> 00:31:58.350 align:middle line:84% A "w" is a glide that corresponds to the vowel "u." 00:31:58.350 --> 00:32:01.800 align:middle line:84% If you think back, way, way back, months ago, 00:32:01.800 --> 00:32:06.030 align:middle line:84% when you were young and carefree and knew a lot about phonology 00:32:06.030 --> 00:32:10.320 align:middle line:84% and had not yet encountered syntax and semantics, 00:32:10.320 --> 00:32:14.370 align:middle line:84% maybe you remember that the value "u" has rounding, right? 00:32:14.370 --> 00:32:15.540 align:middle line:90% Your lips are rounding. 00:32:15.540 --> 00:32:18.400 align:middle line:90% But it's also high and back. 00:32:18.400 --> 00:32:20.730 align:middle line:84% What that means is that your tongue is bunched up 00:32:20.730 --> 00:32:22.800 align:middle line:84% toward the top of the back of your mouth. 00:32:22.800 --> 00:32:27.190 align:middle line:84% So back then, we were doing group exercises like saying 00:32:27.190 --> 00:32:29.620 align:middle line:90% "oo-wee" over and over again. 00:32:29.620 --> 00:32:32.547 align:middle line:84% And when you did that, you discovered 00:32:32.547 --> 00:32:34.380 align:middle line:84% that your lips were rounding and unrounding, 00:32:34.380 --> 00:32:35.880 align:middle line:84% but also that your tongue was shifting 00:32:35.880 --> 00:32:38.172 align:middle line:84% from the back of your mouth to the front of your mouth. 00:32:38.172 --> 00:32:42.600 align:middle line:84% So "w" is a glide that corresponds to the vowel "u." 00:32:42.600 --> 00:32:46.560 align:middle line:84% It's a version of the vowel "u" said very fast, basically, 00:32:46.560 --> 00:32:48.415 align:middle line:84% in which your tongue is bunching up 00:32:48.415 --> 00:32:50.040 align:middle line:84% toward the back of your mouth, reaching 00:32:50.040 --> 00:32:51.748 align:middle line:84% toward the top of the back of your mouth, 00:32:51.748 --> 00:32:53.650 align:middle line:90% and your lips are rounding. 00:32:53.650 --> 00:32:58.396 align:middle line:84% So if you think about what you do when you do a "g," a "guh," 00:32:58.396 --> 00:33:01.300 align:middle line:84% a voiced velar stop, well what you're doing 00:33:01.300 --> 00:33:04.480 align:middle line:84% is your tongue is touching the back of your mouth, right? 00:33:04.480 --> 00:33:07.520 align:middle line:84% It's making a velar closure back there. 00:33:07.520 --> 00:33:11.260 align:middle line:84% So the fact that a "wa" has become a "gwa" in this language 00:33:11.260 --> 00:33:12.570 align:middle line:90% is sort of understandable. 00:33:12.570 --> 00:33:15.850 align:middle line:84% A "gwa" is just a "w" that got out of hand, right? 00:33:15.850 --> 00:33:18.880 align:middle line:84% It's your tongue not just bunching up 00:33:18.880 --> 00:33:22.300 align:middle line:84% toward the back of your mouth, but actually colliding with it, 00:33:22.300 --> 00:33:25.540 align:middle line:90% making a closure back there. 00:33:25.540 --> 00:33:28.150 align:middle line:84% And I'm going through all this because this sound change 00:33:28.150 --> 00:33:31.090 align:middle line:90% didn't just happen in Chamorro. 00:33:31.090 --> 00:33:34.090 align:middle line:90% It also happened in Welsh. 00:33:34.090 --> 00:33:36.520 align:middle line:90% So Welsh has a bunch of words. 00:33:36.520 --> 00:33:38.680 align:middle line:84% I've just given you one, which is "man." 00:33:38.680 --> 00:33:40.480 align:middle line:84% The Welsh word for "man" is "gwir." 00:33:40.480 --> 00:33:42.897 align:middle line:84% It's related to the Latin word "vir," [pronounced "weer"], 00:33:42.897 --> 00:33:45.580 align:middle line:84% which we have as "virile" and English words like that. 00:33:45.580 --> 00:33:49.120 align:middle line:84% The "vir" is actually also, I think, related to the "were" 00:33:49.120 --> 00:33:51.880 align:middle line:84% in "werewolf," which is a man, a man-wolf. 00:33:51.880 --> 00:33:54.550 align:middle line:90% 00:33:54.550 --> 00:33:57.820 align:middle line:84% Welsh has another word which they borrowed from-- 00:33:57.820 --> 00:34:00.460 align:middle line:90% 00:34:00.460 --> 00:34:04.960 align:middle line:84% borrowed from Latin, "gwin," which is "wine." 00:34:04.960 --> 00:34:06.460 align:middle line:84% So they borrowed that from Latin. 00:34:06.460 --> 00:34:09.100 align:middle line:84% They changed the "w" sound at the beginning to a "gwa." 00:34:09.100 --> 00:34:11.139 align:middle line:84% So Latin underwent-- Welsh, sorry, 00:34:11.139 --> 00:34:13.052 align:middle line:90% underwent the same sound change. 00:34:13.052 --> 00:34:14.469 align:middle line:84% There are also some words that got 00:34:14.469 --> 00:34:18.489 align:middle line:84% borrowed from Proto-Germanic into late Latin, that 00:34:18.489 --> 00:34:21.130 align:middle line:84% is, Latin right before it was about to break up and turn 00:34:21.130 --> 00:34:24.639 align:middle line:84% into all the Romance languages, in which the Proto-Germanic 00:34:24.639 --> 00:34:27.580 align:middle line:84% "w" got realized in late Latin as a "gwa." 00:34:27.580 --> 00:34:30.864 align:middle line:84% So the Proto-Germanic word for "war," which was "werra"-- 00:34:30.864 --> 00:34:34.870 align:middle line:84% it's an ancestor of the modern English, "war"-- 00:34:34.870 --> 00:34:38.980 align:middle line:84% got borrowed into late Latin as "gwerra." 00:34:38.980 --> 00:34:43.929 align:middle line:84% And you now see that in English words like guerrilla, 00:34:43.929 --> 00:34:47.132 align:middle line:84% in the sense of a person who fights 00:34:47.132 --> 00:34:48.340 align:middle line:90% a particular type of warfare. 00:34:48.340 --> 00:34:51.250 align:middle line:90% 00:34:51.250 --> 00:34:54.010 align:middle line:84% If you study any Romance languages, 00:34:54.010 --> 00:35:00.730 align:middle line:84% the old Latin word for "war" was "bellum." 00:35:00.730 --> 00:35:03.010 align:middle line:84% But that word, I believe, vanished 00:35:03.010 --> 00:35:04.420 align:middle line:90% from all the Romance languages. 00:35:04.420 --> 00:35:07.370 align:middle line:84% What they all have are descendants of this word. 00:35:07.370 --> 00:35:12.760 align:middle line:84% So if you study Italian or French or Spanish, 00:35:12.760 --> 00:35:16.660 align:middle line:84% you need to learn words for war like "guerre" and "guerra." 00:35:16.660 --> 00:35:18.980 align:middle line:84% Those are all descendants of this Germanic word. 00:35:18.980 --> 00:35:20.890 align:middle line:84% So the late Latins, I guess, were really 00:35:20.890 --> 00:35:22.960 align:middle line:84% impressed with the Germanic attitude toward war. 00:35:22.960 --> 00:35:25.120 align:middle line:84% They borrowed this word, and they completely 00:35:25.120 --> 00:35:27.673 align:middle line:84% replaced their word for "war" with it, as far as I can tell. 00:35:27.673 --> 00:35:29.590 align:middle line:84% AUDIENCE: Do we know why "bellum" disappeared? 00:35:29.590 --> 00:35:30.270 align:middle line:90% NORVIN RICHARDS: No. 00:35:30.270 --> 00:35:30.960 align:middle line:90% At least, I don't. 00:35:30.960 --> 00:35:31.460 align:middle line:90% Yeah. 00:35:31.460 --> 00:35:34.200 align:middle line:84% This just became the cool word for war. 00:35:34.200 --> 00:35:36.630 align:middle line:84% So I mean, we have it in expressions 00:35:36.630 --> 00:35:39.150 align:middle line:84% that we've borrowed from Latin, like "antebellum." 00:35:39.150 --> 00:35:43.260 align:middle line:84% But as far as their normal word for war, 00:35:43.260 --> 00:35:45.540 align:middle line:90% it got replaced with this. 00:35:45.540 --> 00:35:47.670 align:middle line:84% Or we also have words in English where 00:35:47.670 --> 00:35:49.120 align:middle line:90% we have both of these words. 00:35:49.120 --> 00:35:50.820 align:middle line:84% So there's the Proto-Germanic word 00:35:50.820 --> 00:35:52.340 align:middle line:90% for "guard," which was "ward." 00:35:52.340 --> 00:35:55.170 align:middle line:84% We have that as a word, "ward," to ward off a threat 00:35:55.170 --> 00:35:56.940 align:middle line:90% or something like that. 00:35:56.940 --> 00:35:59.610 align:middle line:84% It got borrowed into late Latin as "gward." 00:35:59.610 --> 00:36:04.030 align:middle line:84% That's the modern word for "watch" in lots of languages. 00:36:04.030 --> 00:36:06.660 align:middle line:84% So in Italian, if you want to watch TV, 00:36:06.660 --> 00:36:09.000 align:middle line:90% the verb you use is "guardare." 00:36:09.000 --> 00:36:10.200 align:middle line:90% You guard the TV. 00:36:10.200 --> 00:36:11.310 align:middle line:90% You watch it carefully. 00:36:11.310 --> 00:36:13.980 align:middle line:90% 00:36:13.980 --> 00:36:16.920 align:middle line:84% And English has both of these words now, so we have "ward" 00:36:16.920 --> 00:36:18.210 align:middle line:90% and we also have "guard." 00:36:18.210 --> 00:36:20.520 align:middle line:90% They both have the same root. 00:36:20.520 --> 00:36:22.560 align:middle line:84% They're both from Proto-Germanic "ward," 00:36:22.560 --> 00:36:25.350 align:middle line:84% which came to us straight in the form of "ward," 00:36:25.350 --> 00:36:28.110 align:middle line:84% and then took a little detour into Latin to become "guard." 00:36:28.110 --> 00:36:31.110 align:middle line:84% And then we borrowed it again in a different form. 00:36:31.110 --> 00:36:34.740 align:middle line:90% 00:36:34.740 --> 00:36:36.000 align:middle line:90% Big discovery, then. 00:36:36.000 --> 00:36:38.010 align:middle line:84% Grimm's Law, this change from "w" 00:36:38.010 --> 00:36:42.323 align:middle line:84% to "gwa," so lots of big discoveries. 00:36:42.323 --> 00:36:43.740 align:middle line:84% But one of the big discoveries was 00:36:43.740 --> 00:36:46.032 align:middle line:84% that sound change was regular, that you could do better 00:36:46.032 --> 00:36:49.647 align:middle line:84% than just saying these words look kind of similar 00:36:49.647 --> 00:36:50.230 align:middle line:90% to each other. 00:36:50.230 --> 00:36:51.690 align:middle line:84% I bet these languages are related. 00:36:51.690 --> 00:36:53.160 align:middle line:84% You could actually say, oh, yeah. 00:36:53.160 --> 00:36:54.810 align:middle line:84% These words are-- these languages 00:36:54.810 --> 00:36:58.800 align:middle line:84% are related because there is this general law about how 00:36:58.800 --> 00:37:01.650 align:middle line:84% sounds in language X are related to sounds in language Y. 00:37:01.650 --> 00:37:04.680 align:middle line:84% This was called the Neogrammarian Hypothesis, 00:37:04.680 --> 00:37:06.720 align:middle line:84% and it was one of the big, early discoveries. 00:37:06.720 --> 00:37:10.230 align:middle line:84% It shifts emphasis away from looking 00:37:10.230 --> 00:37:12.752 align:middle line:84% at lists of words that look kind of similar. 00:37:12.752 --> 00:37:14.460 align:middle line:84% What we're looking for are lists of words 00:37:14.460 --> 00:37:17.043 align:middle line:84% that can be related by regular sound change laws, which is not 00:37:17.043 --> 00:37:18.040 align:middle line:90% the same thing. 00:37:18.040 --> 00:37:19.670 align:middle line:84% There could be lots of sound changes. 00:37:19.670 --> 00:37:21.420 align:middle line:84% The sound changes could be really radical. 00:37:21.420 --> 00:37:23.087 align:middle line:84% The result could be that the words don't 00:37:23.087 --> 00:37:24.430 align:middle line:90% look like each other at all. 00:37:24.430 --> 00:37:26.280 align:middle line:84% The point is to discover that there 00:37:26.280 --> 00:37:29.310 align:middle line:84% are regular correspondences between the words 00:37:29.310 --> 00:37:30.000 align:middle line:90% and the lists. 00:37:30.000 --> 00:37:31.120 align:middle line:90% Does that make sense? 00:37:31.120 --> 00:37:34.840 align:middle line:84% That's how you discover that two languages are related. 00:37:34.840 --> 00:37:37.530 align:middle line:84% So just to give you a dramatic example of this, 00:37:37.530 --> 00:37:39.630 align:middle line:84% here are three words for the number 00:37:39.630 --> 00:37:44.160 align:middle line:84% two in three languages whose names I have cunningly 00:37:44.160 --> 00:37:46.248 align:middle line:90% hidden from you. 00:37:46.248 --> 00:37:48.540 align:middle line:84% We'll call them language A, language B, and language C. 00:37:48.540 --> 00:37:50.332 align:middle line:84% And I will tell you that two of these words 00:37:50.332 --> 00:37:53.590 align:middle line:84% are related to each other and the third is not. 00:37:53.590 --> 00:37:55.690 align:middle line:84% You're probably suspecting a trick. 00:37:55.690 --> 00:37:58.540 align:middle line:84% But go ahead, somebody fall for it. 00:37:58.540 --> 00:38:00.190 align:middle line:84% Which of these two languages do you 00:38:00.190 --> 00:38:01.880 align:middle line:90% think are related to each other? 00:38:01.880 --> 00:38:02.380 align:middle line:90% Yes. 00:38:02.380 --> 00:38:03.160 align:middle line:90% AUDIENCE: B and C. 00:38:03.160 --> 00:38:05.202 align:middle line:84% NORVIN RICHARDS: B and C. Yes, see, you're right. 00:38:05.202 --> 00:38:08.500 align:middle line:84% But that's because you know how this kind of thing works. 00:38:08.500 --> 00:38:10.960 align:middle line:84% You're supposed to say A and B. Would somebody go ahead 00:38:10.960 --> 00:38:11.740 align:middle line:90% and say A and B? 00:38:11.740 --> 00:38:12.730 align:middle line:90% I would feel better. 00:38:12.730 --> 00:38:13.480 align:middle line:90% AUDIENCE: Is it A and B? 00:38:13.480 --> 00:38:16.063 align:middle line:84% NORVIN RICHARDS: Yes, yes, it's A and B. Excellent suggestion. 00:38:16.063 --> 00:38:17.380 align:middle line:90% Thank you. 00:38:17.380 --> 00:38:19.780 align:middle line:90% But of course, he's right. 00:38:19.780 --> 00:38:25.330 align:middle line:84% It is B and C. "Er" Is the Mandarin word for two. 00:38:25.330 --> 00:38:27.310 align:middle line:84% "Erku" is the Armenian word for two, 00:38:27.310 --> 00:38:31.120 align:middle line:84% and Greek has "duo" as it's word for two. 00:38:31.120 --> 00:38:33.730 align:middle line:84% And the way you convince yourself of this 00:38:33.730 --> 00:38:37.060 align:middle line:84% is by discovering that wherever Greek has a "dwa," 00:38:37.060 --> 00:38:40.893 align:middle line:84% Armenian has an "erk," which is a pretty radical sound change. 00:38:40.893 --> 00:38:42.310 align:middle line:84% It took people a while to discover 00:38:42.310 --> 00:38:44.800 align:middle line:84% that Armenian was an Indo-European language 00:38:44.800 --> 00:38:48.082 align:middle line:84% because Armenian has had really radical things happen 00:38:48.082 --> 00:38:50.290 align:middle line:84% to it, with the consequence that its words don't look 00:38:50.290 --> 00:38:52.330 align:middle line:90% all that Indo-European anymore. 00:38:52.330 --> 00:38:54.970 align:middle line:84% But it underwent this general sound change, 00:38:54.970 --> 00:38:57.715 align:middle line:84% and the point is you discover that the sound change is 00:38:57.715 --> 00:38:59.590 align:middle line:84% general, and that's how you convince yourself 00:38:59.590 --> 00:39:01.510 align:middle line:90% that it belongs to this club. 00:39:01.510 --> 00:39:02.050 align:middle line:90% Yeah. 00:39:02.050 --> 00:39:02.800 align:middle line:90% AUDIENCE: Why? 00:39:02.800 --> 00:39:05.330 align:middle line:84% NORVIN RICHARDS: Why did it do that? 00:39:05.330 --> 00:39:09.160 align:middle line:84% So there isn't any really good answer to that, 00:39:09.160 --> 00:39:10.840 align:middle line:90% but you could think. 00:39:10.840 --> 00:39:13.420 align:middle line:84% If you imagined first that they decided 00:39:13.420 --> 00:39:17.907 align:middle line:84% that it wasn't good to have words start with "d" and "w." 00:39:17.907 --> 00:39:19.990 align:middle line:84% Like, that's too many consonants at the beginning. 00:39:19.990 --> 00:39:20.830 align:middle line:90% We don't want that. 00:39:20.830 --> 00:39:22.600 align:middle line:90% We're going to add a vowel. 00:39:22.600 --> 00:39:25.750 align:middle line:84% So they'll add a vowel at the beginning. 00:39:25.750 --> 00:39:28.960 align:middle line:84% This alternation between "d" and "r" 00:39:28.960 --> 00:39:32.420 align:middle line:84% is not such a strange thing to do. 00:39:32.420 --> 00:39:34.990 align:middle line:90% So an "r" is kind of-- 00:39:34.990 --> 00:39:38.710 align:middle line:84% it's not unlike the relationship between "w" and "u", 00:39:38.710 --> 00:39:42.700 align:middle line:84% that "d" is an actual stop, an "ruh," it is unlike that. 00:39:42.700 --> 00:39:45.940 align:middle line:84% "d" is an actual stop, and "r," especially if it's a flap, 00:39:45.940 --> 00:39:48.550 align:middle line:84% is you're just sort of whacking the part of your mouth 00:39:48.550 --> 00:39:50.680 align:middle line:84% that where you make the "d" stop. 00:39:50.680 --> 00:39:53.440 align:middle line:84% And then the fact that "w" became a "k," well, 00:39:53.440 --> 00:39:59.320 align:middle line:84% we've seen "w" becoming a "gwa" in Welsh and in Chamorro. 00:39:59.320 --> 00:40:03.760 align:middle line:84% And a "k" has a stop in a place where a "w" would just 00:40:03.760 --> 00:40:04.690 align:middle line:90% be an approximant. 00:40:04.690 --> 00:40:07.900 align:middle line:84% So it's maybe not a completely crazy set of sound changes. 00:40:07.900 --> 00:40:10.540 align:middle line:84% I should say, it's not always possible to understand 00:40:10.540 --> 00:40:13.600 align:middle line:84% why a language underwent the sound changes that it did. 00:40:13.600 --> 00:40:14.843 align:middle line:90% There's a beautiful example. 00:40:14.843 --> 00:40:17.260 align:middle line:84% I have sometimes used this as a problem set in this class. 00:40:17.260 --> 00:40:18.670 align:middle line:84% If you were going to have another problem set, 00:40:18.670 --> 00:40:19.920 align:middle line:90% it might have been about this. 00:40:19.920 --> 00:40:25.523 align:middle line:84% There's an Algonquian language, Arapaho. 00:40:25.523 --> 00:40:27.190 align:middle line:84% If you look at the Algonquian languages, 00:40:27.190 --> 00:40:29.230 align:middle line:84% the Proto-Algonquian can be reconstructed 00:40:29.230 --> 00:40:33.370 align:middle line:90% to a fair degree of confidence. 00:40:33.370 --> 00:40:35.167 align:middle line:84% There are a bunch of Algonquian languages. 00:40:35.167 --> 00:40:36.250 align:middle line:90% They're very well studied. 00:40:36.250 --> 00:40:37.250 align:middle line:90% There are sound changes. 00:40:37.250 --> 00:40:39.490 align:middle line:84% It's not all that hard to relate them to each other. 00:40:39.490 --> 00:40:42.310 align:middle line:84% Arapaho is an Algonquian language, but it has undergone 00:40:42.310 --> 00:40:45.070 align:middle line:90% sound changes like-- 00:40:45.070 --> 00:40:49.510 align:middle line:84% I'm trying to remember them all. "p" has become "chuh." 00:40:49.510 --> 00:40:52.780 align:middle line:84% "m" has become-- what has "m" become? 00:40:52.780 --> 00:40:56.320 align:middle line:90% "m" has become "b." 00:40:56.320 --> 00:40:58.990 align:middle line:84% There's a sound-- so with the consequence 00:40:58.990 --> 00:41:01.030 align:middle line:84% that-- oh, no. "m" has become "chuh." 00:41:01.030 --> 00:41:03.910 align:middle line:84% They've both become "chuh." they've blended together. 00:41:03.910 --> 00:41:05.890 align:middle line:84% So the consequence is that the word 00:41:05.890 --> 00:41:08.320 align:middle line:84% for "dog," which is the one that I can remember off 00:41:08.320 --> 00:41:10.390 align:middle line:90% the top of my head. 00:41:10.390 --> 00:41:14.650 align:middle line:84% There's a word for "dog" that in Wampanoag, which is 00:41:14.650 --> 00:41:17.320 align:middle line:90% spoken around here, is "anum." 00:41:17.320 --> 00:41:18.280 align:middle line:90% This is a schwa. 00:41:18.280 --> 00:41:27.410 align:middle line:84% And in Passamaquoddy is "alum," and in Cree is "ateem." 00:41:27.410 --> 00:41:31.030 align:middle line:84% And so there's an "m' here that corresponds, 00:41:31.030 --> 00:41:33.280 align:middle line:84% and there are vowels here that have undergone 00:41:33.280 --> 00:41:34.630 align:middle line:90% various kinds of sound changes. 00:41:34.630 --> 00:41:36.960 align:middle line:84% There's a sound here, which is kind of mysterious. 00:41:36.960 --> 00:41:40.600 align:middle line:84% It's unclear how to represent it in Proto-Algonquian. 00:41:40.600 --> 00:41:42.940 align:middle line:90% But so here's one Wampanoag. 00:41:42.940 --> 00:41:45.910 align:middle line:90% Here's Passamaquoddy. 00:41:45.910 --> 00:41:48.310 align:middle line:90% Here's Cree. 00:41:48.310 --> 00:41:54.910 align:middle line:84% And the Arapaho word for it is "fench." 00:41:54.910 --> 00:41:55.450 align:middle line:90% Yeah. 00:41:55.450 --> 00:42:00.310 align:middle line:84% So you've got "anum," "alum," "ateem" and then "fench" 00:42:00.310 --> 00:42:01.670 align:middle line:90% is the Arapaho word. 00:42:01.670 --> 00:42:05.230 align:middle line:84% So the Arapahos, I don't know what happened to them. 00:42:05.230 --> 00:42:08.050 align:middle line:84% But they underwent sound changes that it's very hard 00:42:08.050 --> 00:42:09.700 align:middle line:90% to justify to yourself. 00:42:09.700 --> 00:42:11.470 align:middle line:90% It's unclear what happened. 00:42:11.470 --> 00:42:14.110 align:middle line:84% I was once talking with a linguist about this, 00:42:14.110 --> 00:42:16.570 align:middle line:84% and his theory was that what happened 00:42:16.570 --> 00:42:20.950 align:middle line:84% with the Arapaho was that they became a horse culture. 00:42:20.950 --> 00:42:23.080 align:middle line:84% When horses were brought to the US, 00:42:23.080 --> 00:42:25.090 align:middle line:84% the Arapaho got really into horses. 00:42:25.090 --> 00:42:27.790 align:middle line:84% And his theory was there must have 00:42:27.790 --> 00:42:30.550 align:middle line:84% been a generation of Arapaho who just sort of stopped 00:42:30.550 --> 00:42:33.718 align:middle line:84% listening to their parents, It underwent all kinds 00:42:33.718 --> 00:42:35.260 align:middle line:84% of radical sound changes because they 00:42:35.260 --> 00:42:36.820 align:middle line:90% were too busy riding horses. 00:42:36.820 --> 00:42:40.450 align:middle line:90% 00:42:40.450 --> 00:42:42.320 align:middle line:90% This is an example where-- 00:42:42.320 --> 00:42:42.820 align:middle line:90% thank you. 00:42:42.820 --> 00:42:48.850 align:middle line:84% One of you deliberately fell for the trap that I was setting. 00:42:48.850 --> 00:42:51.340 align:middle line:84% An example where you convince yourself 00:42:51.340 --> 00:42:55.810 align:middle line:84% that these two, the Mandarin and the Armenian words, 00:42:55.810 --> 00:42:58.930 align:middle line:84% are not related, despite having some stuff that looks like it's 00:42:58.930 --> 00:42:59.830 align:middle line:90% in common. 00:42:59.830 --> 00:43:04.340 align:middle line:84% It's really the Armenian and the Greek words that are related. 00:43:04.340 --> 00:43:07.250 align:middle line:84% Another example where you have two words that look similar, 00:43:07.250 --> 00:43:09.460 align:middle line:84% but the two languages are not related. 00:43:09.460 --> 00:43:17.410 align:middle line:84% There's a beautiful memoir by a guy named Dixon, Robert Dixon. 00:43:17.410 --> 00:43:20.180 align:middle line:90% He's a linguist who did-- 00:43:20.180 --> 00:43:23.680 align:middle line:84% he wrote a book called Memoirs of a Field Worker 00:43:23.680 --> 00:43:26.620 align:middle line:84% which was about his work in Aboriginal languages 00:43:26.620 --> 00:43:29.560 align:middle line:90% in Australia in the '60s. 00:43:29.560 --> 00:43:31.150 align:middle line:90% It's a nice book. 00:43:31.150 --> 00:43:32.650 align:middle line:84% And one of the things he talks about 00:43:32.650 --> 00:43:37.060 align:middle line:84% is going-- he was traveling around different Aboriginal 00:43:37.060 --> 00:43:39.130 align:middle line:84% areas, talking to different speakers 00:43:39.130 --> 00:43:40.180 align:middle line:90% of different languages. 00:43:40.180 --> 00:43:42.130 align:middle line:84% And there was a particular language 00:43:42.130 --> 00:43:45.205 align:middle line:84% that he managed to find possibly one of the last speakers of, 00:43:45.205 --> 00:43:48.100 align:middle line:84% a language called Mbabaram, which no longer 00:43:48.100 --> 00:43:49.150 align:middle line:90% has any speakers at all. 00:43:49.150 --> 00:43:50.710 align:middle line:90% This was in the '60s. 00:43:50.710 --> 00:43:52.660 align:middle line:84% And he was very interested in this language. 00:43:52.660 --> 00:43:54.100 align:middle line:84% It was supposed to be quite different from some 00:43:54.100 --> 00:43:55.558 align:middle line:84% of the languages that were near it. 00:43:55.558 --> 00:43:57.070 align:middle line:84% And he finally found a speaker who 00:43:57.070 --> 00:43:59.170 align:middle line:84% was finally willing to talk to him and sat down with this guy. 00:43:59.170 --> 00:44:00.790 align:middle line:84% And the guy said, OK, let me start teaching 00:44:00.790 --> 00:44:01.960 align:middle line:90% you some Mbabaram words. 00:44:01.960 --> 00:44:03.930 align:middle line:90% Do you know our word for "dog"? 00:44:03.930 --> 00:44:04.930 align:middle line:90% And he was like, OK, no. 00:44:04.930 --> 00:44:06.100 align:middle line:90% What's your word for "dog"? 00:44:06.100 --> 00:44:09.820 align:middle line:84% And the guy smiled, and he said, "dog." 00:44:09.820 --> 00:44:13.370 align:middle line:84% And Dixon was like, OK, this has to be a joke. 00:44:13.370 --> 00:44:15.980 align:middle line:84% But it turned out it was not a joke. 00:44:15.980 --> 00:44:18.340 align:middle line:84% Most of the languages of the world 00:44:18.340 --> 00:44:19.870 align:middle line:90% probably have a word for "dog." 00:44:19.870 --> 00:44:21.880 align:middle line:84% There are dogs in lots of places. 00:44:21.880 --> 00:44:25.990 align:middle line:84% And "d" and "o" and "g," they're not such uncommon sounds. 00:44:25.990 --> 00:44:29.920 align:middle line:84% And 6,000 whatever languages in the world, and yeah, 00:44:29.920 --> 00:44:32.090 align:middle line:84% this kind of thing happens every so often. 00:44:32.090 --> 00:44:34.400 align:middle line:84% So these are completely unrelated languages. 00:44:34.400 --> 00:44:38.945 align:middle line:84% Mbabaram "dog" can be related to-- 00:44:38.945 --> 00:44:41.470 align:middle line:84% what is that, a proto [INAUDIBLE] form, 00:44:41.470 --> 00:44:45.670 align:middle line:84% "gudaga," which changed in some related languages to other 00:44:45.670 --> 00:44:46.690 align:middle line:90% sounds. 00:44:46.690 --> 00:44:49.750 align:middle line:84% So you have "gudaga" in Yidiny, and you have "guda" in Dyirbal. 00:44:49.750 --> 00:44:52.162 align:middle line:84% And in Mbabaram, they got rid of the first syllable, 00:44:52.162 --> 00:44:53.620 align:middle line:84% and they got rid of the last vowel. 00:44:53.620 --> 00:44:55.420 align:middle line:84% And before you knew it, you had "dog." 00:44:55.420 --> 00:44:57.670 align:middle line:90% English also has a word "dog." 00:44:57.670 --> 00:45:00.382 align:middle line:90% It's not from Australia. 00:45:00.382 --> 00:45:01.840 align:middle line:84% So we have-- it originally referred 00:45:01.840 --> 00:45:06.130 align:middle line:84% to another kind, a particular kind of dog, a mastiff. 00:45:06.130 --> 00:45:08.320 align:middle line:84% And it's taken over as our word for "dog." 00:45:08.320 --> 00:45:12.508 align:middle line:84% Our older word for "dog" was "hound," which we still have. 00:45:12.508 --> 00:45:14.300 align:middle line:84% And this is the kind of thing that happens. 00:45:14.300 --> 00:45:17.350 align:middle line:84% So you know, Persian and English have the same word for "bad." 00:45:17.350 --> 00:45:19.840 align:middle line:84% Malay and Greek have very similar-looking words 00:45:19.840 --> 00:45:21.310 align:middle line:90% for "eye." 00:45:21.310 --> 00:45:23.740 align:middle line:90% And it's just a coincidence. 00:45:23.740 --> 00:45:26.350 align:middle line:84% How do you know whether you're looking at a coincidence? 00:45:26.350 --> 00:45:28.900 align:middle line:84% Well, you ask yourself whether you can posit regular sound 00:45:28.900 --> 00:45:30.010 align:middle line:90% changes. 00:45:30.010 --> 00:45:33.160 align:middle line:84% So English and Kaqchikel Mayan have words for "mess" 00:45:33.160 --> 00:45:35.710 align:middle line:84% that sound more or less the same. 00:45:35.710 --> 00:45:37.990 align:middle line:84% That's the kind of thing, if I just tell you 00:45:37.990 --> 00:45:39.700 align:middle line:84% in isolation I have two languages that 00:45:39.700 --> 00:45:42.130 align:middle line:84% have the same word for "mess," if you want to know, 00:45:42.130 --> 00:45:44.560 align:middle line:84% are those languages related, well the way you find out 00:45:44.560 --> 00:45:46.210 align:middle line:84% is by trying to find out whether you 00:45:46.210 --> 00:45:48.310 align:middle line:84% can come up with sound changes that 00:45:48.310 --> 00:45:50.160 align:middle line:90% apply to more than one word. 00:45:50.160 --> 00:45:52.660 align:middle line:84% Is it generally the case that where this language has an "m" 00:45:52.660 --> 00:45:53.952 align:middle line:90% this other language has an "m'? 00:45:53.952 --> 00:45:57.370 align:middle line:84% Where this language has an "s," this other language has an "s"? 00:45:57.370 --> 00:46:00.050 align:middle line:84% In the case of English and Kaqchikel, the answer is no. 00:46:00.050 --> 00:46:02.470 align:middle line:84% So here are a bunch of other words with "m"s in them, 00:46:02.470 --> 00:46:05.170 align:middle line:84% and they don't have "m"s in Kaqchikel. 00:46:05.170 --> 00:46:07.900 align:middle line:84% You can't rely on it being the case 00:46:07.900 --> 00:46:10.030 align:middle line:84% that every English word with an "m" in it 00:46:10.030 --> 00:46:12.610 align:middle line:84% would have an "m" in Kaqchikel, even 00:46:12.610 --> 00:46:14.150 align:middle line:90% if the languages were related. 00:46:14.150 --> 00:46:16.330 align:middle line:84% But you are entitled to hope that it'll 00:46:16.330 --> 00:46:18.460 align:middle line:90% happen more than once. 00:46:18.460 --> 00:46:20.830 align:middle line:84% So this looks like a coincidence. 00:46:20.830 --> 00:46:23.920 align:middle line:90% 00:46:23.920 --> 00:46:27.160 align:middle line:90% Any questions about any of this? 00:46:27.160 --> 00:46:29.470 align:middle line:84% Just as an in-class exercise, I wanted 00:46:29.470 --> 00:46:31.570 align:middle line:84% to try doing some historical linguistics, 00:46:31.570 --> 00:46:33.880 align:middle line:84% reconstructing some sound changes. 00:46:33.880 --> 00:46:35.810 align:middle line:84% Here are some Polynesian languages. 00:46:35.810 --> 00:46:38.380 align:middle line:84% So they are languages from the Polynesian branch 00:46:38.380 --> 00:46:41.890 align:middle line:84% of the very large Austronesian language family. 00:46:41.890 --> 00:46:43.990 align:middle line:84% Hawaiian, which is spoken in Hawaii, and Maori, 00:46:43.990 --> 00:46:46.990 align:middle line:84% which is from New Zealand, and Tongan, which is from Tonga, 00:46:46.990 --> 00:46:50.080 align:middle line:90% and Samoan, which is from Samoa. 00:46:50.080 --> 00:46:53.500 align:middle line:84% These are all languages spoken in islands of the Pacific, 00:46:53.500 --> 00:46:54.895 align:middle line:84% particularly the eastern Pacific, 00:46:54.895 --> 00:46:57.280 align:middle line:90% or not too far from Hawaii-- 00:46:57.280 --> 00:46:59.680 align:middle line:90% between Hawaii and New Zealand. 00:46:59.680 --> 00:47:02.290 align:middle line:84% And these are all words that look pretty related 00:47:02.290 --> 00:47:03.200 align:middle line:90% to each other. 00:47:03.200 --> 00:47:04.810 align:middle line:84% And so let's see if we can figure out 00:47:04.810 --> 00:47:06.400 align:middle line:90% any sound correspondences. 00:47:06.400 --> 00:47:09.680 align:middle line:90% 00:47:09.680 --> 00:47:12.680 align:middle line:84% Maybe we can start by looking at the things 00:47:12.680 --> 00:47:13.925 align:middle line:90% that I have made green here. 00:47:13.925 --> 00:47:16.460 align:middle line:90% 00:47:16.460 --> 00:47:17.390 align:middle line:90% What do you think? 00:47:17.390 --> 00:47:20.090 align:middle line:84% What sort of proto sound should we 00:47:20.090 --> 00:47:24.015 align:middle line:84% posit for the sound that is in green in these kinds 00:47:24.015 --> 00:47:24.515 align:middle line:90% of examples? 00:47:24.515 --> 00:47:28.897 align:middle line:90% 00:47:28.897 --> 00:47:31.480 align:middle line:84% There are at least two theories that we could take seriously-- 00:47:31.480 --> 00:47:33.340 align:middle line:84% hypotheses we could take seriously, right? 00:47:33.340 --> 00:47:35.830 align:middle line:84% There's a third kind of hypothesis. 00:47:35.830 --> 00:47:39.280 align:middle line:90% Maybe it's an "m," right? 00:47:39.280 --> 00:47:41.990 align:middle line:84% But let's put that possibility aside for a second. 00:47:41.990 --> 00:47:44.270 align:middle line:84% We're not going to do that unless we're driven to it. 00:47:44.270 --> 00:47:46.270 align:middle line:84% So the two theories that we could take seriously 00:47:46.270 --> 00:47:51.110 align:middle line:84% are maybe it's a "k" or maybe it's a "t." 00:47:51.110 --> 00:47:54.920 align:middle line:84% Anybody have a theory about what we ought to do? 00:47:54.920 --> 00:47:56.390 align:middle line:90% Joseph? 00:47:56.390 --> 00:48:03.010 align:middle line:84% AUDIENCE: [INAUDIBLE] --that maybe the majority rules 00:48:03.010 --> 00:48:03.730 align:middle line:90% and it's the "t." 00:48:03.730 --> 00:48:06.063 align:middle line:84% NORVIN RICHARDS: Yes, so we could do this democratically 00:48:06.063 --> 00:48:09.400 align:middle line:84% and say, yeah, there are more "t"s on this slide than there 00:48:09.400 --> 00:48:11.140 align:middle line:90% are "k"s. 00:48:11.140 --> 00:48:12.550 align:middle line:90% Majority rules. 00:48:12.550 --> 00:48:13.425 align:middle line:90% Yeah. 00:48:13.425 --> 00:48:14.880 align:middle line:84% AUDIENCE: I would also go with "t" 00:48:14.880 --> 00:48:20.300 align:middle line:84% because Hawaii is a more geographically isolated area. 00:48:20.300 --> 00:48:21.290 align:middle line:90% NORVIN RICHARDS: Yeah. 00:48:21.290 --> 00:48:24.260 align:middle line:84% All of these-- I mean, these are all islands. 00:48:24.260 --> 00:48:25.550 align:middle line:90% Amazing story. 00:48:25.550 --> 00:48:29.060 align:middle line:84% These people started-- the Proto-Austronesians 00:48:29.060 --> 00:48:30.710 align:middle line:84% are supposed to have started in Taiwan 00:48:30.710 --> 00:48:34.280 align:middle line:84% and headed out in outrigger canoes 00:48:34.280 --> 00:48:37.250 align:middle line:84% and settled all of these islands in the Pacific, 00:48:37.250 --> 00:48:40.100 align:middle line:84% tossing consonants overboard when they got to too heavy, 00:48:40.100 --> 00:48:42.362 align:middle line:90% as far as I can tell. 00:48:42.362 --> 00:48:43.070 align:middle line:90% But you're right. 00:48:43.070 --> 00:48:45.860 align:middle line:84% Hawaiian was near the end of their journey. 00:48:45.860 --> 00:48:48.800 align:middle line:84% So they're supposed to have started in Taiwan. 00:48:48.800 --> 00:48:51.470 align:middle line:84% New Zealand and Tonga and Samoa were 00:48:51.470 --> 00:48:54.210 align:middle line:84% all closer to where they started than Hawaii. 00:48:54.210 --> 00:48:57.590 align:middle line:84% So maybe Hawaii-- we're justified to think 00:48:57.590 --> 00:48:59.750 align:middle line:84% that Hawaiian might have undergone some 00:48:59.750 --> 00:49:03.170 align:middle line:90% of the special sound changes. 00:49:03.170 --> 00:49:05.720 align:middle line:84% I can also tell you something special about Hawaiian. 00:49:05.720 --> 00:49:08.660 align:middle line:84% I don't know whether this will come up later. 00:49:08.660 --> 00:49:10.580 align:middle line:90% Hawaiian does not have a "t." 00:49:10.580 --> 00:49:14.420 align:middle line:84% It's one of the few languages in the world that does not, 00:49:14.420 --> 00:49:17.060 align:middle line:84% which is why, if you've ever heard, 00:49:17.060 --> 00:49:21.740 align:middle line:84% there's a song that is sometimes played around Christmas 00:49:21.740 --> 00:49:25.550 align:middle line:84% which involves the Hawaiian version of "Merry Christmas"-- 00:49:25.550 --> 00:49:29.920 align:middle line:90% "Mele Kalikimaka." 00:49:29.920 --> 00:49:33.890 align:middle line:84% "Mele Kalikimaka" is the Hawaiian version 00:49:33.890 --> 00:49:36.440 align:middle line:84% of "Merry Christmas" because Hawaiian has no t," 00:49:36.440 --> 00:49:39.830 align:middle line:84% and it also has no "s," So they need 00:49:39.830 --> 00:49:42.260 align:middle line:84% to replace the "s" in "Christmas" with something, 00:49:42.260 --> 00:49:44.720 align:middle line:84% and what they replace it with is "k" because that's 00:49:44.720 --> 00:49:45.660 align:middle line:90% the best they can do. 00:49:45.660 --> 00:49:48.980 align:middle line:84% That's the closest they've got to that. 00:49:48.980 --> 00:49:50.570 align:middle line:90% So yes, Hawaiian has no "t." 00:49:50.570 --> 00:49:52.820 align:middle line:84% So the idea that maybe there was a t," 00:49:52.820 --> 00:49:55.400 align:middle line:84% and in Hawaiian, it became a "k," 00:49:55.400 --> 00:49:57.860 align:middle line:90% has some initial plausibility. 00:49:57.860 --> 00:49:59.930 align:middle line:84% So yeah, we'll posit a sound change in Hawaiian 00:49:59.930 --> 00:50:02.400 align:middle line:90% from "t" to "k." 00:50:02.400 --> 00:50:05.850 align:middle line:84% And we'll posit a some Proto-Polynesian forms 00:50:05.850 --> 00:50:07.260 align:middle line:90% that will have "t" in them. 00:50:07.260 --> 00:50:09.690 align:middle line:84% That is, they'll look like the Maori, Tongan, and Samoan 00:50:09.690 --> 00:50:12.180 align:middle line:84% forms, and then we'll posit the sound change 00:50:12.180 --> 00:50:16.330 align:middle line:84% in Hawaiian that changes that "t" into a "k." 00:50:16.330 --> 00:50:17.980 align:middle line:90% So far, so good. 00:50:17.980 --> 00:50:21.620 align:middle line:90% Here are some more forms. 00:50:21.620 --> 00:50:22.910 align:middle line:90% What should we do with these? 00:50:22.910 --> 00:50:28.982 align:middle line:90% 00:50:28.982 --> 00:50:30.815 align:middle line:84% So it's either a "k" or it's a glottal stop. 00:50:30.815 --> 00:50:40.630 align:middle line:90% 00:50:40.630 --> 00:50:46.000 align:middle line:84% Maybe for now we should just sit on this problem because here, 00:50:46.000 --> 00:50:48.070 align:middle line:84% just this principle of doing this kind of thing 00:50:48.070 --> 00:50:50.350 align:middle line:84% democratically, maybe it doesn't help us. 00:50:50.350 --> 00:50:53.975 align:middle line:84% We've got two votes for "k" and two votes for glottal stop. 00:50:53.975 --> 00:50:55.600 align:middle line:84% This might be a place where it's easier 00:50:55.600 --> 00:50:58.690 align:middle line:84% to imagine a "k" becoming a glottal stop 00:50:58.690 --> 00:51:02.200 align:middle line:84% than it is to imagine a glottal stop becoming a "k." 00:51:02.200 --> 00:51:05.110 align:middle line:84% "k"s becoming glottal stops, that's a thing that happens 00:51:05.110 --> 00:51:08.980 align:middle line:84% pretty commonly in various languages. 00:51:08.980 --> 00:51:12.470 align:middle line:84% And in fact, that's what people standardly do in this case. 00:51:12.470 --> 00:51:17.860 align:middle line:84% So in Hawaiian, we posit a "k" changing to a global stop 00:51:17.860 --> 00:51:20.560 align:middle line:90% in these kinds of examples. 00:51:20.560 --> 00:51:22.630 align:middle line:84% Maori and Tongan are the ones that 00:51:22.630 --> 00:51:23.890 align:middle line:90% are being conservative here. 00:51:23.890 --> 00:51:26.480 align:middle line:84% We'll see an additional reason to say that in just a second, 00:51:26.480 --> 00:51:26.980 align:middle line:90% I think. 00:51:26.980 --> 00:51:31.520 align:middle line:90% 00:51:31.520 --> 00:51:33.890 align:middle line:84% Here's the additional reason to say that. 00:51:33.890 --> 00:51:35.960 align:middle line:84% Here are a bunch more words where 00:51:35.960 --> 00:51:41.660 align:middle line:84% Tongan has glottal stops in unpredictable places. 00:51:41.660 --> 00:51:43.445 align:middle line:84% So at the beginning of the word for "day" 00:51:43.445 --> 00:51:46.070 align:middle line:84% and twice in the word for "love" and in the middle of the words 00:51:46.070 --> 00:51:49.025 align:middle line:90% for "leg" and "voice." 00:51:49.025 --> 00:51:51.650 align:middle line:84% And it's not because-- so if you want to learn to speak Tongan, 00:51:51.650 --> 00:51:53.108 align:middle line:84% one of the things you have to learn 00:51:53.108 --> 00:51:55.610 align:middle line:84% is that some words begin with glottal stops, 00:51:55.610 --> 00:51:57.110 align:middle line:90% and others begin with vowels. 00:51:57.110 --> 00:52:00.320 align:middle line:84% So "dawn" starts with a vowel, and "day" starts 00:52:00.320 --> 00:52:01.650 align:middle line:90% with a glottal stop. 00:52:01.650 --> 00:52:05.300 align:middle line:84% And distinguishing these from each other is tricky. 00:52:05.300 --> 00:52:08.150 align:middle line:84% The fact that Tongan is putting these glottal stops in places 00:52:08.150 --> 00:52:09.560 align:middle line:90% that can't be predicted-- 00:52:09.560 --> 00:52:10.100 align:middle line:90% [SNEEZE] 00:52:10.100 --> 00:52:10.880 align:middle line:90% Bless you. 00:52:10.880 --> 00:52:13.700 align:middle line:90% 00:52:13.700 --> 00:52:16.280 align:middle line:84% Leads us to think that what's going on 00:52:16.280 --> 00:52:20.120 align:middle line:84% is that those were glottal stops in Proto-Polynesian, 00:52:20.120 --> 00:52:22.920 align:middle line:84% and they've been lost in all the other languages. 00:52:22.920 --> 00:52:26.480 align:middle line:84% So here's place where predictability 00:52:26.480 --> 00:52:29.000 align:middle line:90% trumps democracy. 00:52:29.000 --> 00:52:32.960 align:middle line:84% We can't predict where Tongan will have these glottal stops. 00:52:32.960 --> 00:52:36.680 align:middle line:84% We can predict where Hawaiian, Maori, and Samoan 00:52:36.680 --> 00:52:37.880 align:middle line:90% will have glottal stops. 00:52:37.880 --> 00:52:41.510 align:middle line:84% So Hawaiian and Samoan will have glottal stops 00:52:41.510 --> 00:52:44.810 align:middle line:84% wherever there's a "k" in Maori and Tongan. 00:52:44.810 --> 00:52:48.530 align:middle line:84% And Maori just won't have glottal stops. 00:52:48.530 --> 00:52:50.960 align:middle line:84% So it's always predictable where the glottal stops 00:52:50.960 --> 00:52:52.100 align:middle line:90% are in the other languages. 00:52:52.100 --> 00:52:54.017 align:middle line:84% It's not predictable where they are in Tongan, 00:52:54.017 --> 00:52:56.280 align:middle line:84% so we'll posit them for the proto language. 00:52:56.280 --> 00:53:00.170 align:middle line:84% So Aloha in Hawaiian, which literally means "love," 00:53:00.170 --> 00:53:05.600 align:middle line:84% in Proto-Polynesian is "'alo'ofa" with two glottal 00:53:05.600 --> 00:53:06.830 align:middle line:90% stops in it. 00:53:06.830 --> 00:53:09.200 align:middle line:84% And you can see, incidentally, another sound change. 00:53:09.200 --> 00:53:14.980 align:middle line:84% The "f" became "h" in Hawaiian and also in Maori. 00:53:14.980 --> 00:53:17.230 align:middle line:84% So Hawaiian has undergone sound changes, 00:53:17.230 --> 00:53:19.630 align:middle line:84% "k" becoming glottal stops, "t" becoming "k," 00:53:19.630 --> 00:53:22.240 align:middle line:90% and glottal stop vanishing. 00:53:22.240 --> 00:53:25.000 align:middle line:84% This is sometimes-- so I'll just do that again dramatically 00:53:25.000 --> 00:53:28.340 align:middle line:90% with a sagittal section here. 00:53:28.340 --> 00:53:32.410 align:middle line:84% So Proto-Polynesian "ata," "dawn," has become "aka." 00:53:32.410 --> 00:53:35.080 align:middle line:84% In Hawaiian, "t" has become "k," with the consequence 00:53:35.080 --> 00:53:37.480 align:middle line:90% that Hawaiian has no "t." 00:53:37.480 --> 00:53:40.840 align:middle line:84% "k" has become glottal stop, so Proto-Polynesian "kula," 00:53:40.840 --> 00:53:44.320 align:middle line:84% which means "red," has become "'ula" in Hawaiian. 00:53:44.320 --> 00:53:47.920 align:middle line:84% And Proto-Polynesian glottal stop has vanished. 00:53:47.920 --> 00:53:51.070 align:middle line:84% This is sometimes called a chain shift. 00:53:51.070 --> 00:53:55.300 align:middle line:84% And there are different ways of thinking about chain shifts, 00:53:55.300 --> 00:53:56.890 align:middle line:84% but they're not-- cross-linguistically 00:53:56.890 --> 00:53:57.490 align:middle line:90% not uncommon. 00:53:57.490 --> 00:53:59.380 align:middle line:84% I'll show you another kind of example of one 00:53:59.380 --> 00:54:03.730 align:middle line:84% in just a second, where you have sound A becoming sound B, 00:54:03.730 --> 00:54:08.050 align:middle line:84% sound B becoming sound C, and sound C becoming sound D. 00:54:08.050 --> 00:54:13.420 align:middle line:84% And it's often difficult to determine the order in which 00:54:13.420 --> 00:54:14.770 align:middle line:90% things like this happen. 00:54:14.770 --> 00:54:17.980 align:middle line:84% One way to think about this is the Hawaiians got rid 00:54:17.980 --> 00:54:19.810 align:middle line:84% of Proto-Polynesian glottal stop, 00:54:19.810 --> 00:54:22.810 align:middle line:84% and then they were like, now that we have no glottal stop, 00:54:22.810 --> 00:54:26.137 align:middle line:84% we're free to pronounce our "k" as glottally as we like. 00:54:26.137 --> 00:54:27.970 align:middle line:84% We don't have to worry about the "k" getting 00:54:27.970 --> 00:54:29.710 align:middle line:90% confused with the glottal stop. 00:54:29.710 --> 00:54:31.630 align:middle line:84% And so the "k" began drifting backwards 00:54:31.630 --> 00:54:32.920 align:middle line:90% and becoming a glottal stop. 00:54:32.920 --> 00:54:34.450 align:middle line:84% And then they were like, wait, wait. 00:54:34.450 --> 00:54:35.720 align:middle line:90% We have no "k" anymore. 00:54:35.720 --> 00:54:36.460 align:middle line:90% Our "k" is gone. 00:54:36.460 --> 00:54:37.450 align:middle line:90% It's just become a glottal stop. 00:54:37.450 --> 00:54:38.050 align:middle line:90% We miss "k." 00:54:38.050 --> 00:54:40.510 align:middle line:84% We will change our "t" into a "k." 00:54:40.510 --> 00:54:42.400 align:middle line:84% And then who knows what would happen, 00:54:42.400 --> 00:54:44.260 align:middle line:84% how they would manage to come up with a "t"? 00:54:44.260 --> 00:54:45.760 align:middle line:84% Maybe if we keep an eye on Hawaiian, 00:54:45.760 --> 00:54:47.830 align:middle line:90% we'll get to see what they do. 00:54:47.830 --> 00:54:50.440 align:middle line:84% It's at least one way to talk about these kinds of chain 00:54:50.440 --> 00:54:52.995 align:middle line:90% shifts. 00:54:52.995 --> 00:54:54.370 align:middle line:84% Another famous chain shift, which 00:54:54.370 --> 00:54:56.260 align:middle line:84% is called the Great English Vowel Shift-- 00:54:56.260 --> 00:54:58.930 align:middle line:84% sorry, the English Great Vowel Shift. 00:54:58.930 --> 00:55:05.230 align:middle line:84% It is the reason that if you study almost any other language 00:55:05.230 --> 00:55:07.960 align:middle line:84% of Europe, you'll learn vowels that 00:55:07.960 --> 00:55:11.680 align:middle line:84% have a certain set of sounds associated with them. 00:55:11.680 --> 00:55:13.300 align:middle line:84% They are the sounds, more or less, 00:55:13.300 --> 00:55:15.550 align:middle line:84% that those symbols have in the IPA. 00:55:15.550 --> 00:55:17.770 align:middle line:84% So in lots of languages, this symbol 00:55:17.770 --> 00:55:20.710 align:middle line:84% is used for a vowel that's pronounced "ee." 00:55:20.710 --> 00:55:25.360 align:middle line:84% But in English, it's pronounced "I." 00:55:25.360 --> 00:55:29.980 align:middle line:84% Or that in lots of languages, this symbol is used for a vowel 00:55:29.980 --> 00:55:33.940 align:middle line:84% that's pronounced "a," but in English, it's pronounced "ee." 00:55:33.940 --> 00:55:36.365 align:middle line:84% This is all because of the Great English Vowel Shift. 00:55:36.365 --> 00:55:37.990 align:middle line:84% What was the Great English Vowel Shift? 00:55:37.990 --> 00:55:39.220 align:middle line:90% It went like this. 00:55:39.220 --> 00:55:41.860 align:middle line:84% Around the 14th century, English had 00:55:41.860 --> 00:55:43.660 align:middle line:90% long vowels in these positions. 00:55:43.660 --> 00:55:45.310 align:middle line:90% English also had short vowels. 00:55:45.310 --> 00:55:46.420 align:middle line:90% We'll put those aside. 00:55:46.420 --> 00:55:48.420 align:middle line:84% So there were long vowels, that were just vowels 00:55:48.420 --> 00:55:50.440 align:middle line:90% that were held for a long time. 00:55:50.440 --> 00:55:54.280 align:middle line:84% And they were in those positions in the mouth. 00:55:54.280 --> 00:55:56.980 align:middle line:84% So you had an "ee," an "ay," an "aa," 00:55:56.980 --> 00:55:59.590 align:middle line:84% and then an "ah," "aw," "oh," and "oo." 00:55:59.590 --> 00:56:01.780 align:middle line:90% We had all those long vowels. 00:56:01.780 --> 00:56:09.710 align:middle line:84% And then the "ee" and the "oo" became dipthongs. 00:56:09.710 --> 00:56:13.160 align:middle line:84% If you think about holding an "ee" for a long time, 00:56:13.160 --> 00:56:16.870 align:middle line:84% your tongue is in a position of maximum tension. 00:56:16.870 --> 00:56:19.480 align:middle line:84% It's front and high, and you're supposed 00:56:19.480 --> 00:56:20.893 align:middle line:90% to hold it for a long time. 00:56:20.893 --> 00:56:22.810 align:middle line:84% And you can understand how something like that 00:56:22.810 --> 00:56:24.268 align:middle line:84% might have changed into a dipthong. 00:56:24.268 --> 00:56:26.800 align:middle line:84% So instead of holding an "ee" for a long time, 00:56:26.800 --> 00:56:29.350 align:middle line:84% you were allowing yourself to begin 00:56:29.350 --> 00:56:31.300 align:middle line:84% the vowel with your tongue in some more 00:56:31.300 --> 00:56:33.490 align:middle line:84% central position and then just sort of gesture 00:56:33.490 --> 00:56:34.690 align:middle line:90% in the direction of an "ee." 00:56:34.690 --> 00:56:39.520 align:middle line:84% So you went from "ee" to "uee" to "ayee" to "ai." 00:56:39.520 --> 00:56:44.710 align:middle line:84% And so "ee" and "oo" became "ai" and "ow." 00:56:44.710 --> 00:56:46.930 align:middle line:84% And then suddenly, English was like, wait. 00:56:46.930 --> 00:56:48.520 align:middle line:84% Didn't we have an "ee" and an "oo"? 00:56:48.520 --> 00:56:49.810 align:middle line:90% Our "ee" and "oo" are gone. 00:56:49.810 --> 00:56:52.030 align:middle line:90% All we have is "ai" and "ow." 00:56:52.030 --> 00:56:55.250 align:middle line:90% And so "ay" and "oh" changed. 00:56:55.250 --> 00:56:59.410 align:middle line:84% So "ay" became "ee" and "oh" became "oo," 00:56:59.410 --> 00:57:05.980 align:middle line:84% filling the gap that the first part of the vowel shift left. 00:57:05.980 --> 00:57:09.370 align:middle line:84% And then long "aa" and "ah" both became "ay." 00:57:09.370 --> 00:57:16.420 align:middle line:84% That's why long "a" in a word like "fate" is pronounced "ay." 00:57:16.420 --> 00:57:22.070 align:middle line:84% It's because, well, the "ay" was off becoming an "ee," 00:57:22.070 --> 00:57:24.340 align:middle line:84% and so now we had a space where there was an "ay." 00:57:24.340 --> 00:57:26.395 align:middle line:90% And so our "ah" became an "ay." 00:57:26.395 --> 00:57:29.260 align:middle line:90% And our "aw" became an "oh." 00:57:29.260 --> 00:57:33.790 align:middle line:84% So all of our vowels have undergone this dance. 00:57:33.790 --> 00:57:38.440 align:middle line:84% And if you look at modern English dialects, 00:57:38.440 --> 00:57:40.750 align:middle line:84% you can see a lot of these kinds of sound changes 00:57:40.750 --> 00:57:41.660 align:middle line:90% continuing today. 00:57:41.660 --> 00:57:44.500 align:middle line:84% So I come from a place where those dipthongs are 00:57:44.500 --> 00:57:51.460 align:middle line:84% getting flattened, where "ai" has a tendency to become "aa." 00:57:51.460 --> 00:57:55.870 align:middle line:84% And so the vowels are continuing to chase each other around 00:57:55.870 --> 00:57:57.742 align:middle line:90% in the vowel space. 00:57:57.742 --> 00:57:59.200 align:middle line:84% I think I talked about this when we 00:57:59.200 --> 00:58:01.210 align:middle line:84% were talking about phonetic inventories, 00:58:01.210 --> 00:58:07.150 align:middle line:84% that there's work on the kinds of vowel inventories 00:58:07.150 --> 00:58:08.050 align:middle line:90% that we find. 00:58:08.050 --> 00:58:10.120 align:middle line:84% So there are plenty of languages out there that 00:58:10.120 --> 00:58:12.662 align:middle line:84% have three vowels, and the three vowels that you 00:58:12.662 --> 00:58:16.330 align:middle line:84% have if you have three vowels are "ee," "oo," and "ah." 00:58:16.330 --> 00:58:18.760 align:middle line:84% That is, there's a front high vowel, 00:58:18.760 --> 00:58:22.282 align:middle line:84% there's a back high vowel, and there's a low vowel. 00:58:22.282 --> 00:58:23.740 align:middle line:84% That is, you have three vowels that 00:58:23.740 --> 00:58:26.327 align:middle line:84% are kind of as far apart from each other as possible. 00:58:26.327 --> 00:58:28.660 align:middle line:84% There aren't languages out there that have three vowels, 00:58:28.660 --> 00:58:33.280 align:middle line:84% and the three vowels are "ee," "ey," and "eh." 00:58:33.280 --> 00:58:34.877 align:middle line:90% That's not a vowel system. 00:58:34.877 --> 00:58:36.460 align:middle line:84% One of my colleagues, Edward Flemming, 00:58:36.460 --> 00:58:39.220 align:middle line:84% has done a lot of work on different kinds of vowel 00:58:39.220 --> 00:58:41.500 align:middle line:84% systems and reducing them to this idea 00:58:41.500 --> 00:58:44.470 align:middle line:84% that you're trying to keep your vowels maximally 00:58:44.470 --> 00:58:48.710 align:middle line:84% dispersed from each other in the vocal tract. 00:58:48.710 --> 00:58:50.710 align:middle line:84% And that kind of thinking can be one of the ways 00:58:50.710 --> 00:58:54.070 align:middle line:84% to think about why English underwent this vowel shift. 00:58:54.070 --> 00:58:58.630 align:middle line:84% It's like, first the high vowels changed into dipthongs, 00:58:58.630 --> 00:59:01.108 align:middle line:84% and then you have this sort of unclaimed real estate 00:59:01.108 --> 00:59:02.650 align:middle line:84% at the top of the vocal tract, right? 00:59:02.650 --> 00:59:04.840 align:middle line:84% So once you no longer have an "ee" or an "oo," 00:59:04.840 --> 00:59:07.330 align:middle line:84% once they've changed to "ay" and "ow," 00:59:07.330 --> 00:59:09.850 align:middle line:84% there's this desire for dispersion 00:59:09.850 --> 00:59:14.740 align:middle line:84% that's pushing you to get some vowels into that space. 00:59:14.740 --> 00:59:17.155 align:middle line:84% That first change kind of triggers a second change. 00:59:17.155 --> 00:59:22.360 align:middle line:90% 00:59:22.360 --> 00:59:25.450 align:middle line:84% If you study languages, one of the things 00:59:25.450 --> 00:59:27.520 align:middle line:84% that you will sometimes be told is 00:59:27.520 --> 00:59:30.490 align:middle line:84% here's a regular rule for how you form, say, 00:59:30.490 --> 00:59:32.440 align:middle line:84% different forms of a noun, singulars 00:59:32.440 --> 00:59:34.970 align:middle line:84% and plurals or different case forms or whatever. 00:59:34.970 --> 00:59:37.480 align:middle line:84% And then here are some irregular forms. 00:59:37.480 --> 00:59:39.820 align:middle line:84% And sometimes the irregular forms 00:59:39.820 --> 00:59:42.440 align:middle line:84% are the consequence of sound changes. 00:59:42.440 --> 00:59:45.940 align:middle line:84% So for example, when I was briefly 00:59:45.940 --> 00:59:48.200 align:middle line:84% attempting to learn Latin in high school, 00:59:48.200 --> 00:59:52.420 align:middle line:84% we were taught there are certain Latin nouns, 00:59:52.420 --> 00:59:55.120 align:middle line:84% so-called third declension nouns, for which you 00:59:55.120 --> 00:59:57.910 align:middle line:84% need to memorize both the nominative singular 00:59:57.910 --> 01:00:00.700 align:middle line:90% form and another form. 01:00:00.700 --> 01:00:02.710 align:middle line:84% We were taught to memorize the genitive singular 01:00:02.710 --> 01:00:09.010 align:middle line:84% form because you can't predict from the nominative singular 01:00:09.010 --> 01:00:10.420 align:middle line:90% what the other forms will be. 01:00:10.420 --> 01:00:12.640 align:middle line:84% So you get words like "rex," which means "king," 01:00:12.640 --> 01:00:16.720 align:middle line:84% and "nox," which means "night," and "vox," which means "voice." 01:00:16.720 --> 01:00:22.390 align:middle line:84% But the stems to which you add any other suffix 01:00:22.390 --> 01:00:24.560 align:middle line:84% are unpredictable from the nominatives. 01:00:24.560 --> 01:00:29.050 align:middle line:84% So king, "rex," the genitive and the accusative and all 01:00:29.050 --> 01:00:33.280 align:middle line:84% the other forms are formed off of a stem "reg-" Or for night, 01:00:33.280 --> 01:00:36.730 align:middle line:84% it's "noct-," and for a voice, it's "voc-." 01:00:36.730 --> 01:00:38.150 align:middle line:90% Anybody else here study Latin? 01:00:38.150 --> 01:00:40.233 align:middle line:84% This is all stuff that you have to learn if you're 01:00:40.233 --> 01:00:41.350 align:middle line:90% trying to learn Latin. 01:00:41.350 --> 01:00:42.850 align:middle line:84% And what I was taught, at least, was 01:00:42.850 --> 01:00:45.580 align:middle line:84% to think of "king," the word for king as "rex, regis." 01:00:45.580 --> 01:00:49.220 align:middle line:84% We were supposed to just sort of cite both of those. 01:00:49.220 --> 01:00:51.940 align:middle line:84% So my old high school Latin teacher, Mrs. McNair, 01:00:51.940 --> 01:00:53.110 align:middle line:90% that's what she taught us. 01:00:53.110 --> 01:00:55.527 align:middle line:84% Mrs. McNair could have taught us-- she probably knew this, 01:00:55.527 --> 01:00:59.830 align:middle line:84% but she didn't reveal it to us-- that although you can't predict 01:00:59.830 --> 01:01:02.350 align:middle line:84% the genitive from the nominative, 01:01:02.350 --> 01:01:04.930 align:middle line:84% you can predict the nominative from the genitive. 01:01:04.930 --> 01:01:08.260 align:middle line:84% So what's really happening in Latin is that you've got these 01:01:08.260 --> 01:01:11.440 align:middle line:84% stems, which are "reg-," "noct-," "voc-," 01:01:11.440 --> 01:01:14.770 align:middle line:84% and that you're forming the nominative by adding an "s" 01:01:14.770 --> 01:01:16.730 align:middle line:84% and then doing some sound changes. 01:01:16.730 --> 01:01:19.840 align:middle line:84% So you take "reg-" and you add an "s." 01:01:19.840 --> 01:01:21.820 align:middle line:84% That by itself would give you "regs." 01:01:21.820 --> 01:01:24.610 align:middle line:84% And then Latin doesn't allow a sequence 01:01:24.610 --> 01:01:26.440 align:middle line:90% of a "g" followed by an "s." 01:01:26.440 --> 01:01:27.840 align:middle line:90% You make them both voiceless. 01:01:27.840 --> 01:01:29.920 align:middle line:84% You change the "g" so that it's "k." 01:01:29.920 --> 01:01:33.640 align:middle line:84% and Latin spells the sequence "ks" with an "x." 01:01:33.640 --> 01:01:38.680 align:middle line:84% Or "nox," yeah, that would end in two stops followed 01:01:38.680 --> 01:01:39.310 align:middle line:90% by an "s." 01:01:39.310 --> 01:01:41.140 align:middle line:84% A "k," a "t"-- a "k" sound, which 01:01:41.140 --> 01:01:45.022 align:middle line:84% is spelled in Latin with a "c," and a "t" and then an "s." 01:01:45.022 --> 01:01:47.260 align:middle line:84% And then you simplify that by getting rid 01:01:47.260 --> 01:01:49.720 align:middle line:90% of the "t" in between. 01:01:49.720 --> 01:01:52.510 align:middle line:84% Or "vox," which is, well, basically, just "vocs," 01:01:52.510 --> 01:01:54.160 align:middle line:90% just adding "s" to the stem. 01:01:54.160 --> 01:01:57.940 align:middle line:84% So Mrs. McNair could have made my life easier 01:01:57.940 --> 01:02:01.095 align:middle line:84% if she had been willing to tell me about that. 01:02:01.095 --> 01:02:02.470 align:middle line:84% So there have to be sound changes 01:02:02.470 --> 01:02:04.990 align:middle line:84% that if you're willing to think about the third declension 01:02:04.990 --> 01:02:06.270 align:middle line:90% nouns this way-- 01:02:06.270 --> 01:02:07.450 align:middle line:90% [COUGHING] excuse me. 01:02:07.450 --> 01:02:10.870 align:middle line:84% You can think of them as just having a nominative singular 01:02:10.870 --> 01:02:14.350 align:middle line:84% form "s" and some sound changes that obscure 01:02:14.350 --> 01:02:17.500 align:middle line:84% the fact that the nominative singular is really not 01:02:17.500 --> 01:02:20.657 align:middle line:90% so complicated. 01:02:20.657 --> 01:02:22.990 align:middle line:84% She would have had to teach us a bunch of sound changes, 01:02:22.990 --> 01:02:26.110 align:middle line:90% but these are some of them. 01:02:26.110 --> 01:02:30.130 align:middle line:84% Or here's another example of this kind. 01:02:30.130 --> 01:02:31.990 align:middle line:84% We had a problem set earlier on Inupiaq, 01:02:31.990 --> 01:02:34.420 align:middle line:84% which is one of a number of languages that 01:02:34.420 --> 01:02:35.770 align:middle line:90% are all related to each other. 01:02:35.770 --> 01:02:37.570 align:middle line:84% They're all members of what is still, 01:02:37.570 --> 01:02:40.360 align:middle line:84% for some reason, called the Eskimo family of languages, 01:02:40.360 --> 01:02:43.150 align:middle line:84% where I say "still" because the word "Eskimo" 01:02:43.150 --> 01:02:47.980 align:middle line:84% is not well-regarded among the people to whom it refers. 01:02:47.980 --> 01:02:51.790 align:middle line:84% It's a name for them that was come up with by the Cree. 01:02:51.790 --> 01:02:55.330 align:middle line:84% There are debates about what it means. 01:02:55.330 --> 01:02:56.930 align:middle line:84% But it doesn't matter what it means. 01:02:56.930 --> 01:02:58.472 align:middle line:84% They don't like being called Eskimos, 01:02:58.472 --> 01:03:01.600 align:middle line:84% so it's unfortunate that linguistics continues 01:03:01.600 --> 01:03:03.160 align:middle line:84% to refer to their language family 01:03:03.160 --> 01:03:05.260 align:middle line:90% as the Proto-Eskimo family. 01:03:05.260 --> 01:03:08.710 align:middle line:84% Here are some words in this language family. 01:03:08.710 --> 01:03:10.840 align:middle line:84% These are proto forms, and so I should really 01:03:10.840 --> 01:03:12.520 align:middle line:90% have stars on them. 01:03:12.520 --> 01:03:14.800 align:middle line:84% So "iglu," which is a word you may have heard, 01:03:14.800 --> 01:03:17.450 align:middle line:90% it's the word for house. 01:03:17.450 --> 01:03:18.910 align:middle line:90% So all of us live in igloos. 01:03:18.910 --> 01:03:23.200 align:middle line:84% We all grew up in igloos, as far as the language is concerned, 01:03:23.200 --> 01:03:24.640 align:middle line:90% or "tuma" or "tavsi." 01:03:24.640 --> 01:03:26.350 align:middle line:90% Here are some words. 01:03:26.350 --> 01:03:29.290 align:middle line:84% And there's a very general plural suffix, 01:03:29.290 --> 01:03:33.490 align:middle line:90% "t," which gets added. 01:03:33.490 --> 01:03:36.670 align:middle line:84% The proto language for this language family had-- 01:03:36.670 --> 01:03:39.220 align:middle line:90% let's see now-- four vowels. 01:03:39.220 --> 01:03:42.107 align:middle line:84% So there was "a," "i," and "u," and there's also schwa. 01:03:42.107 --> 01:03:43.690 align:middle line:84% You can see there's a schwa at the end 01:03:43.690 --> 01:03:46.450 align:middle line:90% of the word for "footprint." 01:03:46.450 --> 01:03:50.770 align:middle line:84% And then in the history between the proto language 01:03:50.770 --> 01:03:55.090 align:middle line:84% and modern Inupiaq, two sound changes took place. 01:03:55.090 --> 01:03:58.492 align:middle line:84% First, "t," like the "t" of the plural, 01:03:58.492 --> 01:03:59.950 align:middle line:84% but not just the "t" of the plural. 01:03:59.950 --> 01:04:01.200 align:middle line:90% It was a general sound change. 01:04:01.200 --> 01:04:05.770 align:middle line:84% Whenever you had a "t" that followed the vowel "i," the "t" 01:04:05.770 --> 01:04:07.403 align:middle line:84% underwent what's called palatalization, 01:04:07.403 --> 01:04:09.070 align:middle line:84% cross-linguistically a very common sound 01:04:09.070 --> 01:04:14.230 align:middle line:84% change where a "t" becomes a "ch" around a front vowel. 01:04:14.230 --> 01:04:16.270 align:middle line:84% We had something like this happen 01:04:16.270 --> 01:04:17.470 align:middle line:90% in English several times. 01:04:17.470 --> 01:04:20.725 align:middle line:84% It's why our word for cheese starts with a "ch" sound. 01:04:20.725 --> 01:04:26.690 align:middle line:84% It comes from Latin caseus, which started with a "k" sound. 01:04:26.690 --> 01:04:29.320 align:middle line:84% So T became "ch" after the vowel "i." 01:04:29.320 --> 01:04:32.290 align:middle line:84% That's what you're seeing at the end of the word for "belts." 01:04:32.290 --> 01:04:33.710 align:middle line:90% Regular sound change. 01:04:33.710 --> 01:04:37.278 align:middle line:84% And then another sound change, Inupiaq 01:04:37.278 --> 01:04:38.320 align:middle line:90% has a three vowel system. 01:04:38.320 --> 01:04:39.760 align:middle line:90% It only has "a," "i," and "u." 01:04:39.760 --> 01:04:41.140 align:middle line:90% They got rid of schwa. 01:04:41.140 --> 01:04:44.133 align:middle line:84% And in particular, they changed schwa to an "i." 01:04:44.133 --> 01:04:45.550 align:middle line:84% But these sound changes took place 01:04:45.550 --> 01:04:47.590 align:middle line:84% in this order with the consequence 01:04:47.590 --> 01:04:50.830 align:middle line:84% that if you want to learn Inupiaq now, what you're taught 01:04:50.830 --> 01:04:55.330 align:middle line:84% is, well, the plural suffix is a "t," except after some nouns, 01:04:55.330 --> 01:04:59.260 align:middle line:84% but not all nouns, that have an "i" at the end of them, 01:04:59.260 --> 01:05:00.730 align:middle line:90% where the "t" becomes a "ch." 01:05:00.730 --> 01:05:04.900 align:middle line:84% And you just have to learn which nouns that end in "i" 01:05:04.900 --> 01:05:06.730 align:middle line:84% change the "t" to a "ch" and which 01:05:06.730 --> 01:05:08.620 align:middle line:84% nouns that end in "i"-- the vowel that's 01:05:08.620 --> 01:05:10.750 align:middle line:90% written with the letter "i"-- 01:05:10.750 --> 01:05:11.410 align:middle line:90% don't do that. 01:05:11.410 --> 01:05:14.800 align:middle line:84% So "belt," you just have to memorize, 01:05:14.800 --> 01:05:16.600 align:middle line:84% that's the kind of "i" final noun where 01:05:16.600 --> 01:05:18.610 align:middle line:90% the plural suffix is a "ch." 01:05:18.610 --> 01:05:22.450 align:middle line:84% "Footprint," that's the kind of "i" final where the plural 01:05:22.450 --> 01:05:25.840 align:middle line:90% suffix is still a "t." 01:05:25.840 --> 01:05:28.930 align:middle line:84% This situation, which is, of course, distressing to people 01:05:28.930 --> 01:05:32.263 align:middle line:84% who want to learn Inupiaq, is the consequence 01:05:32.263 --> 01:05:34.180 align:middle line:84% of these sound changes occurring in this order 01:05:34.180 --> 01:05:35.800 align:middle line:84% and creating this kind of opacity. 01:05:35.800 --> 01:05:38.200 align:middle line:84% So what's really going on is, well, there 01:05:38.200 --> 01:05:39.880 align:middle line:90% used to be another vowel. 01:05:39.880 --> 01:05:43.600 align:middle line:84% And we got rid of that vowel, but we were already 01:05:43.600 --> 01:05:46.250 align:middle line:84% committed to having "ch" in certain places and not others. 01:05:46.250 --> 01:05:49.720 align:middle line:84% I should say there's now dialect variation within Inupiaq. 01:05:49.720 --> 01:05:52.000 align:middle line:84% There are dialects of Inupiaq that said, no, look, 01:05:52.000 --> 01:05:53.110 align:middle line:90% this is silly. 01:05:53.110 --> 01:05:56.470 align:middle line:84% We're going to change the "ch" in "belts" to a "t." 01:05:56.470 --> 01:05:59.320 align:middle line:84% So they still have "t" becoming "ch" in some other places, 01:05:59.320 --> 01:06:01.570 align:middle line:84% but specifically the plural has gotten regularized 01:06:01.570 --> 01:06:03.260 align:middle line:90% so that it's a "t" everywhere. 01:06:03.260 --> 01:06:07.130 align:middle line:90% 01:06:07.130 --> 01:06:08.930 align:middle line:84% Consequence of a history of sound changes 01:06:08.930 --> 01:06:10.340 align:middle line:90% is sometimes opacity. 01:06:10.340 --> 01:06:11.990 align:middle line:84% It's kind of like the Lardil example 01:06:11.990 --> 01:06:13.250 align:middle line:90% that I showed you before. 01:06:13.250 --> 01:06:15.830 align:middle line:84% You have a change from A to B, and you 01:06:15.830 --> 01:06:19.340 align:middle line:84% have another change from C to D. And the consequence 01:06:19.340 --> 01:06:23.690 align:middle line:84% of the change from C to D would create new environments for A 01:06:23.690 --> 01:06:25.640 align:middle line:84% to B, but A to B has already happened. 01:06:25.640 --> 01:06:28.050 align:middle line:90% It doesn't happen again. 01:06:28.050 --> 01:06:29.840 align:middle line:84% So as you are learning languages, 01:06:29.840 --> 01:06:32.030 align:middle line:84% if you encounter facts like this, 01:06:32.030 --> 01:06:34.100 align:middle line:84% entertain the possibility that what's going on 01:06:34.100 --> 01:06:36.500 align:middle line:84% is opacity because of sound change occurring 01:06:36.500 --> 01:06:37.460 align:middle line:90% in a particular order. 01:06:37.460 --> 01:06:39.140 align:middle line:84% I don't know whether you'll consider-- 01:06:39.140 --> 01:06:40.910 align:middle line:90% whether you'll have-- 01:06:40.910 --> 01:06:43.370 align:middle line:84% whether you'll get any comfort from that thought 01:06:43.370 --> 01:06:46.340 align:middle line:84% or whether it'll just annoy you, but that's 01:06:46.340 --> 01:06:49.100 align:middle line:90% sometimes the consequence. 01:06:49.100 --> 01:06:49.850 align:middle line:90% Yeah, Lardil. 01:06:49.850 --> 01:06:52.970 align:middle line:90% We just did Lardil. 01:06:52.970 --> 01:06:53.810 align:middle line:90% Oh, yeah. 01:06:53.810 --> 01:06:56.300 align:middle line:84% Here's another example of the same kind of thing. 01:06:56.300 --> 01:07:00.440 align:middle line:84% Passamaquoddy is an Algonquian language spoken up in Maine. 01:07:00.440 --> 01:07:02.870 align:middle line:84% It underwent a sound change that deleted 01:07:02.870 --> 01:07:06.663 align:middle line:84% odd-numbered short vowels in words, 01:07:06.663 --> 01:07:08.330 align:middle line:84% depending on the consonants around them. 01:07:08.330 --> 01:07:11.180 align:middle line:90% It's called syncope. 01:07:11.180 --> 01:07:14.660 align:middle line:84% If you started with a word that meant "I hook a fish," 01:07:14.660 --> 01:07:17.870 align:middle line:84% that would have been something like "nuh-puhteeheek." 01:07:17.870 --> 01:07:22.820 align:middle line:84% And what you did was to get rid of the first and third vowels 01:07:22.820 --> 01:07:24.360 align:middle line:90% in that word. 01:07:24.360 --> 01:07:27.502 align:middle line:84% So in modern Passamaquoddy, that's now "npuh-teeg." 01:07:27.502 --> 01:07:29.960 align:middle line:84% So you've gotten rid of the first vowel and the third vowel 01:07:29.960 --> 01:07:31.250 align:middle line:90% in the word. 01:07:31.250 --> 01:07:34.490 align:middle line:84% The "nuh-" at the beginning of that is a prefix that agrees 01:07:34.490 --> 01:07:36.551 align:middle line:90% with the subject, "I." 01:07:36.551 --> 01:07:39.225 align:middle line:84% If you get rid of that prefix, like if you want to say he 01:07:39.225 --> 01:07:41.600 align:middle line:84% or she hooks a fish, well then you don't have that prefix 01:07:41.600 --> 01:07:42.980 align:middle line:90% and that changes the count. 01:07:42.980 --> 01:07:47.000 align:middle line:84% So original "puh-tee-hee-geh," again, 01:07:47.000 --> 01:07:49.470 align:middle line:84% you've gotten rid of the first and third vowels in that. 01:07:49.470 --> 01:07:52.490 align:middle line:84% So in modern Passamaquoddy, it's "tee-gee-geh." 01:07:52.490 --> 01:07:54.745 align:middle line:84% So Passamaquoddy speakers, if they 01:07:54.745 --> 01:07:56.120 align:middle line:84% want to learn Passamaquoddy, they 01:07:56.120 --> 01:07:59.270 align:middle line:84% have to learn if you add the prefix, you sometimes 01:07:59.270 --> 01:08:01.110 align:middle line:90% change which vowels are where. 01:08:01.110 --> 01:08:04.430 align:middle line:90% So "I hook a fish" is "npu-tig." 01:08:04.430 --> 01:08:07.310 align:middle line:84% But "he or she hooks a fish" is "tee-hee-guh." 01:08:07.310 --> 01:08:10.490 align:middle line:84% There's a vowel between the "p" and the "t" in the "I" form, 01:08:10.490 --> 01:08:11.870 align:middle line:90% but not in the "he or she" form. 01:08:11.870 --> 01:08:13.880 align:middle line:84% There is a vowel between the "t" and the "h" 01:08:13.880 --> 01:08:16.609 align:middle line:84% in the "he or she" form, but not in the "I" form. 01:08:16.609 --> 01:08:19.220 align:middle line:84% If you look at the online Passamaquoddy dictionary, 01:08:19.220 --> 01:08:21.290 align:middle line:84% you see lots of examples of this kind of thing. 01:08:21.290 --> 01:08:24.529 align:middle line:84% Verbs are listed in their third-person form, 01:08:24.529 --> 01:08:27.828 align:middle line:84% but then there will be a listing for each verb. 01:08:27.828 --> 01:08:29.870 align:middle line:84% This is what it looks like when you add a prefix. 01:08:29.870 --> 01:08:32.600 align:middle line:84% And it's sometimes radically different as a consequence 01:08:32.600 --> 01:08:34.140 align:middle line:90% of this sound change. 01:08:34.140 --> 01:08:37.189 align:middle line:84% So for example, here's another instance of the same thing. 01:08:37.189 --> 01:08:39.260 align:middle line:84% "I'm sorry about it" was originally 01:08:39.260 --> 01:08:42.050 align:middle line:90% "nuh-muh-zah-kay-in." 01:08:42.050 --> 01:08:46.340 align:middle line:84% Getting rid of the first and third vowels in that would have 01:08:46.340 --> 01:08:50.510 align:middle line:84% given you "nmuh-skey-in," which is the modern way to say 01:08:50.510 --> 01:08:51.740 align:middle line:90% "I'm sorry about it." 01:08:51.740 --> 01:08:55.055 align:middle line:84% If you don't add the "nuh" prefix, if you want to say 01:08:55.055 --> 01:08:56.930 align:middle line:84% "he or she is sorry," you're going to get rid 01:08:56.930 --> 01:08:58.550 align:middle line:90% of the first and third vowels. 01:08:58.550 --> 01:09:02.359 align:middle line:84% If nothing else happened, that would give you "sah-key-oo." 01:09:02.359 --> 01:09:07.609 align:middle line:84% But in fact, "n" next to "s" tends to become a "p." 01:09:07.609 --> 01:09:11.330 align:middle line:84% So the word for "I'm sorry about it" is "nmuh-skey-in," 01:09:11.330 --> 01:09:14.512 align:middle line:84% but "he or she is sorry" is "psah-key-oo" 01:09:14.512 --> 01:09:16.970 align:middle line:84% with the consequence that these look so different from each 01:09:16.970 --> 01:09:19.470 align:middle line:84% other that when I have pointed out to Passamaquoddy speakers 01:09:19.470 --> 01:09:21.740 align:middle line:84% that they are the same word, they're astonished. 01:09:21.740 --> 01:09:23.029 align:middle line:90% They're like, oh, yeah, right. 01:09:23.029 --> 01:09:25.970 align:middle line:84% I guess they are, even though they don't look all that much 01:09:25.970 --> 01:09:28.970 align:middle line:90% like each other anymore. 01:09:28.970 --> 01:09:32.660 align:middle line:84% Now, the Passamaquoddy system looks like the result 01:09:32.660 --> 01:09:33.750 align:middle line:90% of a stress system. 01:09:33.750 --> 01:09:38.930 align:middle line:84% So again, the rule is get rid of the odd-numbered short vowels 01:09:38.930 --> 01:09:40.073 align:middle line:90% in the sequence. 01:09:40.073 --> 01:09:42.740 align:middle line:84% And that's the kind of rule that makes sense if you think, yeah, 01:09:42.740 --> 01:09:45.529 align:middle line:84% it had a stress system that stressed 01:09:45.529 --> 01:09:46.970 align:middle line:90% the even-numbered vowels, right? 01:09:46.970 --> 01:09:51.140 align:middle line:84% So "I hook a fish" was "nuh-put-tuh-heek." 01:09:51.140 --> 01:09:53.569 align:middle line:84% And then they got rid of the first and the third vowels 01:09:53.569 --> 01:09:54.950 align:middle line:90% because they weren't stressed. 01:09:54.950 --> 01:09:58.070 align:middle line:84% And "he or she hooks a fish" was "puh-tee-hee-geh," 01:09:58.070 --> 01:10:01.550 align:middle line:84% and you got rid of the schwa in the first one because well, 01:10:01.550 --> 01:10:03.170 align:middle line:90% it wasn't stressed. 01:10:03.170 --> 01:10:04.670 align:middle line:84% Whether you get rid of vowels or not 01:10:04.670 --> 01:10:06.830 align:middle line:84% depends partly on what consonants are around them. 01:10:06.830 --> 01:10:08.000 align:middle line:84% That's why you're not actually getting 01:10:08.000 --> 01:10:09.458 align:middle line:84% rid of the third vowel in this one. 01:10:09.458 --> 01:10:12.410 align:middle line:90% You get "tee-hee-geh." 01:10:12.410 --> 01:10:14.600 align:middle line:84% So if you thought of Passamaquoddy 01:10:14.600 --> 01:10:18.470 align:middle line:84% as having stress on its even-numbered short vowels, 01:10:18.470 --> 01:10:20.210 align:middle line:84% then you can understand why it's getting 01:10:20.210 --> 01:10:21.620 align:middle line:84% rid of its odd-numbered short vowels. 01:10:21.620 --> 01:10:23.870 align:middle line:84% It's getting rid of them because they're not stressed. 01:10:23.870 --> 01:10:25.550 align:middle line:84% And Passamaquoddy does have relatives 01:10:25.550 --> 01:10:28.850 align:middle line:84% that have stress systems like that, where you stress 01:10:28.850 --> 01:10:31.850 align:middle line:90% the even-numbered vowels. 01:10:31.850 --> 01:10:34.920 align:middle line:84% But that is not Passamaquoddy stress system anymore. 01:10:34.920 --> 01:10:37.877 align:middle line:84% So Passamaquoddy underwent all of these sound changes because 01:10:37.877 --> 01:10:39.710 align:middle line:84% of the stress system that they used to have, 01:10:39.710 --> 01:10:41.840 align:middle line:84% that they used to share with some of their relatives. 01:10:41.840 --> 01:10:43.590 align:middle line:84% And then they changed their stress system. 01:10:43.590 --> 01:10:46.310 align:middle line:84% Their modern stress system goes "Stress the first vowel 01:10:46.310 --> 01:10:47.960 align:middle line:84% and stress every other vowel counting 01:10:47.960 --> 01:10:50.120 align:middle line:90% backwards from the end." 01:10:50.120 --> 01:10:53.060 align:middle line:84% So you get words like [INAUDIBLE],, 01:10:53.060 --> 01:10:54.900 align:middle line:84% which is "he or she talks like that," 01:10:54.900 --> 01:10:56.960 align:middle line:84% where you're stressing the first vowel and the second vowel. 01:10:56.960 --> 01:10:59.210 align:middle line:84% First vowel because you always stress the first vowel, 01:10:59.210 --> 01:11:00.650 align:middle line:84% and the second vowel because it's 01:11:00.650 --> 01:11:03.167 align:middle line:90% two vowels back from the end. 01:11:03.167 --> 01:11:05.000 align:middle line:84% So every other vowel, starting from the end. 01:11:05.000 --> 01:11:06.620 align:middle line:84% If you make it one syllable longer, 01:11:06.620 --> 01:11:09.560 align:middle line:84% then you're going to stress the first and the third vowel-- 01:11:09.560 --> 01:11:14.750 align:middle line:84% "wee-guh-west-too," "He or she talks while walking backwards." 01:11:14.750 --> 01:11:17.672 align:middle line:84% "Quiguh-west-too-bin," "You and I like talking." 01:11:17.672 --> 01:11:19.130 align:middle line:84% So stress is on the first syllable, 01:11:19.130 --> 01:11:21.770 align:middle line:84% and it's on the even-numbered syllables counting backwards 01:11:21.770 --> 01:11:23.250 align:middle line:90% from the end of the word. 01:11:23.250 --> 01:11:26.030 align:middle line:84% So they've innovated this new stress system 01:11:26.030 --> 01:11:28.490 align:middle line:84% after having this older stress system which 01:11:28.490 --> 01:11:30.620 align:middle line:84% created all of these alternations 01:11:30.620 --> 01:11:31.802 align:middle line:90% in the positions of vowels. 01:11:31.802 --> 01:11:33.260 align:middle line:84% There's a reason that Passamaquoddy 01:11:33.260 --> 01:11:34.617 align:middle line:90% is difficult to learn. 01:11:34.617 --> 01:11:35.700 align:middle line:90% There are several reasons. 01:11:35.700 --> 01:11:36.290 align:middle line:90% This is one. 01:11:36.290 --> 01:11:39.883 align:middle line:90% 01:11:39.883 --> 01:11:42.300 align:middle line:84% So it used to have a different stress system, during which 01:11:42.300 --> 01:11:43.350 align:middle line:90% the syncope rule applied. 01:11:43.350 --> 01:11:47.760 align:middle line:84% And now it has the system it has now. 01:11:47.760 --> 01:11:49.470 align:middle line:84% I want to end today because I think 01:11:49.470 --> 01:11:51.210 align:middle line:84% we have enough time for me to heap 01:11:51.210 --> 01:11:54.390 align:middle line:90% scorn on a couple of things. 01:11:54.390 --> 01:11:56.017 align:middle line:84% I want to end today with two bad ideas. 01:11:56.017 --> 01:11:57.600 align:middle line:84% So one of the reasons that I'm talking 01:11:57.600 --> 01:11:58.980 align:middle line:84% to you about historical linguistics-- there 01:11:58.980 --> 01:11:59.480 align:middle line:90% are several. 01:11:59.480 --> 01:12:00.775 align:middle line:90% One is that it's fun. 01:12:00.775 --> 01:12:03.150 align:middle line:84% It's kind of interesting to try to trace back these sound 01:12:03.150 --> 01:12:03.720 align:middle line:90% changes. 01:12:03.720 --> 01:12:05.700 align:middle line:84% But another is that historical linguistics 01:12:05.700 --> 01:12:08.430 align:middle line:84% is the kind of thing that is sometimes done badly. 01:12:08.430 --> 01:12:11.262 align:middle line:84% So you'll see in popular press people saying things 01:12:11.262 --> 01:12:13.470 align:middle line:84% like, hey, this language is related to that language, 01:12:13.470 --> 01:12:16.860 align:middle line:84% or this language has borrowed this word from that language. 01:12:16.860 --> 01:12:19.710 align:middle line:84% And if you have any skepticism about that, 01:12:19.710 --> 01:12:21.210 align:middle line:84% what you want them to do is show you 01:12:21.210 --> 01:12:23.640 align:middle line:84% regular sound correspondences or regular rules 01:12:23.640 --> 01:12:25.470 align:middle line:84% for how words are related to other words. 01:12:25.470 --> 01:12:29.010 align:middle line:84% So there was a popular book for a while that 01:12:29.010 --> 01:12:35.940 align:middle line:84% proposed that the Native American languages were all 01:12:35.940 --> 01:12:40.590 align:middle line:84% descended from Chinese, that the Mandarin-- the Chinese 01:12:40.590 --> 01:12:42.990 align:middle line:84% had encountered North America first, 01:12:42.990 --> 01:12:46.410 align:middle line:84% and that their language had taken over North America, 01:12:46.410 --> 01:12:48.000 align:middle line:84% and that all of the modern languages 01:12:48.000 --> 01:12:49.470 align:middle line:90% were descended from Chinese. 01:12:49.470 --> 01:12:52.890 align:middle line:84% And the examples were things like in modern Mandarin, 01:12:52.890 --> 01:12:56.310 align:middle line:90% there is a greeting, "nihao." 01:12:56.310 --> 01:12:59.130 align:middle line:84% And hey, if you watch Hollywood movies, 01:12:59.130 --> 01:13:02.670 align:middle line:84% you will hear Native Americans greeting each other with "how." 01:13:02.670 --> 01:13:05.430 align:middle line:84% And there, you know, so clearly these 01:13:05.430 --> 01:13:06.520 align:middle line:90% are related to each other. 01:13:06.520 --> 01:13:08.700 align:middle line:84% "How" actually is a greeting in Lakota. 01:13:08.700 --> 01:13:12.430 align:middle line:84% It's a greeting that Lakota men use to greet other men. 01:13:12.430 --> 01:13:15.930 align:middle line:84% It's not related to the Mandarin expression "nihao." 01:13:15.930 --> 01:13:18.780 align:middle line:84% And the reason I can say that with so much confidence 01:13:18.780 --> 01:13:22.380 align:middle line:84% is that, well, no one has ever found sound correspondences 01:13:22.380 --> 01:13:25.080 align:middle line:84% or anything like that relating anything else in Lakota 01:13:25.080 --> 01:13:29.010 align:middle line:90% to anything else in Chinese. 01:13:29.010 --> 01:13:32.490 align:middle line:84% There are slightly less silly claims 01:13:32.490 --> 01:13:34.650 align:middle line:84% that people have made using what's sometimes 01:13:34.650 --> 01:13:35.950 align:middle line:90% called megalocomparison. 01:13:35.950 --> 01:13:38.580 align:middle line:84% So let's skip-- so glottochronology-- 01:13:38.580 --> 01:13:40.710 align:middle line:84% I'll leave this on the slide, but glottochronology 01:13:40.710 --> 01:13:43.590 align:middle line:84% is an attempt to sort of carbon date language splits. 01:13:43.590 --> 01:13:48.120 align:middle line:84% So the idea is you take a list of basic vocabulary. 01:13:48.120 --> 01:13:50.490 align:middle line:84% You figure out how many cognates the two languages 01:13:50.490 --> 01:13:54.420 align:middle line:84% share on the list, and you start with the assumption 01:13:54.420 --> 01:13:56.940 align:middle line:84% that cognate loss happens at a constant rate. 01:13:56.940 --> 01:14:00.000 align:middle line:84% This is the figure that has been offered by glottochronologists. 01:14:00.000 --> 01:14:02.850 align:middle line:90% And then you do some math. 01:14:02.850 --> 01:14:06.360 align:middle line:84% And this is something that you see people seriously saying. 01:14:06.360 --> 01:14:08.400 align:middle line:90% So if you look in-- 01:14:08.400 --> 01:14:10.770 align:middle line:84% historical linguistics is usually not done by linguists. 01:14:10.770 --> 01:14:12.960 align:middle line:84% It's often being done by biologists. 01:14:12.960 --> 01:14:15.750 align:middle line:84% They'll say, we know that these languages are related 01:14:15.750 --> 01:14:17.760 align:middle line:84% at a time depth of X. And what they 01:14:17.760 --> 01:14:19.485 align:middle line:84% mean is we've done glottochronology, 01:14:19.485 --> 01:14:21.360 align:middle line:84% which starts from the assumption that you can 01:14:21.360 --> 01:14:26.040 align:middle line:84% do this kind of carbon dating of languages by looking 01:14:26.040 --> 01:14:29.130 align:middle line:84% at cognate lists and figuring out the rate at which two 01:14:29.130 --> 01:14:31.780 align:middle line:84% languages cease to have cognates with each other. 01:14:31.780 --> 01:14:34.200 align:middle line:84% The problem is that the base assumption-- 01:14:34.200 --> 01:14:37.560 align:middle line:84% that cognate loss happens at a constant rate-- is false. 01:14:37.560 --> 01:14:41.925 align:middle line:84% And no one has figured out a way around that. 01:14:41.925 --> 01:14:44.050 align:middle line:84% There doesn't seem to be any way to correct for it. 01:14:44.050 --> 01:14:46.830 align:middle line:84% So if you see people using glottochronology, 01:14:46.830 --> 01:14:48.660 align:middle line:84% you should treat them with skepticism. 01:14:48.660 --> 01:14:53.340 align:middle line:90% 01:14:53.340 --> 01:14:56.130 align:middle line:84% The other bad idea that's in here is megalocomparison. 01:14:56.130 --> 01:14:59.640 align:middle line:84% This is attempts to reconstruct Proto- World, which were very 01:14:59.640 --> 01:15:01.240 align:middle line:90% en vogue at a certain time. 01:15:01.240 --> 01:15:03.030 align:middle line:84% People thought if we do lots and lots 01:15:03.030 --> 01:15:04.650 align:middle line:84% of this cross-linguistic comparison, 01:15:04.650 --> 01:15:07.110 align:middle line:84% we'll be able to figure out the ancestor language for all 01:15:07.110 --> 01:15:09.870 align:middle line:90% of the languages in the world. 01:15:09.870 --> 01:15:14.040 align:middle line:84% The work has never involved discovering 01:15:14.040 --> 01:15:17.190 align:middle line:84% general sound changes that relate languages to each other. 01:15:17.190 --> 01:15:18.937 align:middle line:84% It always involves going back to saying, 01:15:18.937 --> 01:15:20.520 align:middle line:84% "Hey, look, that word in that language 01:15:20.520 --> 01:15:22.437 align:middle line:84% looks kind of like that word in that language! 01:15:22.437 --> 01:15:23.430 align:middle line:90% I bet they're related." 01:15:23.430 --> 01:15:26.070 align:middle line:84% That's how that work is always gone. 01:15:26.070 --> 01:15:27.702 align:middle line:84% So those are two ideas-- bad ideas 01:15:27.702 --> 01:15:29.160 align:middle line:84% that you should be skeptical about. 01:15:29.160 --> 01:15:32.040 align:middle line:84% This has been my public service announcement for today. 01:15:32.040 --> 01:15:34.620 align:middle line:84% Are there any questions about either of those bad ideas? 01:15:34.620 --> 01:15:38.310 align:middle line:84% I put more detail about the bad ideas in the slides. 01:15:38.310 --> 01:15:39.240 align:middle line:90% OK, cool. 01:15:39.240 --> 01:15:43.430 align:middle line:84% Go forth, and I will see you next week. 01:15:43.430 --> 01:15:51.000 align:middle line:90%