WEBVTT 00:00:00.000 --> 00:00:03.992 align:middle line:90% [SQUEAK] [RUSTLING] [CLICKING] 00:00:03.992 --> 00:00:10.380 align:middle line:90% 00:00:10.380 --> 00:00:11.630 align:middle line:90% NORVIN RICHARDS: Welcome back. 00:00:11.630 --> 00:00:14.620 align:middle line:84% Hope everybody had a restful break, spent lots of time 00:00:14.620 --> 00:00:17.140 align:middle line:84% doing syntax in their spare time. 00:00:17.140 --> 00:00:19.810 align:middle line:90% Let's do a little syntax now. 00:00:19.810 --> 00:00:21.743 align:middle line:90% I don't have any announcements. 00:00:21.743 --> 00:00:23.410 align:middle line:84% Are there any questions about the things 00:00:23.410 --> 00:00:25.360 align:middle line:84% that I'm not announcing, things that people 00:00:25.360 --> 00:00:30.000 align:middle line:84% are, like, wondering about, where we start? 00:00:30.000 --> 00:00:30.500 align:middle line:90% OK. 00:00:30.500 --> 00:00:32.292 align:middle line:84% So I just wanted to start with some review, 00:00:32.292 --> 00:00:34.174 align:middle line:84% just to remind everybody where we were. 00:00:34.174 --> 00:00:38.300 align:middle line:84% We were drawing trees like this one. 00:00:38.300 --> 00:00:40.430 align:middle line:84% So here's a tree for "I will tickle the child 00:00:40.430 --> 00:00:42.470 align:middle line:90% with the feather." 00:00:42.470 --> 00:00:45.710 align:middle line:84% It's a tree that has various kinds of information in it 00:00:45.710 --> 00:00:50.720 align:middle line:84% about, well, various things, including substrings 00:00:50.720 --> 00:00:53.000 align:middle line:84% that we think are treated as constituents 00:00:53.000 --> 00:00:55.260 align:middle line:84% for various kinds of syntactic phenomena. 00:00:55.260 --> 00:00:59.120 align:middle line:84% So this tree is meant to reflect the fact 00:00:59.120 --> 00:01:02.840 align:middle line:84% that this sentence, first of all, 00:01:02.840 --> 00:01:06.140 align:middle line:84% has a meaning in which I'm going to use a feather 00:01:06.140 --> 00:01:07.580 align:middle line:90% to tickle a child. 00:01:07.580 --> 00:01:10.370 align:middle line:84% That's the meaning that's been diagrammed here. 00:01:10.370 --> 00:01:14.120 align:middle line:84% And that if you perform certain operations that 00:01:14.120 --> 00:01:17.030 align:middle line:84% make it clear that this is the tree that you've got, like, 00:01:17.030 --> 00:01:19.910 align:middle line:84% topicalizing "the child," moving "the child" to the beginning 00:01:19.910 --> 00:01:21.770 align:middle line:84% of the sentence, "The child I will 00:01:21.770 --> 00:01:24.583 align:middle line:84% tickle with the feather," that's the only-- in fact, 00:01:24.583 --> 00:01:26.000 align:middle line:84% the only meaning that you can have 00:01:26.000 --> 00:01:29.110 align:middle line:84% is the meaning that goes with this tree. 00:01:29.110 --> 00:01:30.800 align:middle line:84% So is this all sounds vaguely familiar? 00:01:30.800 --> 00:01:32.300 align:middle line:84% Are there questions about this tree? 00:01:32.300 --> 00:01:35.030 align:middle line:84% Is anybody looking at this tree and thinking, wait, 00:01:35.030 --> 00:01:37.400 align:middle line:90% which class am I in again? 00:01:37.400 --> 00:01:38.550 align:middle line:90% What is going on? 00:01:38.550 --> 00:01:39.202 align:middle line:90% Why? 00:01:39.202 --> 00:01:41.660 align:middle line:84% I'm going to talk about this tree in a little bit of detail 00:01:41.660 --> 00:01:41.960 align:middle line:90% today. 00:01:41.960 --> 00:01:43.460 align:middle line:84% But I just want to make sure there's 00:01:43.460 --> 00:01:45.875 align:middle line:84% nothing here that's shocking or distressing anybody. 00:01:45.875 --> 00:01:48.470 align:middle line:90% 00:01:48.470 --> 00:01:51.200 align:middle line:84% So one of the things we've talked about 00:01:51.200 --> 00:01:53.360 align:middle line:84% is that some of the properties of that tree 00:01:53.360 --> 00:01:55.560 align:middle line:90% follow from selection. 00:01:55.560 --> 00:01:57.080 align:middle line:84% So we have this idea that there's 00:01:57.080 --> 00:02:02.880 align:middle line:84% a special kind of relation that can hold between heads, 00:02:02.880 --> 00:02:05.450 align:middle line:84% so the smallest things in the tree, 00:02:05.450 --> 00:02:07.970 align:middle line:84% the things that just have words under them, 00:02:07.970 --> 00:02:10.729 align:middle line:84% things like the verb, a relation that 00:02:10.729 --> 00:02:13.400 align:middle line:84% can hold between those heads and other kinds of phrases 00:02:13.400 --> 00:02:14.250 align:middle line:90% in the sentence. 00:02:14.250 --> 00:02:17.810 align:middle line:84% So we've said that the verb "tickle" selects the object 00:02:17.810 --> 00:02:19.010 align:middle line:90% noun phrase, "the child." 00:02:19.010 --> 00:02:23.570 align:middle line:84% And what we mean by that is that the relation between the verb 00:02:23.570 --> 00:02:26.030 align:middle line:84% "tickle" and the object, "the child," 00:02:26.030 --> 00:02:28.790 align:middle line:84% is special in that not every verb can 00:02:28.790 --> 00:02:31.580 align:middle line:90% be followed by an object. 00:02:31.580 --> 00:02:33.440 align:middle line:90% So you can tickle a child. 00:02:33.440 --> 00:02:34.567 align:middle line:90% You can devour a child. 00:02:34.567 --> 00:02:35.150 align:middle line:90% You shouldn't. 00:02:35.150 --> 00:02:36.230 align:middle line:90% But it's grammatical. 00:02:36.230 --> 00:02:39.230 align:middle line:90% You can write a child. 00:02:39.230 --> 00:02:42.230 align:middle line:84% But there are other verbs for which you just can't do this. 00:02:42.230 --> 00:02:43.700 align:middle line:90% You can't "thrive the child." 00:02:43.700 --> 00:02:45.840 align:middle line:90% That doesn't make any sense. 00:02:45.840 --> 00:02:47.918 align:middle line:84% So there are some verbs-- classic way 00:02:47.918 --> 00:02:50.210 align:middle line:84% to say this is there are some verbs that are transitive 00:02:50.210 --> 00:02:52.280 align:middle line:84% and others that are intransitive. 00:02:52.280 --> 00:02:54.830 align:middle line:84% So in order to know whether you can have an object 00:02:54.830 --> 00:02:58.100 align:middle line:84% or not, whether there can be a noun phrase in that position 00:02:58.100 --> 00:03:00.372 align:middle line:84% or not, the sister of the verb, you've 00:03:00.372 --> 00:03:01.580 align:middle line:90% got to know what the verb is. 00:03:01.580 --> 00:03:04.800 align:middle line:84% The verb tells you whether it needs to have an object or not. 00:03:04.800 --> 00:03:06.470 align:middle line:84% So there's this relation of selection 00:03:06.470 --> 00:03:09.020 align:middle line:84% that holds between the verb and the object. 00:03:09.020 --> 00:03:11.630 align:middle line:84% The idea is when we look up words 00:03:11.630 --> 00:03:13.760 align:middle line:84% like "tickle" and "devour" and "write" 00:03:13.760 --> 00:03:17.670 align:middle line:84% in our mental lexicon, we'll see various things about them, 00:03:17.670 --> 00:03:19.700 align:middle line:84% including how they're pronounced, 00:03:19.700 --> 00:03:22.920 align:middle line:84% but also whether they can have objects or not, 00:03:22.920 --> 00:03:26.160 align:middle line:84% whether they select for noun phrase sisters or not. 00:03:26.160 --> 00:03:28.400 align:middle line:84% So when we ask, going back to the tree, 00:03:28.400 --> 00:03:34.953 align:middle line:84% why is the sister of the verb, the noun phrase "the child," 00:03:34.953 --> 00:03:37.370 align:middle line:84% one way of interpreting that question, the answer is well, 00:03:37.370 --> 00:03:40.100 align:middle line:84% it's because the verb "tickle" selects for a noun phrase 00:03:40.100 --> 00:03:41.180 align:middle line:90% sister. 00:03:41.180 --> 00:03:42.940 align:middle line:90% And so you give it one. 00:03:42.940 --> 00:03:46.400 align:middle line:84% Now, that's why the tree has that bit of structure. 00:03:46.400 --> 00:03:48.230 align:middle line:84% And we can tell that kind of story 00:03:48.230 --> 00:03:49.920 align:middle line:84% about a lot of parts of this tree. 00:03:49.920 --> 00:03:51.380 align:middle line:84% So similarly, we're going to want 00:03:51.380 --> 00:03:53.600 align:middle line:84% to say that the preposition "with" selects 00:03:53.600 --> 00:03:56.972 align:middle line:84% for a sister that's a noun phrase, "the feather." 00:03:56.972 --> 00:03:58.430 align:middle line:84% And for that matter, we'll probably 00:03:58.430 --> 00:04:00.860 align:middle line:84% want to say that the T, "will," is selecting 00:04:00.860 --> 00:04:03.170 align:middle line:84% for a verb phrase that's why the T has a verb phrase, 00:04:03.170 --> 00:04:04.463 align:middle line:90% it's its sister. 00:04:04.463 --> 00:04:06.380 align:middle line:84% So there are various places in this tree where 00:04:06.380 --> 00:04:08.300 align:middle line:90% a head has a particular sister. 00:04:08.300 --> 00:04:10.100 align:middle line:84% And it's because the head is selecting 00:04:10.100 --> 00:04:11.180 align:middle line:90% for that as its sister. 00:04:11.180 --> 00:04:16.540 align:middle line:90% 00:04:16.540 --> 00:04:21.160 align:middle line:84% We also said that this relation of selection that is something 00:04:21.160 --> 00:04:23.410 align:middle line:84% that heads can do-- they can select for properties 00:04:23.410 --> 00:04:24.970 align:middle line:90% of their sister-- 00:04:24.970 --> 00:04:28.160 align:middle line:84% that sometimes, at least, they select more specifically 00:04:28.160 --> 00:04:28.660 align:middle line:90% than that. 00:04:28.660 --> 00:04:33.070 align:middle line:84% They select for sisters where the sister's head 00:04:33.070 --> 00:04:34.520 align:middle line:90% has a particular property. 00:04:34.520 --> 00:04:37.240 align:middle line:84% So I think the example I gave you last time was 00:04:37.240 --> 00:04:39.070 align:middle line:90% the verb depends-- 00:04:39.070 --> 00:04:42.860 align:middle line:84% where the verb depends needs a prepositional phrase sister, 00:04:42.860 --> 00:04:44.860 align:middle line:84% but it specifically needs a prepositional phrase 00:04:44.860 --> 00:04:47.230 align:middle line:84% sister in which the preposition is "on." 00:04:47.230 --> 00:04:48.880 align:middle line:90% So you have to depend on things. 00:04:48.880 --> 00:04:51.220 align:middle line:84% You can't depend from them or depend at them. 00:04:51.220 --> 00:04:53.050 align:middle line:84% You can't just have any preposition there. 00:04:53.050 --> 00:04:55.850 align:middle line:90% It has to be "on." 00:04:55.850 --> 00:04:58.030 align:middle line:84% So there are verbs like "depend--" 00:04:58.030 --> 00:04:59.980 align:middle line:84% and there are many verbs like this-- 00:04:59.980 --> 00:05:02.643 align:middle line:84% that select for a particular prepositional phrase-- sisters, 00:05:02.643 --> 00:05:04.810 align:middle line:84% but in particular, prepositional phrase sisters that 00:05:04.810 --> 00:05:07.540 align:middle line:90% have a particular preposition. 00:05:07.540 --> 00:05:09.590 align:middle line:84% Sometimes it's less particular than that. 00:05:09.590 --> 00:05:15.910 align:middle line:90% So I think in class I said-- 00:05:15.910 --> 00:05:18.460 align:middle line:84% (he said, vamping until he could get some chalk out)-- 00:05:18.460 --> 00:05:22.540 align:middle line:84% that the verb "put" seems to select 00:05:22.540 --> 00:05:25.270 align:middle line:84% for a noun phrase sister and prepositional phrase sister. 00:05:25.270 --> 00:05:36.970 align:middle line:84% So you put the book "on the table" or "in the refrigerator" 00:05:36.970 --> 00:05:41.320 align:middle line:84% or "under the car" or whatever, wherever 00:05:41.320 --> 00:05:43.280 align:middle line:90% it is you want to put the book. 00:05:43.280 --> 00:05:45.763 align:middle line:84% So the "put" has to have a noun phrase object, 00:05:45.763 --> 00:05:46.930 align:middle line:90% doesn't have to be the book. 00:05:46.930 --> 00:05:48.013 align:middle line:90% It can be any noun phrase. 00:05:48.013 --> 00:05:51.040 align:middle line:84% And then there can be a prepositional phrase. 00:05:51.040 --> 00:05:52.222 align:middle line:90% "Put" is not like "depend." 00:05:52.222 --> 00:05:54.430 align:middle line:84% There are a variety of prepositions that can be here. 00:05:54.430 --> 00:05:56.305 align:middle line:84% But you can't just have any preposition here. 00:05:56.305 --> 00:05:58.900 align:middle line:84% So you can't say "Put the book during the party." 00:05:58.900 --> 00:06:02.320 align:middle line:90% 00:06:02.320 --> 00:06:06.700 align:middle line:90% That's not a possible sentence. 00:06:06.700 --> 00:06:09.640 align:middle line:84% And we might hope to derive that from properties 00:06:09.640 --> 00:06:11.080 align:middle line:90% of the meaning of "put," right? 00:06:11.080 --> 00:06:13.360 align:middle line:84% So the prepositional phrase that goes with "put" 00:06:13.360 --> 00:06:15.760 align:middle line:84% needs to specify a location, right? 00:06:15.760 --> 00:06:19.420 align:middle line:84% And "During the party" doesn't specify a location unless we're 00:06:19.420 --> 00:06:22.540 align:middle line:84% thinking of time as space, which as long as syntax 00:06:22.540 --> 00:06:25.420 align:middle line:84% is not Doctor Who, I think we're OK. 00:06:25.420 --> 00:06:25.990 align:middle line:90% Yeah? 00:06:25.990 --> 00:06:29.737 align:middle line:84% AUDIENCE: So for clarity, because Merge only is binary-- 00:06:29.737 --> 00:06:30.820 align:middle line:90% NORVIN RICHARDS: Oh, dear. 00:06:30.820 --> 00:06:31.320 align:middle line:90% [LAUGHS] 00:06:31.320 --> 00:06:33.340 align:middle line:90% AUDIENCE: [INAUDIBLE] 00:06:33.340 --> 00:06:34.810 align:middle line:90% NORVIN RICHARDS: Yeah. 00:06:34.810 --> 00:06:36.250 align:middle line:90% So Merge is only binary. 00:06:36.250 --> 00:06:38.590 align:middle line:84% And that means if we're going to have 00:06:38.590 --> 00:06:40.510 align:middle line:84% a tree for these kinds of sentences, 00:06:40.510 --> 00:06:43.960 align:middle line:84% we're going to need to put the verb "put." 00:06:43.960 --> 00:06:45.970 align:middle line:84% And since we want the noun phrase 00:06:45.970 --> 00:06:48.490 align:middle line:84% to be next-- that's what this seems to tell us-- 00:06:48.490 --> 00:06:49.840 align:middle line:90% we'll have a noun phrase here. 00:06:49.840 --> 00:06:54.390 align:middle line:90% 00:06:54.390 --> 00:06:59.037 align:middle line:84% And then we'll have a prepositional phrase 00:06:59.037 --> 00:07:00.870 align:middle line:84% That's one possible answer to your question. 00:07:00.870 --> 00:07:02.203 align:middle line:90% If Merge is going to be binary-- 00:07:02.203 --> 00:07:06.648 align:middle line:90% 00:07:06.648 --> 00:07:08.190 align:middle line:84% if Merge is going to be binary, we're 00:07:08.190 --> 00:07:10.710 align:middle line:84% going to want to say, yeah, "put" 00:07:10.710 --> 00:07:13.582 align:middle line:84% is something that needs two things. 00:07:13.582 --> 00:07:15.790 align:middle line:84% We're going to have to give them to it one at a time. 00:07:15.790 --> 00:07:17.410 align:middle line:84% We can't give them both to it at once. 00:07:17.410 --> 00:07:19.070 align:middle line:84% Otherwise, if we wanted to do that, 00:07:19.070 --> 00:07:20.820 align:middle line:84% we would need to use ternary Merge, right? 00:07:20.820 --> 00:07:22.660 align:middle line:84% We'd need to merge three things at once. 00:07:22.660 --> 00:07:24.270 align:middle line:84% And so we'll merge them one at a time. 00:07:24.270 --> 00:07:25.770 align:middle line:84% We'll first give it the noun phrase. 00:07:25.770 --> 00:07:27.300 align:middle line:84% And then we'll give it the prepositional phrase. 00:07:27.300 --> 00:07:29.220 align:middle line:84% Why do we first give it the noun phrase and then 00:07:29.220 --> 00:07:30.262 align:middle line:90% the prepositional phrase? 00:07:30.262 --> 00:07:33.000 align:middle line:84% Why don't we say, "Put on the table the book"? 00:07:33.000 --> 00:07:34.530 align:middle line:90% We might actually get to that. 00:07:34.530 --> 00:07:36.850 align:middle line:84% But we're not going to get to it today. 00:07:36.850 --> 00:07:39.150 align:middle line:84% So this is one possible answer to your question. 00:07:39.150 --> 00:07:40.960 align:middle line:84% Another possible answer to your question, of course, 00:07:40.960 --> 00:07:42.510 align:middle line:84% would be to say, ah, we've discovered 00:07:42.510 --> 00:07:45.120 align:middle line:84% that we were wrong to say that there's binary Merge, right? 00:07:45.120 --> 00:07:48.288 align:middle line:84% That here's a place where we need there to be ternary merge. 00:07:48.288 --> 00:07:49.830 align:middle line:84% This is going to interact with things 00:07:49.830 --> 00:07:51.820 align:middle line:84% we're going to want to say about how selection works. 00:07:51.820 --> 00:07:53.320 align:middle line:84% So I have now a couple of times said 00:07:53.320 --> 00:07:56.160 align:middle line:84% a thing gets to select for the properties of its sister, 00:07:56.160 --> 00:07:56.970 align:middle line:90% right? 00:07:56.970 --> 00:07:59.310 align:middle line:84% And here it's selecting for properties of its sister. 00:07:59.310 --> 00:08:00.990 align:middle line:84% But this is not its sister, right? 00:08:00.990 --> 00:08:02.760 align:middle line:90% This is something else. 00:08:02.760 --> 00:08:04.800 align:middle line:90% It's-- never mind. 00:08:04.800 --> 00:08:07.200 align:middle line:84% I can't figure out the family relation between these two 00:08:07.200 --> 00:08:08.040 align:middle line:90% points on the tree. 00:08:08.040 --> 00:08:11.380 align:middle line:90% It's-- what is it? 00:08:11.380 --> 00:08:12.628 align:middle line:90% It's its aunt, yeah. 00:08:12.628 --> 00:08:14.670 align:middle line:84% The prepositional phrase is the aunt of the verb. 00:08:14.670 --> 00:08:16.850 align:middle line:84% That's not how actual syntacticians 00:08:16.850 --> 00:08:18.100 align:middle line:90% talk about this kind of thing. 00:08:18.100 --> 00:08:20.550 align:middle line:90% So they're not sisters. 00:08:20.550 --> 00:08:22.320 align:middle line:84% So two kinds of things we can say. 00:08:22.320 --> 00:08:25.650 align:middle line:84% Oh, what we're discovering is that we need ternary Merge. 00:08:25.650 --> 00:08:27.750 align:middle line:84% We need these to both be sisters. 00:08:27.750 --> 00:08:30.180 align:middle line:84% Another would be to say, oh, we're 00:08:30.180 --> 00:08:33.789 align:middle line:84% discovering that selection isn't necessarily 00:08:33.789 --> 00:08:35.039 align:middle line:90% for properties of your sister. 00:08:35.039 --> 00:08:38.350 align:middle line:84% It's something more like you select things. 00:08:38.350 --> 00:08:41.418 align:middle line:84% The first things that you merge must be things that you select. 00:08:41.418 --> 00:08:43.960 align:middle line:84% We're going to circle around and talk about that in a second. 00:08:43.960 --> 00:08:46.905 align:middle line:84% So to say something more elaborate about how you select. 00:08:46.905 --> 00:08:48.280 align:middle line:84% So that's a really good question. 00:08:48.280 --> 00:08:49.740 align:middle line:84% What we really need to do is develop 00:08:49.740 --> 00:08:51.240 align:middle line:84% tests that will tell us whether this 00:08:51.240 --> 00:08:53.340 align:middle line:84% is the right tree for something like this, 00:08:53.340 --> 00:08:56.910 align:middle line:84% or whether we would rather have something like this, where 00:08:56.910 --> 00:08:59.910 align:middle line:84% there's a verb and a noun phrase and prepositional phrase, 00:08:59.910 --> 00:09:03.540 align:middle line:84% and Merge doesn't have to be binary. 00:09:03.540 --> 00:09:07.290 align:middle line:84% Just to cheat a little bit, what we would find out 00:09:07.290 --> 00:09:08.790 align:middle line:84% when we develop these kinds of tests 00:09:08.790 --> 00:09:10.873 align:middle line:84% is that this is not the kind of tree that we want. 00:09:10.873 --> 00:09:12.510 align:middle line:84% We want trees that are like this. 00:09:12.510 --> 00:09:14.520 align:middle line:84% But we're going to have to do some work 00:09:14.520 --> 00:09:16.810 align:middle line:90% to develop those tests. 00:09:16.810 --> 00:09:17.310 align:middle line:90% Raquel? 00:09:17.310 --> 00:09:19.050 align:middle line:84% AUDIENCE: I have a scary thought. 00:09:19.050 --> 00:09:20.112 align:middle line:90% NORVIN RICHARDS: Oh, no. 00:09:20.112 --> 00:09:22.770 align:middle line:84% AUDIENCE: The thought is like-- so about selected-- 00:09:22.770 --> 00:09:23.100 align:middle line:90% NORVIN RICHARDS: Yeah. 00:09:23.100 --> 00:09:24.240 align:middle line:84% AUDIENCE: I feel like there are some verbs, 00:09:24.240 --> 00:09:25.157 align:middle line:90% like, you say "board." 00:09:25.157 --> 00:09:26.100 align:middle line:90% Like you board a ship. 00:09:26.100 --> 00:09:28.403 align:middle line:84% You kind of have, like, a secret selection inside of it 00:09:28.403 --> 00:09:29.820 align:middle line:84% because you could think of "board" 00:09:29.820 --> 00:09:31.005 align:middle line:90% as like get on the ship. 00:09:31.005 --> 00:09:32.130 align:middle line:90% NORVIN RICHARDS: Oh, I see. 00:09:32.130 --> 00:09:33.838 align:middle line:84% AUDIENCE: And so you'd be like, ah, "get" 00:09:33.838 --> 00:09:36.020 align:middle line:84% is selecting for on the ship, a certain type 00:09:36.020 --> 00:09:37.020 align:middle line:90% of prepositional phrase. 00:09:37.020 --> 00:09:37.260 align:middle line:90% NORVIN RICHARDS: Mm-hmm. 00:09:37.260 --> 00:09:40.230 align:middle line:84% AUDIENCE: Or you could have just "board," which kind of includes 00:09:40.230 --> 00:09:40.920 align:middle line:90% it inside of it. 00:09:40.920 --> 00:09:43.500 align:middle line:84% Like, also in Spanish, if you have words like "buscar," which 00:09:43.500 --> 00:09:46.050 align:middle line:84% is, like, "to look for," and so it's, like, 00:09:46.050 --> 00:09:50.490 align:middle line:84% could you think of "put" as a verb that maybe-- 00:09:50.490 --> 00:09:52.120 align:middle line:84% like if English had been different, 00:09:52.120 --> 00:09:55.308 align:middle line:84% you could have a verb like "You put the refrigerator." 00:09:55.308 --> 00:09:57.600 align:middle line:84% Everyone knows that you mean "put in the refrigerator." 00:09:57.600 --> 00:09:58.650 align:middle line:90% NORVIN RICHARDS: Yeah. 00:09:58.650 --> 00:09:59.483 align:middle line:90% I see what you mean. 00:09:59.483 --> 00:10:04.620 align:middle line:84% So you're pointing out something important, which is that-- 00:10:04.620 --> 00:10:06.030 align:middle line:90% here, let's do "look for." 00:10:06.030 --> 00:10:07.890 align:middle line:84% But in English, we're going to want 00:10:07.890 --> 00:10:10.320 align:middle line:84% to have a verb "look" that selects 00:10:10.320 --> 00:10:13.455 align:middle line:84% for a prepositional phrase, where the preposition is "for." 00:10:13.455 --> 00:10:16.320 align:middle line:90% 00:10:16.320 --> 00:10:18.420 align:middle line:84% But that there are other languages like Spanish 00:10:18.420 --> 00:10:20.628 align:middle line:84% in which there's a single verb that means "look for." 00:10:20.628 --> 00:10:23.730 align:middle line:84% For that matter, in English, there's a verb "seek." 00:10:23.730 --> 00:10:25.350 align:middle line:90% So it's kind of dramatic. 00:10:25.350 --> 00:10:28.717 align:middle line:84% But if I go into a bookstore and say, "I'm looking for a book," 00:10:28.717 --> 00:10:30.300 align:middle line:84% that's the normal thing for me to say. 00:10:30.300 --> 00:10:33.970 align:middle line:84% I could also go into a bookstore and say "I am seeking a book." 00:10:33.970 --> 00:10:36.990 align:middle line:84% They would probably direct me to the fantasy, science fiction 00:10:36.990 --> 00:10:40.420 align:middle line:90% section if I did that. 00:10:40.420 --> 00:10:43.170 align:middle line:84% But those are both two ways to say the same thing. 00:10:43.170 --> 00:10:46.580 align:middle line:84% So maybe the way to think about this is, yeah, Spanish 00:10:46.580 --> 00:10:48.330 align:middle line:84% doesn't have a verb that means "look for." 00:10:48.330 --> 00:10:50.100 align:middle line:90% It has a verb that means "seek." 00:10:50.100 --> 00:10:54.060 align:middle line:90% And-- there. 00:10:54.060 --> 00:10:54.780 align:middle line:90% Yeah. 00:10:54.780 --> 00:10:56.197 align:middle line:84% Or maybe another way to say this-- 00:10:56.197 --> 00:10:58.560 align:middle line:84% and this is going to be important as you guys are doing 00:10:58.560 --> 00:11:01.085 align:middle line:84% your work with the languages that you're working on-- 00:11:01.085 --> 00:11:02.460 align:middle line:84% is that you shouldn't necessarily 00:11:02.460 --> 00:11:05.250 align:middle line:84% expect the selectional properties of a verb 00:11:05.250 --> 00:11:08.280 align:middle line:84% to be the same from one language to the next. 00:11:08.280 --> 00:11:11.205 align:middle line:84% That if you are inclined to think of Spanish-- 00:11:11.205 --> 00:11:13.140 align:middle line:90% is it "buscar"?-- 00:11:13.140 --> 00:11:15.570 align:middle line:84% as the equivalent of "look for," that we 00:11:15.570 --> 00:11:18.660 align:middle line:84% want to think of that as-- as-- that 00:11:18.660 --> 00:11:20.655 align:middle line:84% English has a verb "look" that selects 00:11:20.655 --> 00:11:21.780 align:middle line:90% for a prepositional phrase. 00:11:21.780 --> 00:11:24.090 align:middle line:84% Spanish has a verb "buscar" that's like "seek." 00:11:24.090 --> 00:11:27.070 align:middle line:84% It selects for a noun phrase complement. 00:11:27.070 --> 00:11:27.570 align:middle line:90% Yeah. 00:11:27.570 --> 00:11:29.370 align:middle line:84% So you want to be careful as you are going 00:11:29.370 --> 00:11:31.170 align:middle line:90% from language to language. 00:11:31.170 --> 00:11:34.920 align:middle line:84% You can't trust selectional restrictions. 00:11:34.920 --> 00:11:38.700 align:middle line:84% And indeed, so I've used "put" as the example 00:11:38.700 --> 00:11:42.690 align:middle line:84% of a verb that has two things that it seems to select. 00:11:42.690 --> 00:11:44.700 align:middle line:84% But there are probably examples of verbs 00:11:44.700 --> 00:11:48.690 align:middle line:84% that select two things, where the two things are, say, 00:11:48.690 --> 00:11:49.930 align:middle line:90% both noun phrases. 00:11:49.930 --> 00:11:57.450 align:middle line:84% So take a verb like "give the children books." 00:11:57.450 --> 00:12:01.500 align:middle line:84% So here we want there to be two noun phrases coming together 00:12:01.500 --> 00:12:02.070 align:middle line:90% with "give." 00:12:02.070 --> 00:12:04.110 align:middle line:90% And "give" selects those. 00:12:04.110 --> 00:12:06.960 align:middle line:84% It's a property of "give" that it can 00:12:06.960 --> 00:12:08.490 align:middle line:90% have two noun phrases after it. 00:12:08.490 --> 00:12:09.900 align:middle line:90% And not every verb can do that. 00:12:09.900 --> 00:12:12.840 align:middle line:90% "Put," for example, can't. 00:12:12.840 --> 00:12:15.900 align:middle line:84% So even verbs that are fairly similar in their meaning, 00:12:15.900 --> 00:12:16.980 align:middle line:90% like "donate"-- 00:12:16.980 --> 00:12:18.750 align:middle line:84% so you can't "donate the children books." 00:12:18.750 --> 00:12:21.780 align:middle line:84% You have to donate books to the children. 00:12:21.780 --> 00:12:25.080 align:middle line:84% So yes, as so often in this class, 00:12:25.080 --> 00:12:27.690 align:middle line:84% I'm showing you the easiest examples here. 00:12:27.690 --> 00:12:29.580 align:middle line:84% And as so often in this class, you guys 00:12:29.580 --> 00:12:30.870 align:middle line:84% are immediately saying, wait, wait, what 00:12:30.870 --> 00:12:32.162 align:middle line:90% about the complicated examples? 00:12:32.162 --> 00:12:34.200 align:middle line:84% So, yes, there are more complicated examples. 00:12:34.200 --> 00:12:35.490 align:middle line:84% We'll try to get to them, cases where you're 00:12:35.490 --> 00:12:36.740 align:middle line:90% selecting more than one thing. 00:12:36.740 --> 00:12:38.072 align:middle line:90% We'll come back to that. 00:12:38.072 --> 00:12:39.280 align:middle line:90% We'll talk about it more now. 00:12:39.280 --> 00:12:39.780 align:middle line:90% Yes? 00:12:39.780 --> 00:12:41.973 align:middle line:84% AUDIENCE: So for give the children books, 00:12:41.973 --> 00:12:43.140 align:middle line:90% how would you continue that? 00:12:43.140 --> 00:12:45.670 align:middle line:90% [INAUDIBLE]? 00:12:45.670 --> 00:12:47.940 align:middle line:84% NORVIN RICHARDS: So if we use the technology 00:12:47.940 --> 00:12:51.240 align:middle line:84% that we have right now, we're kind of compelled 00:12:51.240 --> 00:12:52.890 align:middle line:90% to have a tree like this one. 00:12:52.890 --> 00:13:02.610 align:middle line:90% 00:13:02.610 --> 00:13:04.350 align:middle line:84% I wrote this as capital N because I 00:13:04.350 --> 00:13:06.690 align:middle line:84% knew I was about to write a noun phrase. 00:13:06.690 --> 00:13:08.220 align:middle line:90% "Give the children books." 00:13:08.220 --> 00:13:12.660 align:middle line:84% So where "give" needs to-- to combine with two noun phrases. 00:13:12.660 --> 00:13:13.770 align:middle line:90% And Merge is binary. 00:13:13.770 --> 00:13:16.490 align:middle line:84% And so we'll do binary merge twice. 00:13:16.490 --> 00:13:18.990 align:middle line:84% That's a kind of tree that we could draw with the techniques 00:13:18.990 --> 00:13:21.630 align:middle line:90% that we have right now. 00:13:21.630 --> 00:13:29.230 align:middle line:84% I hope I hedged enough in there to make you feel emotionally 00:13:29.230 --> 00:13:33.160 align:middle line:84% grounded enough that if it turns out later that that's not 00:13:33.160 --> 00:13:36.580 align:middle line:84% the right tree, that you won't be too disappointed 00:13:36.580 --> 00:13:38.830 align:middle line:84% because we're going to talk more about this stuff. 00:13:38.830 --> 00:13:41.432 align:middle line:84% And our understanding will get more sophisticated. 00:13:41.432 --> 00:13:43.390 align:middle line:84% But given everything I'm telling you right now, 00:13:43.390 --> 00:13:45.890 align:middle line:90% that's the tree we expect. 00:13:45.890 --> 00:13:47.690 align:middle line:90% Yes? 00:13:47.690 --> 00:13:51.775 align:middle line:84% AUDIENCE: Do these trees have to have the words in sentence? 00:13:51.775 --> 00:13:52.900 align:middle line:90% NORVIN RICHARDS: I'm sorry. 00:13:52.900 --> 00:13:55.480 align:middle line:90% I missed the middle of that. 00:13:55.480 --> 00:13:58.150 align:middle line:84% AUDIENCE: Do these trees have to have 00:13:58.150 --> 00:14:00.880 align:middle line:84% exactly each and every word in each sentence? 00:14:00.880 --> 00:14:05.590 align:middle line:90% Or could it fill in some words? 00:14:05.590 --> 00:14:07.870 align:middle line:84% For example, I learned in elementary school 00:14:07.870 --> 00:14:10.180 align:middle line:84% the subject of every command is "you." 00:14:10.180 --> 00:14:11.485 align:middle line:90% NORVIN RICHARDS: Oh, mm-hmm. 00:14:11.485 --> 00:14:15.976 align:middle line:84% AUDIENCE: Could you have "you" in parentheses 00:14:15.976 --> 00:14:17.085 align:middle line:90% in one of these trees? 00:14:17.085 --> 00:14:18.210 align:middle line:90% NORVIN RICHARDS: Oh, I see. 00:14:18.210 --> 00:14:19.918 align:middle line:84% AUDIENCE: "Give (to) the children books." 00:14:19.918 --> 00:14:20.970 align:middle line:90% NORVIN RICHARDS: Yeah. 00:14:20.970 --> 00:14:21.810 align:middle line:90% Oh, I see. 00:14:21.810 --> 00:14:23.790 align:middle line:90% Oh. 00:14:23.790 --> 00:14:27.220 align:middle line:84% That's a really interesting and difficult question. 00:14:27.220 --> 00:14:29.790 align:middle line:90% The short answer is yes. 00:14:29.790 --> 00:14:31.740 align:middle line:84% There absolutely are places where 00:14:31.740 --> 00:14:34.930 align:middle line:84% we're going to want to say I'm showing you trees in which I'm 00:14:34.930 --> 00:14:36.180 align:middle line:90% not showing you trees anymore. 00:14:36.180 --> 00:14:37.710 align:middle line:90% I'm showing you a nebula. 00:14:37.710 --> 00:14:40.440 align:middle line:90% Give me just a second. 00:14:40.440 --> 00:14:45.600 align:middle line:84% I'm showing you trees in which every terminal node, 00:14:45.600 --> 00:14:49.350 align:middle line:84% every head has one word under it. 00:14:49.350 --> 00:14:51.750 align:middle line:84% And you're wondering, could there 00:14:51.750 --> 00:14:54.930 align:middle line:84% be trees where there are nodes that 00:14:54.930 --> 00:14:57.060 align:middle line:84% don't have anything under them or don't have 00:14:57.060 --> 00:15:00.570 align:middle line:90% anything pronounced under them? 00:15:00.570 --> 00:15:05.070 align:middle line:84% And indeed the short answer is yes, there surely 00:15:05.070 --> 00:15:06.780 align:middle line:90% are going to be such trees. 00:15:06.780 --> 00:15:10.320 align:middle line:90% Your example is a good one. 00:15:10.320 --> 00:15:15.730 align:middle line:84% Do I want to go off and talk about that? 00:15:15.730 --> 00:15:17.910 align:middle line:84% Yeah, let's talk about that for just a second. 00:15:17.910 --> 00:15:24.510 align:middle line:90% 00:15:24.510 --> 00:15:26.100 align:middle line:90% We talked a little bit before-- 00:15:26.100 --> 00:15:28.080 align:middle line:84% we're going to come back to this later. 00:15:28.080 --> 00:15:31.680 align:middle line:84% But we talked a little bit before about-- 00:15:31.680 --> 00:15:35.100 align:middle line:90% yeah, we can do this fast-- 00:15:35.100 --> 00:15:39.750 align:middle line:84% constraints on things that pronouns and names 00:15:39.750 --> 00:15:41.380 align:middle line:90% can refer to. 00:15:41.380 --> 00:15:44.400 align:middle line:84% So for example, we talked about the fact that if you say, 00:15:44.400 --> 00:15:49.800 align:middle line:84% "She likes Sally," that's a perfectly fine sentence. 00:15:49.800 --> 00:15:55.360 align:middle line:84% But it can't mean that she and Sally are the same person, 00:15:55.360 --> 00:15:55.860 align:middle line:90% right? 00:15:55.860 --> 00:15:58.590 align:middle line:84% So "she" has to refer to somebody other than Sally. 00:15:58.590 --> 00:16:02.828 align:middle line:84% And we developed a rule that has that consequence. 00:16:02.828 --> 00:16:05.370 align:middle line:84% And I promised you that we would talk about that rule in more 00:16:05.370 --> 00:16:07.390 align:middle line:90% depth later on. 00:16:07.390 --> 00:16:10.660 align:middle line:90% Here's another kind of case. 00:16:10.660 --> 00:16:20.040 align:middle line:84% If you say something like "Sally defended herself," 00:16:20.040 --> 00:16:23.100 align:middle line:84% there's this special expression "herself." 00:16:23.100 --> 00:16:25.950 align:middle line:84% And it has to refer to Sally, right? 00:16:25.950 --> 00:16:29.490 align:middle line:84% So "Sally defended herself" means Sally defended Sally. 00:16:29.490 --> 00:16:33.420 align:middle line:84% So "herself" refers back to whoever the subject is. 00:16:33.420 --> 00:16:35.610 align:middle line:84% If this had been "Mary defended herself," well, 00:16:35.610 --> 00:16:37.500 align:middle line:84% it would have to be Mary defending Mary. 00:16:37.500 --> 00:16:44.970 align:middle line:84% If it had been, "I defended myself," well, 00:16:44.970 --> 00:16:48.000 align:middle line:84% then you have to use this special form of these words, 00:16:48.000 --> 00:16:49.290 align:middle line:90% these "-self" words. 00:16:49.290 --> 00:16:50.400 align:middle line:90% They're called reflexives. 00:16:50.400 --> 00:16:55.950 align:middle line:90% 00:16:55.950 --> 00:16:58.110 align:middle line:84% These reflexives have to refer back 00:16:58.110 --> 00:16:59.670 align:middle line:84% to something else in the sentence. 00:16:59.670 --> 00:17:02.867 align:middle line:84% In these examples, they always refer to the subject. 00:17:02.867 --> 00:17:04.200 align:middle line:90% And they have to do that, right? 00:17:04.200 --> 00:17:15.280 align:middle line:84% So you can't say things like, "I defended herself" or "himself," 00:17:15.280 --> 00:17:15.780 align:middle line:90% right? 00:17:15.780 --> 00:17:17.069 align:middle line:90% These are out, yeah? 00:17:17.069 --> 00:17:20.490 align:middle line:84% So the reflexive that's at the end of these sentences 00:17:20.490 --> 00:17:23.250 align:middle line:84% has to refer back to the subject in these cases. 00:17:23.250 --> 00:17:25.230 align:middle line:84% And the form of the reflexive tells you 00:17:25.230 --> 00:17:26.730 align:middle line:84% something about what the subject is. 00:17:26.730 --> 00:17:29.640 align:middle line:90% So you get, it's got-- 00:17:29.640 --> 00:17:30.540 align:middle line:90% so-- hmm. 00:17:30.540 --> 00:17:32.100 align:middle line:84% You get "herself" when the subject 00:17:32.100 --> 00:17:33.870 align:middle line:90% is feminine and third person. 00:17:33.870 --> 00:17:36.870 align:middle line:84% And you got "myself" when the subject is first person. 00:17:36.870 --> 00:17:42.750 align:middle line:84% And if it were "you defended," it would 00:17:42.750 --> 00:17:45.570 align:middle line:90% be "defended yourself," yeah? 00:17:45.570 --> 00:17:47.530 align:middle line:84% So this reflexive is telling you that it's 00:17:47.530 --> 00:17:48.780 align:middle line:90% referring back to "you," yeah? 00:17:48.780 --> 00:17:50.700 align:middle line:84% This means you defended you, yeah. 00:17:50.700 --> 00:17:53.367 align:middle line:84% I'm telling you all this because your example 00:17:53.367 --> 00:17:55.200 align:middle line:84% from elementary school is a really good one. 00:17:55.200 --> 00:18:02.450 align:middle line:84% If I tell you, defend yourself, it's a command. 00:18:02.450 --> 00:18:04.950 align:middle line:84% Well, everything that we've now learned about these reflexes 00:18:04.950 --> 00:18:08.010 align:middle line:84% leads us to hope that there's a "you" in that sentence 00:18:08.010 --> 00:18:10.710 align:middle line:84% because, well, that's what reflexives seemed to do. 00:18:10.710 --> 00:18:13.020 align:middle line:90% They refer back to the subject. 00:18:13.020 --> 00:18:16.017 align:middle line:84% And they have a form that tells you something 00:18:16.017 --> 00:18:17.100 align:middle line:90% about what the subject is. 00:18:17.100 --> 00:18:18.690 align:middle line:84% In this case, they tell you that the subject 00:18:18.690 --> 00:18:21.240 align:middle line:84% is "you," which is what you were taught in elementary school, 00:18:21.240 --> 00:18:22.330 align:middle line:90% yeah. 00:18:22.330 --> 00:18:23.970 align:middle line:90% So you say "Defend yourself!" 00:18:23.970 --> 00:18:24.810 align:middle line:90% And that's fine. 00:18:24.810 --> 00:18:31.720 align:middle line:84% And you cannot say "Defend himself!" or "Defend myself!"-- 00:18:31.720 --> 00:18:32.220 align:middle line:90% right? 00:18:32.220 --> 00:18:34.080 align:middle line:84% Just as you cannot say you defended-- 00:18:34.080 --> 00:18:35.620 align:middle line:84% you can say you defended yourself, 00:18:35.620 --> 00:18:41.430 align:middle line:84% but not "You defended myself" or "himself" or "herself," yeah? 00:18:41.430 --> 00:18:43.080 align:middle line:84% So indeed, there are some good reasons 00:18:43.080 --> 00:18:44.872 align:middle line:84% to think that whoever told you that there's 00:18:44.872 --> 00:18:47.010 align:middle line:84% a "you" in this sentence was right. 00:18:47.010 --> 00:18:50.560 align:middle line:84% You can detect the "you" by using these reflexives, yeah. 00:18:50.560 --> 00:18:54.240 align:middle line:84% So there are, indeed, words that are not pronounced. 00:18:54.240 --> 00:18:56.740 align:middle line:84% And if we were going to draw a tree for defend yourself, 00:18:56.740 --> 00:18:58.407 align:middle line:84% we'd probably want to say, yeah, there's 00:18:58.407 --> 00:18:59.647 align:middle line:90% a "you" in subject position. 00:18:59.647 --> 00:19:01.980 align:middle line:84% And there would have to be a rule saying don't pronounce 00:19:01.980 --> 00:19:02.897 align:middle line:90% that particular "you." 00:19:02.897 --> 00:19:05.480 align:middle line:84% So the short answer to your question 00:19:05.480 --> 00:19:08.060 align:middle line:84% is, yes, there are parts of trees that are not pronounced. 00:19:08.060 --> 00:19:12.410 align:middle line:84% And we may get a chance to talk more about. 00:19:12.410 --> 00:19:12.920 align:middle line:90% All right. 00:19:12.920 --> 00:19:14.270 align:middle line:90% That was a tangent. 00:19:14.270 --> 00:19:15.950 align:middle line:90% Any other questions? 00:19:15.950 --> 00:19:17.750 align:middle line:84% Take us back to this main line here. 00:19:17.750 --> 00:19:21.948 align:middle line:90% 00:19:21.948 --> 00:19:22.490 align:middle line:90% Where are we? 00:19:22.490 --> 00:19:24.170 align:middle line:90% Yeah, so selection. 00:19:24.170 --> 00:19:26.120 align:middle line:84% Heads select for properties of their sisters, 00:19:26.120 --> 00:19:27.740 align:middle line:84% although several of you want to know 00:19:27.740 --> 00:19:29.570 align:middle line:84% what happens when they seem to have more than one thing 00:19:29.570 --> 00:19:30.470 align:middle line:90% that they select. 00:19:30.470 --> 00:19:33.050 align:middle line:90% And I'm trying to-- 00:19:33.050 --> 00:19:35.750 align:middle line:84% well, I'm not exactly ruthlessly suppressing that question. 00:19:35.750 --> 00:19:37.430 align:middle line:84% But I'm telling you what we currently 00:19:37.430 --> 00:19:40.460 align:middle line:84% seem to have to say about it and maybe flagging 00:19:40.460 --> 00:19:43.970 align:middle line:84% the possibility that we'll have to say different things later. 00:19:43.970 --> 00:19:46.660 align:middle line:90% So far so good? 00:19:46.660 --> 00:19:48.910 align:middle line:90% So, yeah, right. 00:19:48.910 --> 00:19:51.520 align:middle line:84% And when selection is for something specific-- 00:19:51.520 --> 00:19:54.640 align:middle line:84% so this was the point I was trying to make with this slide. 00:19:54.640 --> 00:19:57.250 align:middle line:84% When a verb selects for something specific, 00:19:57.250 --> 00:20:00.040 align:middle line:84% a verb can, for example, say I need a prepositional phrase. 00:20:00.040 --> 00:20:02.620 align:middle line:84% And I need the preposition to be "on." 00:20:02.620 --> 00:20:04.348 align:middle line:84% Or I need a prepositional phrase-- 00:20:04.348 --> 00:20:05.140 align:middle line:90% this is like "put." 00:20:05.140 --> 00:20:06.310 align:middle line:90% I need a prepositional phrase. 00:20:06.310 --> 00:20:08.477 align:middle line:84% And I need the preposition to be something locative. 00:20:08.477 --> 00:20:11.500 align:middle line:84% It needs to describe a location, yeah. 00:20:11.500 --> 00:20:13.780 align:middle line:84% So "depend" selects for a prepositional phrase 00:20:13.780 --> 00:20:15.770 align:middle line:90% with the head "on." 00:20:15.770 --> 00:20:18.490 align:middle line:84% We're never going to find, for example, a verb that selects 00:20:18.490 --> 00:20:19.690 align:middle line:90% for a prepositional phrase-- 00:20:19.690 --> 00:20:21.398 align:middle line:84% and I don't care what the preposition is, 00:20:21.398 --> 00:20:23.085 align:middle line:84% but the object must be "tomatoes." 00:20:23.085 --> 00:20:24.460 align:middle line:84% We won't find anything like that. 00:20:24.460 --> 00:20:28.360 align:middle line:84% So verbs select for their sisters-- 00:20:28.360 --> 00:20:31.120 align:middle line:84% sisters in all the cases I'm going to show you. 00:20:31.120 --> 00:20:32.650 align:middle line:90% They select for their sisters. 00:20:32.650 --> 00:20:34.570 align:middle line:84% But what they select specifically 00:20:34.570 --> 00:20:36.610 align:middle line:84% is sisters with a particular head. 00:20:36.610 --> 00:20:38.440 align:middle line:84% They don't say things like, I select 00:20:38.440 --> 00:20:41.380 align:middle line:84% for a sister whose complement must be this 00:20:41.380 --> 00:20:44.140 align:middle line:84% or which must contain a tomato or which 00:20:44.140 --> 00:20:45.837 align:middle line:90% must be modified by an adverb. 00:20:45.837 --> 00:20:47.170 align:middle line:90% Heads don't do things like that. 00:20:47.170 --> 00:20:49.150 align:middle line:84% They select for properties of their sisters, 00:20:49.150 --> 00:20:52.025 align:middle line:84% specifically the heads of their sisters. 00:20:52.025 --> 00:20:54.400 align:middle line:84% And once we know that-- this is what I said in class last 00:20:54.400 --> 00:20:55.300 align:middle line:90% time-- 00:20:55.300 --> 00:21:00.460 align:middle line:84% we can use that fact about selection 00:21:00.460 --> 00:21:05.620 align:middle line:84% to detect other heads, so other kinds of things 00:21:05.620 --> 00:21:08.440 align:middle line:84% that seem to be in this selection relation 00:21:08.440 --> 00:21:11.200 align:middle line:84% with other words in the sentence, other heads 00:21:11.200 --> 00:21:11.960 align:middle line:90% in the sentence. 00:21:11.960 --> 00:21:14.890 align:middle line:84% So when we see other cases where there's 00:21:14.890 --> 00:21:18.518 align:middle line:84% a particular word in the sentence whose value seems 00:21:18.518 --> 00:21:20.560 align:middle line:84% to be determined by another word in the sentence, 00:21:20.560 --> 00:21:23.050 align:middle line:84% we get to think, oh, OK, maybe that's a selection relation. 00:21:23.050 --> 00:21:25.810 align:middle line:84% This was the example I gave you last time. 00:21:25.810 --> 00:21:30.280 align:middle line:84% Verbs seem to be able to select for properties 00:21:30.280 --> 00:21:33.010 align:middle line:90% of the clause that follows them. 00:21:33.010 --> 00:21:34.990 align:middle line:84% So there are verbs like "think" that 00:21:34.990 --> 00:21:36.580 align:middle line:90% can be followed by a clause. 00:21:36.580 --> 00:21:38.980 align:middle line:84% Not every verb can be followed by a clause, right? 00:21:38.980 --> 00:21:41.530 align:middle line:90% So "I devoured the pizza." 00:21:41.530 --> 00:21:43.570 align:middle line:84% "Devour" can be followed by a noun phrase. 00:21:43.570 --> 00:21:46.550 align:middle line:84% But "I devoured that I have won the lottery" 00:21:46.550 --> 00:21:47.570 align:middle line:90% doesn't make sense. 00:21:47.570 --> 00:21:50.780 align:middle line:84% So this looks like a selection relation. 00:21:50.780 --> 00:21:53.260 align:middle line:84% And specifically, it's a selection relation 00:21:53.260 --> 00:21:56.980 align:middle line:84% that's picky about these words that introduce the clause that 00:21:56.980 --> 00:21:59.440 align:middle line:84% come at the beginning of the clause, words 00:21:59.440 --> 00:22:00.670 align:middle line:90% like "that" and "whether." 00:22:00.670 --> 00:22:04.240 align:middle line:84% So "think" can be followed by a clause where 00:22:04.240 --> 00:22:06.340 align:middle line:84% the word that introduces the clauses is "that." 00:22:06.340 --> 00:22:07.715 align:middle line:84% You can say things like, "I think 00:22:07.715 --> 00:22:09.070 align:middle line:90% that I have won the lottery." 00:22:09.070 --> 00:22:11.605 align:middle line:84% "Wonder" can be followed by a clause where 00:22:11.605 --> 00:22:13.480 align:middle line:84% the word that introduces the following clause 00:22:13.480 --> 00:22:15.310 align:middle line:84% is "whether," as in, "I wonder whether I 00:22:15.310 --> 00:22:16.480 align:middle line:90% have won the lottery." 00:22:16.480 --> 00:22:19.660 align:middle line:84% And these verbs can't be switched. 00:22:19.660 --> 00:22:21.920 align:middle line:84% So you can say "I think that I have won the lottery." 00:22:21.920 --> 00:22:24.620 align:middle line:84% You cannot say "I think whether I have won the lottery." 00:22:24.620 --> 00:22:26.870 align:middle line:84% You can say "I wonder whether I have won the lottery." 00:22:26.870 --> 00:22:29.380 align:middle line:84% You cannot say "I wonder that I have won the lottery." 00:22:29.380 --> 00:22:32.132 align:middle line:84% And I was saying this looks kind of like "depend" and "on". 00:22:32.132 --> 00:22:34.090 align:middle line:84% "Think" needs a clause that starts with "that." 00:22:34.090 --> 00:22:36.215 align:middle line:84% "Wonder" needs a clause that starts with "whether." 00:22:36.215 --> 00:22:38.650 align:middle line:84% "Depend" needs a prepositional phrase that has "on" in it. 00:22:38.650 --> 00:22:41.380 align:middle line:84% And just like with "depend" and "on," we said, yeah, "depend" 00:22:41.380 --> 00:22:45.250 align:middle line:84% is selecting for a prepositional phrase with a particular head. 00:22:45.250 --> 00:22:46.810 align:middle line:84% We get to say, oh, OK, apparently 00:22:46.810 --> 00:22:48.520 align:middle line:84% "that" and "whether" are the heads 00:22:48.520 --> 00:22:51.290 align:middle line:84% of the clause that's getting selected by the verb. 00:22:51.290 --> 00:22:53.350 align:middle line:84% So "think" and "wonder" are selecting 00:22:53.350 --> 00:22:55.390 align:middle line:84% for clauses that are headed by words 00:22:55.390 --> 00:22:56.500 align:middle line:90% like "that" and "whether." 00:22:56.500 --> 00:22:59.050 align:middle line:84% We have a word for words like "that" and "whether"-- 00:22:59.050 --> 00:23:03.010 align:middle line:90% we call them "complementizers." 00:23:03.010 --> 00:23:05.210 align:middle line:84% I think I apologized for that word last time. 00:23:05.210 --> 00:23:07.460 align:middle line:84% I'll just apologize for it again because it's probably 00:23:07.460 --> 00:23:09.310 align:middle line:84% not the word you learned for those things 00:23:09.310 --> 00:23:11.170 align:middle line:84% in elementary school or high school. 00:23:11.170 --> 00:23:14.335 align:middle line:84% You might have heard them called conjunctions, 00:23:14.335 --> 00:23:16.810 align:middle line:84% so "subordinating conjunctions," lots of things people 00:23:16.810 --> 00:23:18.100 align:middle line:90% call them. 00:23:18.100 --> 00:23:20.290 align:middle line:84% But in linguistics, we call them complementizers. 00:23:20.290 --> 00:23:23.590 align:middle line:84% So I'll have to ask you to join me in calling them that. 00:23:23.590 --> 00:23:26.800 align:middle line:84% So the abbreviation for complementizer is C. 00:23:26.800 --> 00:23:31.780 align:middle line:84% And the phrase is the sister of verbs like "think" and "wonder" 00:23:31.780 --> 00:23:34.240 align:middle line:84% is a complementizer phrase, a CP. 00:23:34.240 --> 00:23:36.550 align:middle line:84% It's a phrase that's headed by "that" or "whether," 00:23:36.550 --> 00:23:39.640 align:middle line:90% these complementizers. 00:23:39.640 --> 00:23:40.708 align:middle line:90% So I'll review. 00:23:40.708 --> 00:23:42.625 align:middle line:84% Hopefully it sounds at least vaguely familiar. 00:23:42.625 --> 00:23:46.360 align:middle line:90% 00:23:46.360 --> 00:23:48.738 align:middle line:84% So now here we are back at the original tree. 00:23:48.738 --> 00:23:50.530 align:middle line:84% "I will tickle the child with the feather." 00:23:50.530 --> 00:23:52.270 align:middle line:84% We've talked a lot about selection relations. 00:23:52.270 --> 00:23:54.312 align:middle line:84% We've said that selection relations are relations 00:23:54.312 --> 00:24:00.160 align:middle line:84% between heads and phrases, many cases, the sister of the head. 00:24:00.160 --> 00:24:02.770 align:middle line:84% And in particular, selection relations 00:24:02.770 --> 00:24:04.960 align:middle line:84% seem to be picky about when there 00:24:04.960 --> 00:24:08.890 align:middle line:84% is pickiness about the head of the phrase that's 00:24:08.890 --> 00:24:09.648 align:middle line:90% getting selected. 00:24:09.648 --> 00:24:12.190 align:middle line:84% So we've seen examples of that now with prepositional phrases 00:24:12.190 --> 00:24:17.980 align:middle line:84% and now with Complementizer Phrases, CPs, not clauses. 00:24:17.980 --> 00:24:22.370 align:middle line:84% But-- and this is near the end of what we did last time-- 00:24:22.370 --> 00:24:26.230 align:middle line:84% there is more to life than selection, 00:24:26.230 --> 00:24:27.850 align:middle line:84% not much more, but a little more. 00:24:27.850 --> 00:24:34.970 align:middle line:84% So we don't want to say that "tickle" or "child" is 00:24:34.970 --> 00:24:37.550 align:middle line:84% selecting "with a feather," right? 00:24:37.550 --> 00:24:42.890 align:middle line:84% So when I told you that tickle is selecting the child, 00:24:42.890 --> 00:24:45.260 align:middle line:84% one of the ways I tried to convince you 00:24:45.260 --> 00:24:47.840 align:middle line:84% of that was to say, yeah, not every verb 00:24:47.840 --> 00:24:49.460 align:middle line:90% can be followed by an object. 00:24:49.460 --> 00:24:51.570 align:middle line:84% There are transitive verbs and intransitive verbs. 00:24:51.570 --> 00:24:53.028 align:middle line:84% That's the kind of thing that makes 00:24:53.028 --> 00:24:55.010 align:middle line:84% us think we're looking at a selection relation. 00:24:55.010 --> 00:24:58.580 align:middle line:84% But any verb can be followed by "with the feather." 00:24:58.580 --> 00:25:00.067 align:middle line:84% You can do anything with a feather. 00:25:00.067 --> 00:25:01.650 align:middle line:84% You can tickle a child with a feather. 00:25:01.650 --> 00:25:03.020 align:middle line:84% Or you can devour a child with a feather. 00:25:03.020 --> 00:25:04.170 align:middle line:84% You can write a novel with a feather. 00:25:04.170 --> 00:25:05.545 align:middle line:84% Or you can thrive with a feather. 00:25:05.545 --> 00:25:08.090 align:middle line:84% Feathers, they're just very, very flexible tools. 00:25:08.090 --> 00:25:09.500 align:middle line:90% You can do anything with them. 00:25:09.500 --> 00:25:12.752 align:middle line:84% Again, some of these things make more sense than others. 00:25:12.752 --> 00:25:14.960 align:middle line:84% So it's not clear what you mean when you say that you 00:25:14.960 --> 00:25:17.180 align:middle line:90% will thrive with a feather. 00:25:17.180 --> 00:25:18.480 align:middle line:90% Or for that matter-- 00:25:18.480 --> 00:25:20.120 align:middle line:84% well, it's actually fairly clear what 00:25:20.120 --> 00:25:22.578 align:middle line:84% you mean when you say you'll devour a child with a feather. 00:25:22.578 --> 00:25:26.178 align:middle line:84% It just doesn't bear thinking about too closely. 00:25:26.178 --> 00:25:28.220 align:middle line:84% But basically you can do anything with a feather. 00:25:28.220 --> 00:25:31.100 align:middle line:84% There's nothing like verbs are transitive and intransitive. 00:25:31.100 --> 00:25:33.020 align:middle line:84% So this particular prepositional phrase, 00:25:33.020 --> 00:25:37.320 align:middle line:84% this sort of instrument, can go together with anything. 00:25:37.320 --> 00:25:41.810 align:middle line:84% And so we don't seem to want to say that there is a selection 00:25:41.810 --> 00:25:44.000 align:middle line:84% relation between this prepositional phrase 00:25:44.000 --> 00:25:45.080 align:middle line:90% and anything at all. 00:25:45.080 --> 00:25:48.030 align:middle line:90% 00:25:48.030 --> 00:25:50.210 align:middle line:84% So there's a distinction that we draw 00:25:50.210 --> 00:25:52.602 align:middle line:84% between what are called arguments 00:25:52.602 --> 00:25:53.810 align:middle line:90% and what are called adjuncts. 00:25:53.810 --> 00:25:57.440 align:middle line:84% So arguments, like "the child" in "I will tickle the child 00:25:57.440 --> 00:26:00.530 align:middle line:84% with a feather," are things that are selected by something, 00:26:00.530 --> 00:26:02.030 align:middle line:90% in this case, the verb. 00:26:02.030 --> 00:26:06.380 align:middle line:84% And the adjuncts are just phrases that wandered in. 00:26:06.380 --> 00:26:08.010 align:middle line:84% They're not selected by anything. 00:26:08.010 --> 00:26:10.585 align:middle line:84% They're just there because they want to be. 00:26:10.585 --> 00:26:11.960 align:middle line:84% They're not selected by anything. 00:26:11.960 --> 00:26:14.100 align:middle line:84% You get to put them in because you feel like it. 00:26:14.100 --> 00:26:16.820 align:middle line:84% They're often modifiers sort of like this one. 00:26:16.820 --> 00:26:19.812 align:middle line:90% 00:26:19.812 --> 00:26:20.770 align:middle line:90% So there are arguments. 00:26:20.770 --> 00:26:23.415 align:middle line:90% And then there are adjuncts. 00:26:23.415 --> 00:26:24.790 align:middle line:84% How do you tell whether something 00:26:24.790 --> 00:26:26.320 align:middle line:90% is an argument or an adjunct? 00:26:26.320 --> 00:26:29.320 align:middle line:84% I think I warned you last time this is something that people 00:26:29.320 --> 00:26:30.790 align:middle line:90% often get confused by. 00:26:30.790 --> 00:26:34.450 align:middle line:84% And so I want to be clear about it. 00:26:34.450 --> 00:26:38.260 align:middle line:84% Again, when we decided that "the child" is selected by "tickle," 00:26:38.260 --> 00:26:40.330 align:middle line:84% what we were doing was saying not 00:26:40.330 --> 00:26:43.430 align:middle line:84% every verb can be followed by an object. 00:26:43.430 --> 00:26:46.330 align:middle line:84% So if you're going to have a child right after the verb, 00:26:46.330 --> 00:26:47.727 align:middle line:84% you have to know what verb it is. 00:26:47.727 --> 00:26:49.810 align:middle line:84% There are transitive verbs and intransitive verbs. 00:26:49.810 --> 00:26:51.940 align:middle line:84% And some verbs are OK with the following object, 00:26:51.940 --> 00:26:54.580 align:middle line:90% and others or not. 00:26:54.580 --> 00:26:58.520 align:middle line:84% That's the signature of a selection relation. 00:26:58.520 --> 00:27:00.700 align:middle line:84% So the arguments are picky about which 00:27:00.700 --> 00:27:04.000 align:middle line:84% heads they can combine with, and adjuncts are not. 00:27:04.000 --> 00:27:05.830 align:middle line:84% What people sometimes get confused about 00:27:05.830 --> 00:27:09.318 align:middle line:84% is thinking that it's the opposite, that the pickiness 00:27:09.318 --> 00:27:10.610 align:middle line:90% goes in the opposite direction. 00:27:10.610 --> 00:27:14.440 align:middle line:84% So people think here's a verb that needs to have an object. 00:27:14.440 --> 00:27:15.940 align:middle line:84% And I've probably said things that 00:27:15.940 --> 00:27:17.290 align:middle line:90% led you down this garden path. 00:27:17.290 --> 00:27:20.110 align:middle line:84% So there are indeed verbs that are obligatorily transitive. 00:27:20.110 --> 00:27:22.210 align:middle line:84% I think the example I gave before was "devour." 00:27:22.210 --> 00:27:31.090 align:middle line:84% So "The dragon devoured pizza"--- 00:27:31.090 --> 00:27:34.270 align:middle line:84% because we've devoured enough children for one class, 00:27:34.270 --> 00:27:35.200 align:middle line:90% possibly too many-- 00:27:35.200 --> 00:27:36.580 align:middle line:90% "The dragon devoured the pizza." 00:27:36.580 --> 00:27:39.670 align:middle line:84% "Devour" actually has to combine with an object. 00:27:39.670 --> 00:27:42.460 align:middle line:84% Not only is this a selection relation 00:27:42.460 --> 00:27:44.440 align:middle line:84% because not just any verb can have an object, 00:27:44.440 --> 00:27:48.040 align:middle line:84% but actually, "devour" has to combine with an object. 00:27:48.040 --> 00:27:50.290 align:middle line:84% It's ungrammatical without the pizza. 00:27:50.290 --> 00:27:54.310 align:middle line:84% So if I take out the pizza, it becomes star, yeah? 00:27:54.310 --> 00:27:58.560 align:middle line:84% So "The dragon devoured," that's not a sentence. 00:27:58.560 --> 00:28:00.890 align:middle line:84% So that's an argument, which is actually obligatory. 00:28:00.890 --> 00:28:02.390 align:middle line:90% "Devour" has to be transitive. 00:28:02.390 --> 00:28:05.450 align:middle line:84% But there are plenty of verbs that-- in fact, it's 00:28:05.450 --> 00:28:09.380 align:middle line:84% more common for verbs to be optionally transitive. 00:28:09.380 --> 00:28:11.900 align:middle line:84% So "devour" needs to have an object. 00:28:11.900 --> 00:28:16.342 align:middle line:84% But "eat," which means almost the same thing as "devour," 00:28:16.342 --> 00:28:17.550 align:middle line:90% the transitivity is optional. 00:28:17.550 --> 00:28:18.758 align:middle line:90% You can say "I ate an apple." 00:28:18.758 --> 00:28:21.140 align:middle line:90% But you can also say "I ate." 00:28:21.140 --> 00:28:24.080 align:middle line:84% Obligatorily transitive verbs are actually kind of rare. 00:28:24.080 --> 00:28:26.660 align:middle line:84% And the best ones are kind of violent. 00:28:26.660 --> 00:28:29.600 align:middle line:84% They are things like "eviscerate" and "devour" 00:28:29.600 --> 00:28:30.290 align:middle line:90% and "mutilate." 00:28:30.290 --> 00:28:32.420 align:middle line:84% Those are the really clear transitive verbs. 00:28:32.420 --> 00:28:34.760 align:middle line:84% They're-- very handy in syntax classes. 00:28:34.760 --> 00:28:35.300 align:middle line:90% Yes? 00:28:35.300 --> 00:28:40.046 align:middle line:84% AUDIENCE: So are we suggesting the fact that since [INAUDIBLE] 00:28:40.046 --> 00:28:44.187 align:middle line:84% property of a feather, or is that any adjunct can modify? 00:28:44.187 --> 00:28:46.520 align:middle line:84% NORVIN RICHARDS: We're talking as though any adjunct can 00:28:46.520 --> 00:28:47.780 align:middle line:90% modify anything. 00:28:47.780 --> 00:28:52.950 align:middle line:84% And so this is another one of those places where 00:28:52.950 --> 00:28:57.690 align:middle line:84% we have to carefully distinguish between syntax and semantics. 00:28:57.690 --> 00:28:59.820 align:middle line:84% There are surely things that it's 00:28:59.820 --> 00:29:03.670 align:middle line:84% very difficult to imagine anybody doing with a feather, 00:29:03.670 --> 00:29:04.290 align:middle line:90% right? 00:29:04.290 --> 00:29:07.810 align:middle line:90% But those kinds of cases-- 00:29:07.810 --> 00:29:11.250 align:middle line:90% so what's an example? 00:29:11.250 --> 00:29:16.950 align:middle line:84% "She proved Fermat's last theorem with a feather," right? 00:29:16.950 --> 00:29:20.010 align:middle line:84% I don't know how she would do that, right? 00:29:20.010 --> 00:29:22.320 align:middle line:90% But that's a math problem. 00:29:22.320 --> 00:29:23.770 align:middle line:90% That's not a syntax problem. 00:29:23.770 --> 00:29:26.160 align:middle line:90% So the sentence is grammatical. 00:29:26.160 --> 00:29:28.020 align:middle line:90% We just can't figure out-- 00:29:28.020 --> 00:29:29.980 align:middle line:84% I wouldn't be able to draw a picture of it. 00:29:29.980 --> 00:29:30.940 align:middle line:90% Does that make sense? 00:29:30.940 --> 00:29:32.970 align:middle line:84% So, yeah, what we're saying is that that prepositional phrase 00:29:32.970 --> 00:29:34.920 align:middle line:84% can combine with anything in principle. 00:29:34.920 --> 00:29:37.500 align:middle line:84% Though there may be kinds of combinations 00:29:37.500 --> 00:29:40.480 align:middle line:84% that will give us meanings that are kind of hard to understand. 00:29:40.480 --> 00:29:41.250 align:middle line:90% Yeah, Joseph? 00:29:41.250 --> 00:29:43.660 align:middle line:84% AUDIENCE: [INAUDIBLE] stylistically common that 00:29:43.660 --> 00:29:45.070 align:middle line:90% violent words-- 00:29:45.070 --> 00:29:47.070 align:middle line:84% NORVIN RICHARDS: --are the most transitive ones? 00:29:47.070 --> 00:29:48.580 align:middle line:90% I think it is, yeah. 00:29:48.580 --> 00:29:51.000 align:middle line:84% Yeah, there's something about violence and transitivity, 00:29:51.000 --> 00:29:52.473 align:middle line:90% yeah. 00:29:52.473 --> 00:29:53.890 align:middle line:84% Something to think about as you're 00:29:53.890 --> 00:29:58.090 align:middle line:84% working on your languages, if you want to. 00:29:58.090 --> 00:30:00.880 align:middle line:84% So arguments are picky about what heads they combined with. 00:30:00.880 --> 00:30:03.628 align:middle line:84% But there are-- to look at things 00:30:03.628 --> 00:30:05.170 align:middle line:84% from the other perspective, if you're 00:30:05.170 --> 00:30:08.980 align:middle line:84% asking for a particular head, if you see a head 00:30:08.980 --> 00:30:11.140 align:middle line:84% and it's followed by a phrase, the question you 00:30:11.140 --> 00:30:13.720 align:middle line:84% ask yourself is not "Is that phrase optional?" 00:30:13.720 --> 00:30:15.770 align:middle line:90% People often ask that question. 00:30:15.770 --> 00:30:18.785 align:middle line:84% And it's a question that gets you in the wrong direction. 00:30:18.785 --> 00:30:21.160 align:middle line:84% There are plenty of examples of things that are, in fact, 00:30:21.160 --> 00:30:22.493 align:middle line:90% arguments, but they're optional. 00:30:22.493 --> 00:30:28.240 align:middle line:90% 00:30:28.240 --> 00:30:31.325 align:middle line:84% And then I think this is the last thing we did last time. 00:30:31.325 --> 00:30:33.700 align:middle line:84% I was trying to illustrate arguments and adjuncts to you. 00:30:33.700 --> 00:30:36.970 align:middle line:84% And we were in the middle of deciding on the boat. 00:30:36.970 --> 00:30:40.280 align:middle line:84% So let's decide on the boat again, fairly quickly. 00:30:40.280 --> 00:30:42.910 align:middle line:84% So what we decided was that the sentence 00:30:42.910 --> 00:30:46.460 align:middle line:84% "I decided on the boat" can mean at least two things. 00:30:46.460 --> 00:30:49.180 align:middle line:84% It can mean either I had some decision to make, 00:30:49.180 --> 00:30:52.420 align:middle line:84% and I made it while I was on the boat, right? 00:30:52.420 --> 00:30:54.562 align:middle line:84% Or it could mean I chose the boat, right? 00:30:54.562 --> 00:30:57.020 align:middle line:84% I was trying to choose between the boat and something else. 00:30:57.020 --> 00:30:59.860 align:middle line:90% And I chose the boat. 00:30:59.860 --> 00:31:04.960 align:middle line:84% And what we said was, if we ask, "The prepositional phrase 00:31:04.960 --> 00:31:08.080 align:middle line:84% 'on the boat,' is that an argument or an adjunct?" 00:31:08.080 --> 00:31:11.290 align:middle line:84% the answer to that question is yes. 00:31:11.290 --> 00:31:12.580 align:middle line:90% It can be an adjunct. 00:31:12.580 --> 00:31:14.950 align:middle line:84% "I made my decision" when it has the-- 00:31:14.950 --> 00:31:16.870 align:middle line:84% when the sentence means "I made my decision 00:31:16.870 --> 00:31:21.460 align:middle line:84% while I was on the boat," then "on the boat" is an adjunct. 00:31:21.460 --> 00:31:22.585 align:middle line:90% You can kind of see why. 00:31:22.585 --> 00:31:24.460 align:middle line:84% Just like you can do anything with a feather, 00:31:24.460 --> 00:31:26.410 align:middle line:84% you can do anything on a boat, right? 00:31:26.410 --> 00:31:31.030 align:middle line:84% So any phrase you can modify it with "on the boat." 00:31:31.030 --> 00:31:32.920 align:middle line:84% This particular prepositional phrase, 00:31:32.920 --> 00:31:37.730 align:middle line:84% then, is an adjunct in that reading, yeah? 00:31:37.730 --> 00:31:41.170 align:middle line:84% Have to be careful because, of course, if I say things 00:31:41.170 --> 00:31:51.730 align:middle line:84% like "It depends on the boat," or for that-- well, let's use 00:31:51.730 --> 00:31:53.322 align:middle line:84% that one, it depends on the boat. 00:31:53.322 --> 00:31:55.030 align:middle line:84% This is a case where there is a selection 00:31:55.030 --> 00:31:58.438 align:middle line:84% relation between "depend" and "on," right? 00:31:58.438 --> 00:31:59.980 align:middle line:84% We've been talking about that before, 00:31:59.980 --> 00:32:02.272 align:middle line:84% that "depend" select for prepositional phrase for which 00:32:02.272 --> 00:32:03.400 align:middle line:90% the head is "on," yeah? 00:32:03.400 --> 00:32:05.602 align:middle line:84% So you can't just look at a prepositional-- 00:32:05.602 --> 00:32:07.810 align:middle line:84% if someone asks you out of the blue, is "on the boat" 00:32:07.810 --> 00:32:09.313 align:middle line:90% an argument or an adjunct-- 00:32:09.313 --> 00:32:11.230 align:middle line:84% in a way, that's the point of this slide too-- 00:32:11.230 --> 00:32:13.980 align:middle line:84% the answer is, wait, what's the rest of the sentence, right? 00:32:13.980 --> 00:32:16.450 align:middle line:84% Because you have to find out what the relationship is 00:32:16.450 --> 00:32:18.658 align:middle line:84% between that prepositional phrase and the other words 00:32:18.658 --> 00:32:20.960 align:middle line:90% that are around it. 00:32:20.960 --> 00:32:23.770 align:middle line:84% So in the reading where "I decided on the boat" 00:32:23.770 --> 00:32:25.840 align:middle line:84% means "I made my decision while I 00:32:25.840 --> 00:32:28.000 align:middle line:84% was on the boat," "on the boat" is an adjunct. 00:32:28.000 --> 00:32:29.710 align:middle line:84% It's just describing the location 00:32:29.710 --> 00:32:32.770 align:middle line:84% where the rest of the sentence took place. 00:32:32.770 --> 00:32:36.760 align:middle line:84% And any sentence can take place on the boat, maybe. 00:32:36.760 --> 00:32:41.290 align:middle line:84% As opposed to the reading where it means "I chose the boat," 00:32:41.290 --> 00:32:45.240 align:middle line:84% there it's an argument because, again, 00:32:45.240 --> 00:32:49.120 align:middle line:84% just the fact that "decide on" means choose 00:32:49.120 --> 00:32:53.740 align:middle line:84% is kind of an idiosyncratic property of the verb "decide" 00:32:53.740 --> 00:32:56.200 align:middle line:90% and the preposition "on." 00:32:56.200 --> 00:32:59.880 align:middle line:84% It's like if you look up "decide" in the lexicon, 00:32:59.880 --> 00:33:01.540 align:middle line:84% it's going to have there, we're going 00:33:01.540 --> 00:33:04.780 align:middle line:84% to have to list the information, that it can optionally 00:33:04.780 --> 00:33:06.910 align:middle line:84% select an argument that starts with "on" 00:33:06.910 --> 00:33:10.180 align:middle line:84% and that the result of that combination means "choose." 00:33:10.180 --> 00:33:11.970 align:middle line:84% If you think about other things that 00:33:11.970 --> 00:33:13.720 align:middle line:84% mean something pretty similar to "decide," 00:33:13.720 --> 00:33:18.430 align:middle line:84% like "make up your mind," so "I made up my mind on the boat" 00:33:18.430 --> 00:33:21.610 align:middle line:84% only means the first thing, I think, right? 00:33:21.610 --> 00:33:24.108 align:middle line:84% "I made up my mind on the boat" means "I made my decision 00:33:24.108 --> 00:33:25.150 align:middle line:90% while I was on the boat." 00:33:25.150 --> 00:33:27.100 align:middle line:84% It doesn't mean "I chose the boat," right? 00:33:27.100 --> 00:33:29.450 align:middle line:84% I made a choice between the boat and something else. 00:33:29.450 --> 00:33:31.030 align:middle line:90% And I chose the boat, right? 00:33:31.030 --> 00:33:33.520 align:middle line:84% So it's not a property of expressions 00:33:33.520 --> 00:33:36.430 align:middle line:84% that mean "decide" that they always combined with "on" 00:33:36.430 --> 00:33:37.480 align:middle line:90% to mean this. 00:33:37.480 --> 00:33:40.300 align:middle line:84% It's a idiosyncratic property of "decide." 00:33:40.300 --> 00:33:43.267 align:middle line:84% It's kind of like we have to list for particular verbs 00:33:43.267 --> 00:33:45.100 align:middle line:84% that they're transitive or that they're not. 00:33:45.100 --> 00:33:46.642 align:middle line:84% We have to list for the word "decide" 00:33:46.642 --> 00:33:48.225 align:middle line:84% that it combines with "on" to give you 00:33:48.225 --> 00:33:49.780 align:middle line:84% a meaning with something like choose. 00:33:49.780 --> 00:33:50.280 align:middle line:90% I'm sorry. 00:33:50.280 --> 00:33:51.280 align:middle line:90% I'm talking over you. 00:33:51.280 --> 00:33:52.170 align:middle line:84% AUDIENCE: Is this the same effect 00:33:52.170 --> 00:33:53.060 align:middle line:90% as "look," where you "look up"? 00:33:53.060 --> 00:33:54.170 align:middle line:90% NORVIN RICHARDS: Yes. 00:33:54.170 --> 00:33:55.590 align:middle line:84% Yeah, that's another good example. 00:33:55.590 --> 00:33:59.270 align:middle line:84% So you can "look up a reference." 00:33:59.270 --> 00:34:01.880 align:middle line:90% So you can "look up." 00:34:01.880 --> 00:34:05.870 align:middle line:84% And that "up" is probably an adjunct. 00:34:05.870 --> 00:34:08.150 align:middle line:90% You can do anything up. 00:34:08.150 --> 00:34:10.205 align:middle line:84% But to "look up a reference," the fact 00:34:10.205 --> 00:34:14.000 align:middle line:84% that that means "look through some stuff and find it," 00:34:14.000 --> 00:34:18.418 align:middle line:84% that's an idiosyncratic fact about looking up. 00:34:18.418 --> 00:34:20.210 align:middle line:84% That's another good example of an argument. 00:34:20.210 --> 00:34:22.010 align:middle line:90% Yeah, nice example. 00:34:22.010 --> 00:34:23.340 align:middle line:90% Yeah? 00:34:23.340 --> 00:34:23.840 align:middle line:90% OK. 00:34:23.840 --> 00:34:26.590 align:middle line:90% 00:34:26.590 --> 00:34:29.560 align:middle line:84% And then-- I think this is where we stopped-- 00:34:29.560 --> 00:34:32.290 align:middle line:84% I asked you to think about sentences like "I decided 00:34:32.290 --> 00:34:34.620 align:middle line:90% on the boat on the plane." 00:34:34.620 --> 00:34:38.250 align:middle line:84% And I asked you to suppress your natural urge 00:34:38.250 --> 00:34:43.239 align:middle line:84% to make things more complicated and to think of alternatives. 00:34:43.239 --> 00:34:46.469 align:middle line:84% So consider situations in which there's 00:34:46.469 --> 00:34:48.900 align:middle line:84% only one boat and only one plane and neither of them 00:34:48.900 --> 00:34:51.810 align:middle line:90% is on the other. 00:34:51.810 --> 00:34:56.370 align:middle line:84% There are two things that you could imagine this meaning. 00:34:56.370 --> 00:35:01.140 align:middle line:84% It could mean I chose the boat while I was on the plane. 00:35:01.140 --> 00:35:03.060 align:middle line:84% Or it could mean, in principle, I 00:35:03.060 --> 00:35:07.280 align:middle line:84% chose the plane while I was on the boat, 00:35:07.280 --> 00:35:09.350 align:middle line:84% if we've decided that "decide" and "on" 00:35:09.350 --> 00:35:11.830 align:middle line:90% can combine in these ways. 00:35:11.830 --> 00:35:14.890 align:middle line:84% But I think the fact generally was that it only 00:35:14.890 --> 00:35:16.648 align:middle line:90% has one of those readings. 00:35:16.648 --> 00:35:18.190 align:middle line:84% Is that the reading that people have? 00:35:18.190 --> 00:35:19.750 align:middle line:90% I'll write the readings down. 00:35:19.750 --> 00:35:31.720 align:middle line:84% One is chose the boat while on the plane. 00:35:31.720 --> 00:35:38.980 align:middle line:84% And the other was chose the plane while I was on the boat. 00:35:38.980 --> 00:35:42.770 align:middle line:90% 00:35:42.770 --> 00:35:43.270 align:middle line:90% Yeah? 00:35:43.270 --> 00:35:46.340 align:middle line:90% 00:35:46.340 --> 00:35:50.600 align:middle line:84% How many of you can get it to mean this first thing? 00:35:50.600 --> 00:35:53.770 align:middle line:84% How many of you can get it to mean the second thing? 00:35:53.770 --> 00:35:55.807 align:middle line:90% A few of you. 00:35:55.807 --> 00:35:58.390 align:middle line:84% The few of you who can get it to mean the second thing, please 00:35:58.390 --> 00:36:00.060 align:middle line:84% meet with your classmates after class 00:36:00.060 --> 00:36:02.560 align:middle line:84% and get them to convince you, or maybe you can convince them 00:36:02.560 --> 00:36:04.040 align:middle line:90% that they're wrong. 00:36:04.040 --> 00:36:10.330 align:middle line:84% I think we talked about this a little bit in class, possibly, 00:36:10.330 --> 00:36:14.770 align:middle line:84% that, for me, at least-- and people should stop me if this 00:36:14.770 --> 00:36:16.840 align:middle line:90% is not true-- 00:36:16.840 --> 00:36:20.230 align:middle line:84% I can get it to mean the second thing, 00:36:20.230 --> 00:36:22.480 align:middle line:90% but I need an opera score. 00:36:22.480 --> 00:36:25.280 align:middle line:84% I have to say it in a particular way. 00:36:25.280 --> 00:36:29.530 align:middle line:84% I have to say something like "I decided, on the boat, 00:36:29.530 --> 00:36:30.730 align:middle line:90% on the plane." 00:36:30.730 --> 00:36:32.890 align:middle line:90% There have to be commas. 00:36:32.890 --> 00:36:36.070 align:middle line:84% And I have to kind of squash "on the boat" 00:36:36.070 --> 00:36:40.790 align:middle line:84% a little bit in order to get it to have this second meaning. 00:36:40.790 --> 00:36:42.320 align:middle line:84% Those of you who were considering 00:36:42.320 --> 00:36:45.110 align:middle line:84% raising your hands the second time, is that true of you? 00:36:45.110 --> 00:36:48.650 align:middle line:84% Or are you people who can just say, "I decided on the boat 00:36:48.650 --> 00:36:51.370 align:middle line:90% on the plane," and mean this? 00:36:51.370 --> 00:36:54.810 align:middle line:90% 00:36:54.810 --> 00:36:57.855 align:middle line:84% Some of you are raising thumbs and nodding 00:36:57.855 --> 00:36:58.730 align:middle line:90% and things like that. 00:36:58.730 --> 00:36:59.720 align:middle line:90% Is there anybody who-- 00:36:59.720 --> 00:37:01.720 align:middle line:84% that thing that I just said when I pronounced it 00:37:01.720 --> 00:37:03.387 align:middle line:84% in that particular way, is there anybody 00:37:03.387 --> 00:37:05.502 align:middle line:84% who did not think, oh, yeah, that's the way I 00:37:05.502 --> 00:37:07.960 align:middle line:84% was pronouncing it in my head when I said that I could have 00:37:07.960 --> 00:37:09.880 align:middle line:90% it be this way? 00:37:09.880 --> 00:37:10.730 align:middle line:90% Let me say it again. 00:37:10.730 --> 00:37:13.210 align:middle line:84% Is there anybody who, if you say, "I decided on the boat 00:37:13.210 --> 00:37:14.470 align:middle line:90% on the plane"-- 00:37:14.470 --> 00:37:15.610 align:middle line:90% saying it that way-- 00:37:15.610 --> 00:37:18.730 align:middle line:84% can it mean this, the second thing? 00:37:18.730 --> 00:37:20.650 align:middle line:90% Yes? 00:37:20.650 --> 00:37:23.970 align:middle line:84% AUDIENCE: I feel like that way, the same sentence is 00:37:23.970 --> 00:37:27.810 align:middle line:90% more neutral than suggesting. 00:37:27.810 --> 00:37:29.838 align:middle line:84% I still have to say it in a different way 00:37:29.838 --> 00:37:34.090 align:middle line:84% to get the second meaning, but when I say like that, 00:37:34.090 --> 00:37:36.207 align:middle line:84% I don't immediately get the part [INAUDIBLE].. 00:37:36.207 --> 00:37:37.040 align:middle line:90% NORVIN RICHARDS: Oh. 00:37:37.040 --> 00:37:37.873 align:middle line:90% Oh, I see. 00:37:37.873 --> 00:37:39.290 align:middle line:84% How do you have to say it in order 00:37:39.290 --> 00:37:40.462 align:middle line:90% to get the second reading? 00:37:40.462 --> 00:37:42.420 align:middle line:84% AUDIENCE: I absolutely didn't think about that. 00:37:42.420 --> 00:37:43.740 align:middle line:90% NORVIN RICHARDS: Oh, OK. 00:37:43.740 --> 00:37:44.310 align:middle line:90% Fair enough. 00:37:44.310 --> 00:37:45.600 align:middle line:90% Fair enough. 00:37:45.600 --> 00:37:48.662 align:middle line:84% This is one of many places where I 00:37:48.662 --> 00:37:51.120 align:middle line:84% think I'm going to be content with the fact that all of you 00:37:51.120 --> 00:37:53.220 align:middle line:84% raised your hands when I asked whether it could mean this 00:37:53.220 --> 00:37:54.960 align:middle line:84% and only a few of you raised your hands when I asked 00:37:54.960 --> 00:37:56.085 align:middle line:90% whether it could mean this. 00:37:56.085 --> 00:37:58.230 align:middle line:84% I think maybe the ones of you who 00:37:58.230 --> 00:38:01.050 align:middle line:84% can get it to mean this are doing it by pronouncing it 00:38:01.050 --> 00:38:03.840 align:middle line:84% in a especially interesting way, which is itself 00:38:03.840 --> 00:38:06.340 align:middle line:90% an interesting fact. 00:38:06.340 --> 00:38:08.610 align:middle line:84% There is-- I think I have alluded to this a couple 00:38:08.610 --> 00:38:09.720 align:middle line:90% of times-- 00:38:09.720 --> 00:38:11.790 align:middle line:84% a topic of study that people work on, 00:38:11.790 --> 00:38:14.160 align:middle line:84% trying to figure out what the rules are for how 00:38:14.160 --> 00:38:16.380 align:middle line:84% the pitch of your voice rises and falls as you speak 00:38:16.380 --> 00:38:18.017 align:middle line:84% and where you put in pauses and so on. 00:38:18.017 --> 00:38:20.100 align:middle line:84% It's the study of what's sometimes called prosody. 00:38:20.100 --> 00:38:24.610 align:middle line:84% It's very complicated and hard, but it's really interesting. 00:38:24.610 --> 00:38:28.173 align:middle line:84% So maybe the short thing for me to say is, on the prosody 00:38:28.173 --> 00:38:30.090 align:middle line:84% where there's nothing special, where we're not 00:38:30.090 --> 00:38:33.210 align:middle line:84% doing any pausing or downgrading of anything, where we're just 00:38:33.210 --> 00:38:34.950 align:middle line:84% reading this straight through, I think 00:38:34.950 --> 00:38:38.110 align:middle line:90% everybody prefers this to this. 00:38:38.110 --> 00:38:42.440 align:middle line:84% Let me say that for the time being. 00:38:42.440 --> 00:38:45.520 align:middle line:84% So "I decided on the boat on the plane." 00:38:45.520 --> 00:38:48.820 align:middle line:84% When you've got these two "on" phrases after "decide," 00:38:48.820 --> 00:38:52.000 align:middle line:84% the sort of tendency anyway, unless you 00:38:52.000 --> 00:38:54.107 align:middle line:84% do fancy prosodic things, is to have 00:38:54.107 --> 00:38:55.690 align:middle line:84% the first one be the argument and then 00:38:55.690 --> 00:38:56.940 align:middle line:90% the second one be the adjunct. 00:38:56.940 --> 00:39:00.240 align:middle line:90% 00:39:00.240 --> 00:39:03.300 align:middle line:84% So what we're finding out, then, is that if you've got a head 00:39:03.300 --> 00:39:05.510 align:middle line:84% and it has both an argument and an adjunct, 00:39:05.510 --> 00:39:08.935 align:middle line:84% the argument is closer to the head. 00:39:08.935 --> 00:39:10.310 align:middle line:84% We're going to come back to that. 00:39:10.310 --> 00:39:13.790 align:middle line:84% But a way to think about it-- and we'll come back to it 00:39:13.790 --> 00:39:15.530 align:middle line:84% and say this more formally in a second-- 00:39:15.530 --> 00:39:16.760 align:middle line:90% it's as though-- 00:39:16.760 --> 00:39:19.760 align:middle line:84% I think I have a tree that I can make use of here. 00:39:19.760 --> 00:39:21.650 align:middle line:84% It's as though, if you're going to have 00:39:21.650 --> 00:39:23.900 align:middle line:84% a verb like "decide" that's going 00:39:23.900 --> 00:39:27.770 align:middle line:84% to have two prepositional phrases after it-- 00:39:27.770 --> 00:39:29.840 align:middle line:84% and forget what these are for now, 00:39:29.840 --> 00:39:34.370 align:middle line:84% and also we'll have "on the boat" and "on the plane." 00:39:34.370 --> 00:39:36.950 align:middle line:90% 00:39:36.950 --> 00:39:38.030 align:middle line:90% So you've got "decide." 00:39:38.030 --> 00:39:40.447 align:middle line:84% And it's going to have two prepositional phrases after it. 00:39:40.447 --> 00:39:43.190 align:middle line:84% And "decide" optionally selects for a prepositional phrase 00:39:43.190 --> 00:39:45.560 align:middle line:90% with "on" as the head. 00:39:45.560 --> 00:39:48.200 align:middle line:84% And what we're seeing is it's as though-- if you 00:39:48.200 --> 00:39:49.745 align:middle line:84% pick the version of "decide" that's 00:39:49.745 --> 00:39:53.900 align:middle line:84% like-- if you decide to have "decide" select for something, 00:39:53.900 --> 00:39:57.440 align:middle line:84% it's as though, when you're merging these two 00:39:57.440 --> 00:40:00.830 align:middle line:84% prepositional phrases to projections of the verb, 00:40:00.830 --> 00:40:03.980 align:middle line:84% you need to merge the one that's selected first. 00:40:03.980 --> 00:40:07.460 align:middle line:84% There's some kind of urgency about this selection relation. 00:40:07.460 --> 00:40:10.340 align:middle line:84% You don't just get to freely decide 00:40:10.340 --> 00:40:12.650 align:middle line:84% what order to merge these two prepositional phrases in, 00:40:12.650 --> 00:40:14.180 align:middle line:90% the argument and the adjunct. 00:40:14.180 --> 00:40:16.820 align:middle line:84% You need to hurry up and give it its selection requirements. 00:40:16.820 --> 00:40:18.140 align:middle line:90% Something like that. 00:40:18.140 --> 00:40:20.060 align:middle line:84% And we'll come back and talk more about that. 00:40:20.060 --> 00:40:24.290 align:middle line:90% 00:40:24.290 --> 00:40:27.380 align:middle line:90% Incidentally-- well, who cares? 00:40:27.380 --> 00:40:32.630 align:middle line:84% Incidentally, that way of talking about things-- 00:40:32.630 --> 00:40:33.770 align:middle line:90% no, sorry. 00:40:33.770 --> 00:40:35.740 align:middle line:84% Let me wait until we get more formal about it 00:40:35.740 --> 00:40:36.737 align:middle line:90% to make the next point. 00:40:36.737 --> 00:40:38.570 align:middle line:84% The next point is that this may help us some 00:40:38.570 --> 00:40:40.460 align:middle line:84% with the kinds of cases that got me to write 00:40:40.460 --> 00:40:42.470 align:middle line:84% that tree in the first place, the kinds of cases 00:40:42.470 --> 00:40:45.120 align:middle line:84% where a head seems to select for more than one thing. 00:40:45.120 --> 00:40:48.630 align:middle line:90% So we'll come back to that. 00:40:48.630 --> 00:40:52.010 align:middle line:84% So that's where we were last time, I think. 00:40:52.010 --> 00:40:54.500 align:middle line:84% [LAUGHS] Does anybody have any questions about any of that? 00:40:54.500 --> 00:40:55.490 align:middle line:84% This has been an attempt to review 00:40:55.490 --> 00:40:57.948 align:middle line:84% because I know you've had a week during which you might not 00:40:57.948 --> 00:41:00.680 align:middle line:84% have spent every waking moment thinking about syntax. 00:41:00.680 --> 00:41:03.770 align:middle line:84% I don't know why, but I guess young people today, 00:41:03.770 --> 00:41:06.290 align:middle line:90% they have their priorities. 00:41:06.290 --> 00:41:07.600 align:middle line:90% Are there questions about this? 00:41:07.600 --> 00:41:10.370 align:middle line:84% Are there things that aren't clear? 00:41:10.370 --> 00:41:11.620 align:middle line:90% OK. 00:41:11.620 --> 00:41:14.237 align:middle line:90% Yeah? 00:41:14.237 --> 00:41:16.820 align:middle line:84% AUDIENCE: I guess it seems to me that "decide" and "decide on" 00:41:16.820 --> 00:41:18.320 align:middle line:84% are almost like two different verbs. 00:41:18.320 --> 00:41:19.237 align:middle line:90% NORVIN RICHARDS: Yeah. 00:41:19.237 --> 00:41:21.752 align:middle line:84% AUDIENCE: So why couldn't you merge "decide" with "on" 00:41:21.752 --> 00:41:24.690 align:middle line:84% and have that be a verb phrase and the argument of that 00:41:24.690 --> 00:41:25.763 align:middle line:90% is [INAUDIBLE]? 00:41:25.763 --> 00:41:26.680 align:middle line:90% NORVIN RICHARDS: Yeah. 00:41:26.680 --> 00:41:27.280 align:middle line:90% Yeah. 00:41:27.280 --> 00:41:27.850 align:middle line:90% Yeah. 00:41:27.850 --> 00:41:32.080 align:middle line:84% So you're wondering for "decide on the boat," 00:41:32.080 --> 00:41:38.830 align:middle line:84% why not do "decide on" and then combine 00:41:38.830 --> 00:41:43.847 align:middle line:84% that whole thing with a noun phrase, "the boat," 00:41:43.847 --> 00:41:44.680 align:middle line:90% something like that? 00:41:44.680 --> 00:41:45.263 align:middle line:90% AUDIENCE: Yes. 00:41:45.263 --> 00:41:46.810 align:middle line:90% NORVIN RICHARDS: Yeah. 00:41:46.810 --> 00:41:49.210 align:middle line:84% Notice that these are making different 00:41:49.210 --> 00:41:55.300 align:middle line:84% claims about whether "on the boat" is a constituent or not. 00:41:55.300 --> 00:41:57.760 align:middle line:84% And so, for example, one of the constituency 00:41:57.760 --> 00:42:02.560 align:middle line:84% tests that we fooled around with was the one where 00:42:02.560 --> 00:42:06.460 align:middle line:84% I'm amazed by something that you've said, 00:42:06.460 --> 00:42:09.140 align:middle line:84% and I repeat part of what you've said in amazement. 00:42:09.140 --> 00:42:12.880 align:middle line:84% So if you say, "I decided on the boat," one thing I can surely 00:42:12.880 --> 00:42:15.860 align:middle line:90% say is, "The boat?" 00:42:15.860 --> 00:42:17.630 align:middle line:84% But I think I can also say, "On the boat?" 00:42:17.630 --> 00:42:20.333 align:middle line:90% 00:42:20.333 --> 00:42:22.250 align:middle line:84% AUDIENCE: Does that switch "up the reference"? 00:42:22.250 --> 00:42:24.890 align:middle line:84% NORVIN RICHARDS: So I don't think so, right? 00:42:24.890 --> 00:42:27.420 align:middle line:84% "Up the reference"-- that's a good example. 00:42:27.420 --> 00:42:32.690 align:middle line:90% "Look up the reference." 00:42:32.690 --> 00:42:36.230 align:middle line:84% If I'm amazed when you say, "I looked up the reference," 00:42:36.230 --> 00:42:38.420 align:middle line:90% if I say, "Up the reference?" 00:42:38.420 --> 00:42:40.670 align:middle line:90% that's weird. 00:42:40.670 --> 00:42:43.410 align:middle line:84% I have a lifetime's experience being weird. 00:42:43.410 --> 00:42:47.630 align:middle line:84% And I think that's worse than "I decided 00:42:47.630 --> 00:42:50.090 align:middle line:90% on the boat," "On the boat?" 00:42:50.090 --> 00:42:50.840 align:middle line:90% I think. 00:42:50.840 --> 00:42:52.585 align:middle line:90% But, yeah, that's the claim. 00:42:52.585 --> 00:42:53.960 align:middle line:84% Notice there's another difference 00:42:53.960 --> 00:42:56.335 align:middle line:84% between "look up the reference" and "decide on the boat." 00:42:56.335 --> 00:42:59.900 align:middle line:84% You can also say, "look the reference up." 00:42:59.900 --> 00:43:03.050 align:middle line:84% You cannot say "decide the boat on." 00:43:03.050 --> 00:43:06.380 align:middle line:84% So we probably don't want these to be the same. 00:43:06.380 --> 00:43:08.990 align:middle line:84% It's another reason to think that they're different. 00:43:08.990 --> 00:43:11.600 align:middle line:84% If you're familiar with German, German 00:43:11.600 --> 00:43:13.400 align:middle line:84% has a zillion particles like this, 00:43:13.400 --> 00:43:15.650 align:middle line:84% where you're putting things between the verb 00:43:15.650 --> 00:43:19.010 align:middle line:84% and the separable prefix, they're called. 00:43:19.010 --> 00:43:19.510 align:middle line:90% Yeah? 00:43:19.510 --> 00:43:23.112 align:middle line:90% 00:43:23.112 --> 00:43:25.930 align:middle line:90% OK. 00:43:25.930 --> 00:43:28.660 align:middle line:84% So now, hopefully I've got you willing to entertain, at least 00:43:28.660 --> 00:43:30.310 align:middle line:84% temporarily, the possibility that there 00:43:30.310 --> 00:43:33.790 align:middle line:84% are things that we can describe as arguments and adjuncts. 00:43:33.790 --> 00:43:36.550 align:middle line:84% Arguments are the things that are selected by other heads. 00:43:36.550 --> 00:43:40.330 align:middle line:84% Adjuncts are phrases that just wandered in by mistake. 00:43:40.330 --> 00:43:41.470 align:middle line:90% Nothing is selecting them. 00:43:41.470 --> 00:43:43.608 align:middle line:84% They're just here because they want to be. 00:43:43.608 --> 00:43:45.400 align:middle line:84% They're not here because they're satisfying 00:43:45.400 --> 00:43:47.072 align:middle line:90% any needs of anything at all. 00:43:47.072 --> 00:43:49.030 align:middle line:84% They're parasites, basically, is what they are. 00:43:49.030 --> 00:43:53.452 align:middle line:84% They end up in here inside the clause. 00:43:53.452 --> 00:43:54.910 align:middle line:84% And I've shown you one kind of test 00:43:54.910 --> 00:43:59.260 align:middle line:84% for them, which is that if you have a head that is combining 00:43:59.260 --> 00:44:02.140 align:middle line:84% both with an argument and with an adjunct, 00:44:02.140 --> 00:44:05.170 align:middle line:84% it looks like there's at least a tendency for the argument 00:44:05.170 --> 00:44:07.880 align:middle line:84% to want to be closer to the head than the adjunct. 00:44:07.880 --> 00:44:10.060 align:middle line:84% So "decide on the boat on the plane," 00:44:10.060 --> 00:44:12.760 align:middle line:84% the argument-y reading, the choose reading, 00:44:12.760 --> 00:44:14.110 align:middle line:90% goes with the first one. 00:44:14.110 --> 00:44:17.170 align:middle line:84% That means chose the boat on the plane. 00:44:17.170 --> 00:44:22.450 align:middle line:84% I want to show you another test for arguments versus adjuncts. 00:44:22.450 --> 00:44:23.560 align:middle line:90% Here's another sentence. 00:44:23.560 --> 00:44:28.010 align:middle line:84% "I decided on the boat, and Mary decided on the plane." 00:44:28.010 --> 00:44:34.650 align:middle line:84% That can mean at least two things, I think-- 00:44:34.650 --> 00:44:37.260 align:middle line:90% possibly four, but at least two. 00:44:37.260 --> 00:44:39.930 align:middle line:84% One thing that it can mean is I chose the boat, 00:44:39.930 --> 00:44:41.980 align:middle line:90% and Mary chose the plane. 00:44:41.980 --> 00:44:44.150 align:middle line:84% Another thing it can mean is Mary 00:44:44.150 --> 00:44:46.170 align:middle line:84% and I both had decisions to make. 00:44:46.170 --> 00:44:49.302 align:middle line:84% We had to decide whether to major in linguistics or not. 00:44:49.302 --> 00:44:51.260 align:middle line:84% And I made my decision while I was in the boat, 00:44:51.260 --> 00:44:54.200 align:middle line:84% and she made hers while she was on the plane. 00:44:54.200 --> 00:44:57.320 align:middle line:90% So it could mean that, too. 00:44:57.320 --> 00:45:00.650 align:middle line:84% I guess it could also mean I chose the boat, 00:45:00.650 --> 00:45:02.690 align:middle line:84% and Mary decided to become a linguistics major 00:45:02.690 --> 00:45:05.150 align:middle line:90% while she was on the plane? 00:45:05.150 --> 00:45:06.525 align:middle line:90% Maybe? 00:45:06.525 --> 00:45:07.650 align:middle line:90% I'm not so sure about that. 00:45:07.650 --> 00:45:09.733 align:middle line:84% But anyway, it can mean at least those two things. 00:45:09.733 --> 00:45:13.790 align:middle line:90% 00:45:13.790 --> 00:45:17.320 align:middle line:90% Now consider this sentence. 00:45:17.320 --> 00:45:20.185 align:middle line:84% "I decided on the boat, and Mary did so on the plane." 00:45:20.185 --> 00:45:24.520 align:middle line:90% 00:45:24.520 --> 00:45:27.900 align:middle line:90% Does that mean two things? 00:45:27.900 --> 00:45:29.330 align:middle line:90% I don't think so. 00:45:29.330 --> 00:45:30.440 align:middle line:90% What does that one mean? 00:45:30.440 --> 00:45:33.180 align:middle line:90% 00:45:33.180 --> 00:45:33.780 align:middle line:90% Joseph? 00:45:33.780 --> 00:45:34.988 align:middle line:90% AUDIENCE: That's the example. 00:45:34.988 --> 00:45:37.370 align:middle line:90% 00:45:37.370 --> 00:45:39.310 align:middle line:84% I decided to become a linguistics major 00:45:39.310 --> 00:45:40.523 align:middle line:90% while I was on the boat. 00:45:40.523 --> 00:45:42.730 align:middle line:84% And Mary chose not to, but she [INAUDIBLE].. 00:45:42.730 --> 00:45:43.540 align:middle line:90% NORVIN RICHARDS: Yeah, exactly. 00:45:43.540 --> 00:45:45.248 align:middle line:84% So I made the right decision on the boat, 00:45:45.248 --> 00:45:47.170 align:middle line:84% and Mary made the wrong decision on the plane. 00:45:47.170 --> 00:45:50.440 align:middle line:90% Yeah, it can mean that. 00:45:50.440 --> 00:45:50.980 align:middle line:90% Yes. 00:45:50.980 --> 00:45:53.320 align:middle line:84% So there's a phenomenon called VP-pronominalization. 00:45:53.320 --> 00:45:59.000 align:middle line:84% That is you're taking the VP "decide," 00:45:59.000 --> 00:46:02.350 align:middle line:84% and you're replacing it with "do so," 00:46:02.350 --> 00:46:04.750 align:middle line:90% which is kind of like a pronoun. 00:46:04.750 --> 00:46:07.060 align:middle line:84% It's like a pronoun in that it's an expression that 00:46:07.060 --> 00:46:09.123 align:middle line:84% can stand in for lots of different kinds 00:46:09.123 --> 00:46:09.790 align:middle line:90% of verb phrases. 00:46:09.790 --> 00:46:13.420 align:middle line:84% So just like a pronoun like "she" can stand in 00:46:13.420 --> 00:46:16.720 align:middle line:84% for any female person, "do so" can 00:46:16.720 --> 00:46:20.980 align:middle line:84% stand in for lots of kinds of verb phrases. 00:46:20.980 --> 00:46:24.910 align:middle line:84% And what we're learning is that if you do that, 00:46:24.910 --> 00:46:29.217 align:middle line:84% the phrase that's outside of "do so"-- "on the plane"-- 00:46:29.217 --> 00:46:32.830 align:middle line:90% 00:46:32.830 --> 00:46:34.120 align:middle line:90% can only be an adjunct. 00:46:34.120 --> 00:46:36.562 align:middle line:90% It can't be an argument. 00:46:36.562 --> 00:46:38.770 align:middle line:84% And there are various ways we could think about this, 00:46:38.770 --> 00:46:40.270 align:middle line:90% but here's one. 00:46:40.270 --> 00:46:43.495 align:middle line:84% We could say, look, this is the argument. 00:46:43.495 --> 00:46:46.790 align:middle line:90% 00:46:46.790 --> 00:46:49.360 align:middle line:90% This is the adjunct. 00:46:49.360 --> 00:46:52.390 align:middle line:84% We said there's a special relation of selection 00:46:52.390 --> 00:46:55.213 align:middle line:84% between "decide" and the argument that 00:46:55.213 --> 00:46:57.130 align:middle line:84% doesn't hold between "decide" and the adjunct. 00:46:57.130 --> 00:46:59.150 align:middle line:84% The adjunct is just here because it wants to be. 00:46:59.150 --> 00:47:01.370 align:middle line:84% So there's a selection relation between these two things. 00:47:01.370 --> 00:47:02.800 align:middle line:84% And then the adjunct is just something 00:47:02.800 --> 00:47:05.470 align:middle line:84% you merge because, well, you've got it, and you can merge it. 00:47:05.470 --> 00:47:08.690 align:middle line:90% 00:47:08.690 --> 00:47:09.940 align:middle line:90% That's what we've been saying. 00:47:09.940 --> 00:47:11.440 align:middle line:84% And maybe what we're learning here 00:47:11.440 --> 00:47:15.400 align:middle line:84% is that if you look up "decide" in the lexicon, 00:47:15.400 --> 00:47:17.710 align:middle line:84% you're going to see that it has the option of selecting 00:47:17.710 --> 00:47:20.480 align:middle line:84% for a prepositional phrase headed by "on." 00:47:20.480 --> 00:47:22.930 align:middle line:84% But if you look up "do so" in the lexicon, 00:47:22.930 --> 00:47:25.270 align:middle line:84% well, it doesn't select anything. 00:47:25.270 --> 00:47:28.450 align:middle line:84% There aren't any selectional relations with "do so." 00:47:28.450 --> 00:47:31.240 align:middle line:90% So "do so" is kind of blank. 00:47:31.240 --> 00:47:32.770 align:middle line:84% It just means there's a verb phrase, 00:47:32.770 --> 00:47:36.520 align:middle line:84% and look around at the rest of the sentence to figure out 00:47:36.520 --> 00:47:38.380 align:middle line:90% what the verb phrase is. 00:47:38.380 --> 00:47:40.550 align:middle line:90% But there's a verb phrase. 00:47:40.550 --> 00:47:42.350 align:middle line:90% So it's not doing any selecting. 00:47:42.350 --> 00:47:45.670 align:middle line:84% So if you see a prepositional phrase after "do so," 00:47:45.670 --> 00:47:48.220 align:middle line:84% it's not an argument of anything. 00:47:48.220 --> 00:47:50.110 align:middle line:90% It had better be an adjunct. 00:47:50.110 --> 00:47:52.270 align:middle line:84% I guess that's what we're learning here. 00:47:52.270 --> 00:47:53.770 align:middle line:84% As I told you, this is another test 00:47:53.770 --> 00:47:54.910 align:middle line:90% for arguments and adjuncts. 00:47:54.910 --> 00:47:58.270 align:middle line:84% If you do VP-pronominalization-- if you replace the verb phrase 00:47:58.270 --> 00:48:01.300 align:middle line:84% with "do so" and you have a phrase left over, 00:48:01.300 --> 00:48:03.880 align:middle line:84% then that thing had better not be an argument. 00:48:03.880 --> 00:48:06.910 align:middle line:90% It had better be an adjunct. 00:48:06.910 --> 00:48:09.220 align:middle line:84% The action with arguments and adjuncts 00:48:09.220 --> 00:48:12.340 align:middle line:84% so far has been with prepositional phrases. 00:48:12.340 --> 00:48:14.770 align:middle line:84% We've talked as though direct objects, for example, 00:48:14.770 --> 00:48:15.972 align:middle line:90% are always arguments. 00:48:15.972 --> 00:48:18.430 align:middle line:84% And so what we expect is that a direct object will never be 00:48:18.430 --> 00:48:19.722 align:middle line:90% able to be left over like this. 00:48:19.722 --> 00:48:26.170 align:middle line:84% That is, you won't be able to say, "I chose the boat, 00:48:26.170 --> 00:48:35.000 align:middle line:84% and Mary did so the plane," because "do so" 00:48:35.000 --> 00:48:36.230 align:middle line:90% doesn't select for anything. 00:48:36.230 --> 00:48:37.880 align:middle line:84% Direct objects are always selected for. 00:48:37.880 --> 00:48:40.190 align:middle line:84% You get them just if the verb is a transitive verb. 00:48:40.190 --> 00:48:41.690 align:middle line:84% And "do so" isn't a transitive verb. 00:48:41.690 --> 00:48:44.420 align:middle line:90% It's almost not a verb at all. 00:48:44.420 --> 00:48:46.250 align:middle line:84% Does that accord with people's intuitions, 00:48:46.250 --> 00:48:47.450 align:middle line:90% that you can't say that? 00:48:47.450 --> 00:48:48.050 align:middle line:90% Yeah. 00:48:48.050 --> 00:48:48.550 align:middle line:90% OK. 00:48:48.550 --> 00:48:50.570 align:middle line:84% In fact, I'm going to star it before anybody 00:48:50.570 --> 00:48:52.880 align:middle line:90% gets the wrong idea. 00:48:52.880 --> 00:48:53.540 align:middle line:90% Bad sentence. 00:48:53.540 --> 00:48:54.230 align:middle line:90% Bad sentence. 00:48:54.230 --> 00:48:55.660 align:middle line:90% No biscuit! 00:48:55.660 --> 00:48:56.658 align:middle line:90% Yeah? 00:48:56.658 --> 00:48:59.530 align:middle line:90% OK. 00:48:59.530 --> 00:49:02.445 align:middle line:90% All right, so-- yeah? 00:49:02.445 --> 00:49:03.820 align:middle line:84% AUDIENCE: You could say, "I chose 00:49:03.820 --> 00:49:05.470 align:middle line:90% the boat and Mary the plane." 00:49:05.470 --> 00:49:09.025 align:middle line:84% NORVIN RICHARDS: Yes, good example. 00:49:09.025 --> 00:49:13.759 align:middle line:90% 00:49:13.759 --> 00:49:17.373 align:middle line:90% AUDIENCE: [INAUDIBLE] 00:49:17.373 --> 00:49:18.290 align:middle line:90% NORVIN RICHARDS: Yeah. 00:49:18.290 --> 00:49:20.780 align:middle line:84% So, "I chose a boat and Mary a plane," 00:49:20.780 --> 00:49:28.350 align:middle line:84% or "I decided on a boat and Mary on a plane." 00:49:28.350 --> 00:49:29.060 align:middle line:90% Can you say that? 00:49:29.060 --> 00:49:33.954 align:middle line:90% 00:49:33.954 --> 00:49:37.373 align:middle line:84% "I live in Somerville and Mary in Allston." 00:49:37.373 --> 00:49:38.790 align:middle line:84% You can have prepositional phrases 00:49:38.790 --> 00:49:42.040 align:middle line:84% in this construction, but maybe not selected ones. 00:49:42.040 --> 00:49:42.540 align:middle line:90% Ugh. 00:49:42.540 --> 00:49:47.242 align:middle line:84% Because "I decided on the boat and Mary on the plane," 00:49:47.242 --> 00:49:48.700 align:middle line:84% that feels kind of adjunct-y to me. 00:49:48.700 --> 00:49:50.786 align:middle line:90% I don't know about guys. 00:49:50.786 --> 00:49:52.970 align:middle line:84% AUDIENCE: It works, but it's still vague. 00:49:52.970 --> 00:49:55.033 align:middle line:84% NORVIN RICHARDS: Works but it's still vague? 00:49:55.033 --> 00:49:55.950 align:middle line:90% Is that what you said? 00:49:55.950 --> 00:49:57.180 align:middle line:90% AUDIENCE: [INAUDIBLE] 00:49:57.180 --> 00:49:58.170 align:middle line:84% NORVIN RICHARDS: It could mean either one? 00:49:58.170 --> 00:49:58.530 align:middle line:90% AUDIENCE: Either one. 00:49:58.530 --> 00:49:59.400 align:middle line:90% NORVIN RICHARDS: OK, good. 00:49:59.400 --> 00:50:00.870 align:middle line:84% I was sort of hoping that was true 00:50:00.870 --> 00:50:02.828 align:middle line:84% because I don't know any reason why it wouldn't 00:50:02.828 --> 00:50:04.290 align:middle line:90% be able to mean either one. 00:50:04.290 --> 00:50:06.133 align:middle line:84% This is a really interesting phenomenon. 00:50:06.133 --> 00:50:07.300 align:middle line:90% There's a lot of work on it. 00:50:07.300 --> 00:50:10.140 align:middle line:90% It's called gapping. 00:50:10.140 --> 00:50:14.550 align:middle line:84% And dissertations have been written. 00:50:14.550 --> 00:50:16.830 align:middle line:90% [LAUGHS] It's interesting stuff. 00:50:16.830 --> 00:50:17.648 align:middle line:90% Yeah? 00:50:17.648 --> 00:50:19.440 align:middle line:84% AUDIENCE: When you first read the sentence, 00:50:19.440 --> 00:50:23.250 align:middle line:84% I actually interpreted it as "I chose the boat and Mary did so 00:50:23.250 --> 00:50:24.110 align:middle line:90% on the plane." 00:50:24.110 --> 00:50:25.110 align:middle line:90% NORVIN RICHARDS: Oh, no. 00:50:25.110 --> 00:50:26.385 align:middle line:90% What did Mary do on the plane? 00:50:26.385 --> 00:50:27.540 align:middle line:90% AUDIENCE: She chose the boat. 00:50:27.540 --> 00:50:29.010 align:middle line:84% NORVIN RICHARDS: Oh, she chose the boat on the plane. 00:50:29.010 --> 00:50:29.970 align:middle line:90% Oh, yes. 00:50:29.970 --> 00:50:31.110 align:middle line:90% Well, that would be OK. 00:50:31.110 --> 00:50:32.475 align:middle line:90% [LAUGHS] Phew. 00:50:32.475 --> 00:50:34.350 align:middle line:84% I thought you meant that she chose the plane. 00:50:34.350 --> 00:50:36.930 align:middle line:84% Yes, no, that's-- no, that's absolutely right. 00:50:36.930 --> 00:50:38.733 align:middle line:90% OK, phew. 00:50:38.733 --> 00:50:39.900 align:middle line:90% Yeah, that's a nice example. 00:50:39.900 --> 00:50:43.950 align:middle line:90% "I decided on the boat." 00:50:43.950 --> 00:50:45.360 align:middle line:90% Oh, here-- we can use this tree. 00:50:45.360 --> 00:50:48.465 align:middle line:84% "I decided on the boat and Mary did so on the plane." 00:50:48.465 --> 00:50:51.702 align:middle line:90% 00:50:51.702 --> 00:50:58.020 align:middle line:84% I was taking that to mean "decide" in the first clause 00:50:58.020 --> 00:51:00.360 align:middle line:84% is taking an adjunct, "on the boat." 00:51:00.360 --> 00:51:02.610 align:middle line:84% And in the second clause, well, "do so" 00:51:02.610 --> 00:51:04.740 align:middle line:90% is just replacing "decide." 00:51:04.740 --> 00:51:06.690 align:middle line:84% And "on the plane" is another adjunct. 00:51:06.690 --> 00:51:09.720 align:middle line:84% You're saying, yeah, "on the plane" is an adjunct all right. 00:51:09.720 --> 00:51:12.480 align:middle line:84% But "do so" is replacing the entire verb phrase 00:51:12.480 --> 00:51:16.770 align:middle line:84% "decide on the boat," where "on the boat" is an argument. 00:51:16.770 --> 00:51:18.420 align:middle line:84% And yes, that's another reading that it 00:51:18.420 --> 00:51:20.040 align:middle line:90% ought to be able to have. 00:51:20.040 --> 00:51:21.780 align:middle line:84% I was just ignoring that reading. 00:51:21.780 --> 00:51:23.430 align:middle line:84% Thank you for pulling that reading out 00:51:23.430 --> 00:51:24.960 align:middle line:84% and making me pay attention to it. 00:51:24.960 --> 00:51:26.770 align:middle line:90% You're absolutely right. 00:51:26.770 --> 00:51:27.640 align:middle line:90% Yeah? 00:51:27.640 --> 00:51:28.330 align:middle line:90% Cool. 00:51:28.330 --> 00:51:29.200 align:middle line:90% All right. 00:51:29.200 --> 00:51:33.100 align:middle line:90% So arguments versus adjuncts-- 00:51:33.100 --> 00:51:35.080 align:middle line:84% "decide," "on the boat," "on the plane." 00:51:35.080 --> 00:51:37.420 align:middle line:90% "On the boat" is an argument. 00:51:37.420 --> 00:51:40.360 align:middle line:84% This means I chose the boat on the plane, at least 00:51:40.360 --> 00:51:41.860 align:middle line:90% on the most natural prosody. 00:51:41.860 --> 00:51:47.710 align:middle line:84% And we've now seen two ways of distinguishing arguments 00:51:47.710 --> 00:51:48.310 align:middle line:90% from adjuncts. 00:51:48.310 --> 00:51:51.460 align:middle line:84% One is if you have anything left over after 00:51:51.460 --> 00:51:53.230 align:middle line:90% VP-pronominalization-- 00:51:53.230 --> 00:51:55.030 align:middle line:84% anything that's outside, "do so." 00:51:55.030 --> 00:51:56.740 align:middle line:84% Anything that's still there had better 00:51:56.740 --> 00:51:59.290 align:middle line:84% be an adjunct and not an argument because "do so" 00:51:59.290 --> 00:52:02.090 align:middle line:90% doesn't select for things. 00:52:02.090 --> 00:52:05.050 align:middle line:84% And if you have both an argument and an adjunct, 00:52:05.050 --> 00:52:09.460 align:middle line:84% the argument tends to be closer to the verb than the adjunct. 00:52:09.460 --> 00:52:12.070 align:middle line:84% Those are the two tests that I've given you so far. 00:52:12.070 --> 00:52:17.970 align:middle line:90% 00:52:17.970 --> 00:52:21.085 align:middle line:84% And then here's a mini-constituency test 00:52:21.085 --> 00:52:23.460 align:middle line:84% that's meant to get you to take seriously the possibility 00:52:23.460 --> 00:52:25.680 align:middle line:84% that there is a constituent, "decide on the boat." 00:52:25.680 --> 00:52:28.560 align:middle line:84% Look, you can coordinate that constituent 00:52:28.560 --> 00:52:30.330 align:middle line:84% with another constituent of the same kind. 00:52:30.330 --> 00:52:32.100 align:middle line:84% You can say, "Mary will decide on the boat 00:52:32.100 --> 00:52:34.075 align:middle line:90% and read a novel on the plane." 00:52:34.075 --> 00:52:34.575 align:middle line:90% That's fine. 00:52:34.575 --> 00:52:37.680 align:middle line:90% 00:52:37.680 --> 00:52:41.610 align:middle line:84% So, terminology break-- the lower thing is the complement. 00:52:41.610 --> 00:52:43.860 align:middle line:90% It's the sister of the head. 00:52:43.860 --> 00:52:45.840 align:middle line:84% So "complement" is a name that we 00:52:45.840 --> 00:52:48.190 align:middle line:84% use for sisters of heads which, as we've seen, 00:52:48.190 --> 00:52:52.980 align:middle line:84% is a position that's kind of privileged for arguments, 00:52:52.980 --> 00:52:54.250 align:middle line:90% for things that are selected. 00:52:54.250 --> 00:52:59.070 align:middle line:84% So if there's only going to be one thing that 00:52:59.070 --> 00:53:01.860 align:middle line:84% can be-- if there's a competition between an argument 00:53:01.860 --> 00:53:03.663 align:middle line:84% and an adjunct to be the complement, 00:53:03.663 --> 00:53:05.580 align:middle line:84% well, it's the argument that's the complement. 00:53:05.580 --> 00:53:07.538 align:middle line:84% That's the thing that gets to be the sister. 00:53:07.538 --> 00:53:08.580 align:middle line:90% That's what we're seeing. 00:53:08.580 --> 00:53:12.770 align:middle line:90% 00:53:12.770 --> 00:53:15.940 align:middle line:84% So here's the principle that I introduced informally 00:53:15.940 --> 00:53:18.550 align:middle line:84% a second ago and I'll now say a little more formally. 00:53:18.550 --> 00:53:21.310 align:middle line:84% If you've got a head that's selecting for an argument, what 00:53:21.310 --> 00:53:25.930 align:middle line:84% we're seeing is you need to take care of the selection 00:53:25.930 --> 00:53:29.470 align:middle line:84% requirements of the head first before you do anything else. 00:53:29.470 --> 00:53:31.900 align:middle line:84% You must first eat your broccoli before you 00:53:31.900 --> 00:53:33.130 align:middle line:90% can have your dessert. 00:53:33.130 --> 00:53:34.810 align:middle line:84% You have to take your core classes 00:53:34.810 --> 00:53:37.540 align:middle line:84% before you can switch to taking all linguistics classes. 00:53:37.540 --> 00:53:40.670 align:middle line:90% That's the way life is. 00:53:40.670 --> 00:53:43.630 align:middle line:84% So if you're going to read books quickly, 00:53:43.630 --> 00:53:45.430 align:middle line:90% "read" is a transitive verb. 00:53:45.430 --> 00:53:49.210 align:middle line:84% It can combine with a noun phrase. 00:53:49.210 --> 00:53:52.930 align:middle line:84% And so you're going to want to give it its noun phrase first. 00:53:52.930 --> 00:53:55.540 align:middle line:84% That's the first thing you'll merge with "read." 00:53:55.540 --> 00:53:56.950 align:middle line:90% "Quickly" is an adjunct. 00:53:56.950 --> 00:53:58.720 align:middle line:90% You can do anything quickly. 00:53:58.720 --> 00:54:00.460 align:middle line:84% It's not getting selected by "read." 00:54:00.460 --> 00:54:02.860 align:middle line:84% We won't find verbs that can combine with "quickly" 00:54:02.860 --> 00:54:03.940 align:middle line:90% and verbs that can't. 00:54:03.940 --> 00:54:08.380 align:middle line:84% And so, yeah, you need to satisfy the needs of "read" 00:54:08.380 --> 00:54:11.200 align:middle line:84% first before you go merging "quickly," which is just 00:54:11.200 --> 00:54:13.160 align:middle line:90% kind of an optional thing. 00:54:13.160 --> 00:54:15.760 align:middle line:84% And that's why the story is going to be-- 00:54:15.760 --> 00:54:18.160 align:middle line:84% we can say, "He read the book quickly," 00:54:18.160 --> 00:54:21.580 align:middle line:84% but not, "He read quickly the book," in English. 00:54:21.580 --> 00:54:22.780 align:middle line:90% Yeah? 00:54:22.780 --> 00:54:25.150 align:middle line:84% AUDIENCE: In languages that have flexible word order-- 00:54:25.150 --> 00:54:26.370 align:middle line:90% NORVIN RICHARDS: Yes. 00:54:26.370 --> 00:54:27.760 align:middle line:90% [LAUGHS] Let me sit down. 00:54:27.760 --> 00:54:29.920 align:middle line:84% AUDIENCE: How do these sentences-- 00:54:29.920 --> 00:54:34.420 align:middle line:84% the sentence trees work out if you can rearrange? 00:54:34.420 --> 00:54:37.252 align:middle line:84% NORVIN RICHARDS: So you're pointing to the fact 00:54:37.252 --> 00:54:39.460 align:middle line:84% that there are many languages out there in which word 00:54:39.460 --> 00:54:42.940 align:middle line:84% order is a lot more flexible than it is in English. 00:54:42.940 --> 00:54:44.920 align:middle line:90% And you know what? 00:54:44.920 --> 00:54:47.140 align:middle line:84% Even in English, it sometimes gets 00:54:47.140 --> 00:54:48.680 align:middle line:90% more complicated than this. 00:54:48.680 --> 00:54:52.060 align:middle line:84% So let me show you the ways in which it gets more complicated 00:54:52.060 --> 00:54:52.930 align:middle line:90% in English. 00:54:52.930 --> 00:54:59.020 align:middle line:84% And when we get to that, we will have a tool 00:54:59.020 --> 00:55:00.610 align:middle line:84% that we can use to describe what's 00:55:00.610 --> 00:55:03.670 align:middle line:84% going on in the languages you have in mind. 00:55:03.670 --> 00:55:07.090 align:middle line:84% So let me just get you to hold off on your question for now. 00:55:07.090 --> 00:55:10.810 align:middle line:84% Are there any other questions about this so far? 00:55:10.810 --> 00:55:11.590 align:middle line:90% OK. 00:55:11.590 --> 00:55:12.090 align:middle line:90% All right. 00:55:12.090 --> 00:55:14.700 align:middle line:90% 00:55:14.700 --> 00:55:17.010 align:middle line:84% In fact, I think we might be about to do it right now. 00:55:17.010 --> 00:55:17.510 align:middle line:90% Yeah. 00:55:17.510 --> 00:55:18.870 align:middle line:90% So where are we? 00:55:18.870 --> 00:55:20.990 align:middle line:84% Trees are constructed by binary Merge. 00:55:20.990 --> 00:55:23.817 align:middle line:84% Merge is constrained by selection via this thing I just 00:55:23.817 --> 00:55:25.650 align:middle line:84% called the Projection Principle, which says, 00:55:25.650 --> 00:55:27.330 align:middle line:84% if a head selects for something, then 00:55:27.330 --> 00:55:29.910 align:middle line:84% that should be the first thing you merge. 00:55:29.910 --> 00:55:31.970 align:middle line:84% Before, some of you were tormenting me 00:55:31.970 --> 00:55:33.720 align:middle line:84% with questions about heads that seem to be 00:55:33.720 --> 00:55:35.190 align:middle line:90% selecting more than one thing. 00:55:35.190 --> 00:55:37.440 align:middle line:84% And you can kind of interpret the Projection Principle 00:55:37.440 --> 00:55:39.450 align:middle line:84% in a way that's sympathetic to that. 00:55:39.450 --> 00:55:41.550 align:middle line:84% You get to say, yeah, if you have a head that 00:55:41.550 --> 00:55:44.760 align:middle line:84% selects for more than one thing, "give the children books" 00:55:44.760 --> 00:55:49.440 align:middle line:84% or "put the book on the table," it has two needs. 00:55:49.440 --> 00:55:51.300 align:middle line:84% And you had better satisfy those needs 00:55:51.300 --> 00:55:54.210 align:middle line:84% as fast as you can, modulo binary merge. 00:55:54.210 --> 00:55:56.227 align:middle line:90% So give it one of its arguments. 00:55:56.227 --> 00:55:57.810 align:middle line:84% And then it still has another argument 00:55:57.810 --> 00:56:00.630 align:middle line:84% it needs, so you better merge that one next. 00:56:00.630 --> 00:56:03.810 align:middle line:84% The tools that we have right now make that a possible answer 00:56:03.810 --> 00:56:06.750 align:middle line:90% to those kinds of questions. 00:56:06.750 --> 00:56:08.993 align:middle line:84% Questions still floating around about whether those 00:56:08.993 --> 00:56:10.410 align:middle line:84% are the right trees, but those are 00:56:10.410 --> 00:56:12.660 align:middle line:90% the trees we expect right now. 00:56:12.660 --> 00:56:14.120 align:middle line:90% Yep. 00:56:14.120 --> 00:56:15.460 align:middle line:90% OK. 00:56:15.460 --> 00:56:16.400 align:middle line:90% All right. 00:56:16.400 --> 00:56:18.130 align:middle line:90% Yeah? 00:56:18.130 --> 00:56:18.630 align:middle line:90% Yes? 00:56:18.630 --> 00:56:20.670 align:middle line:84% AUDIENCE: So for a verb like "give," 00:56:20.670 --> 00:56:22.880 align:middle line:84% and you are trying to make a construction like "give 00:56:22.880 --> 00:56:27.450 align:middle line:84% the children books," would it make sense 00:56:27.450 --> 00:56:30.090 align:middle line:84% to instead say that "books" as an argument 00:56:30.090 --> 00:56:36.850 align:middle line:84% is an argument for what you get after you merge "gives" 00:56:36.850 --> 00:56:39.240 align:middle line:90% with that first argument? 00:56:39.240 --> 00:56:41.400 align:middle line:90% It's kind of like-- 00:56:41.400 --> 00:56:43.200 align:middle line:84% NORVIN RICHARDS: Yeah, I see what you mean. 00:56:43.200 --> 00:56:43.700 align:middle line:90% Sorry. 00:56:43.700 --> 00:56:45.696 align:middle line:90% Go ahead. 00:56:45.696 --> 00:56:48.880 align:middle line:84% AUDIENCE: Kind of like cutting in lambda calculus. 00:56:48.880 --> 00:56:49.795 align:middle line:90% NORVIN RICHARDS: Yes. 00:56:49.795 --> 00:56:50.670 align:middle line:90% Yes, it is like that. 00:56:50.670 --> 00:56:53.170 align:middle line:84% That's a nice way to talk about it. 00:56:53.170 --> 00:56:57.288 align:middle line:90% So I've been talking as though-- 00:56:57.288 --> 00:57:00.300 align:middle line:84% let's see if I can say it coherently. 00:57:00.300 --> 00:57:04.590 align:middle line:84% I've been talking as though when we look up "give," "give" 00:57:04.590 --> 00:57:07.273 align:middle line:84% is going to say, well, I need two noun phrases. 00:57:07.273 --> 00:57:08.940 align:middle line:84% Or at least there is a version of "give" 00:57:08.940 --> 00:57:10.710 align:middle line:90% that takes two noun phrases. 00:57:10.710 --> 00:57:14.708 align:middle line:84% You're asking me, wait, couldn't-- 00:57:14.708 --> 00:57:17.250 align:middle line:84% first of all, don't we want to say something more complicated 00:57:17.250 --> 00:57:17.880 align:middle line:90% than that? 00:57:17.880 --> 00:57:19.657 align:middle line:84% And we saw this with "put" as well. 00:57:19.657 --> 00:57:21.240 align:middle line:84% It doesn't just want two noun phrases. 00:57:21.240 --> 00:57:23.760 align:middle line:84% It wants two noun phrases in a particular order. 00:57:23.760 --> 00:57:25.920 align:middle line:84% So you want to "give the children books." 00:57:25.920 --> 00:57:29.070 align:middle line:84% You don't want to "give the books children," 00:57:29.070 --> 00:57:31.770 align:middle line:90% meaning give children books. 00:57:31.770 --> 00:57:34.860 align:middle line:84% So the first noun phrase in this list 00:57:34.860 --> 00:57:36.810 align:middle line:84% is the noun phrase that gets the books, 00:57:36.810 --> 00:57:37.980 align:middle line:90% gets the second noun phrase. 00:57:37.980 --> 00:57:39.760 align:middle line:90% It's not the other way around. 00:57:39.760 --> 00:57:42.070 align:middle line:84% And in "put the book on the table," 00:57:42.070 --> 00:57:44.010 align:middle line:84% "put" maybe selects for a noun phrase 00:57:44.010 --> 00:57:48.082 align:middle line:84% and a prepositional phrase, but we 00:57:48.082 --> 00:57:50.040 align:middle line:84% need to say something more structured than that 00:57:50.040 --> 00:57:51.990 align:middle line:84% because it has to be the noun phrase first and then 00:57:51.990 --> 00:57:52.800 align:middle line:90% the prepositional phrase. 00:57:52.800 --> 00:57:54.280 align:middle line:84% It can't be the other way around. 00:57:54.280 --> 00:57:57.660 align:middle line:84% And you're wondering, can we say something like, 00:57:57.660 --> 00:58:02.790 align:middle line:84% "give" is a function that takes a noun phrase 00:58:02.790 --> 00:58:06.840 align:middle line:84% and gets converted into another function that also takes a noun 00:58:06.840 --> 00:58:08.295 align:middle line:90% phrase, something like that. 00:58:08.295 --> 00:58:11.042 align:middle line:90% 00:58:11.042 --> 00:58:12.750 align:middle line:84% That's sort of the equivalent of currying 00:58:12.750 --> 00:58:17.490 align:middle line:84% in or Schönfinkelization or whatever. 00:58:17.490 --> 00:58:19.890 align:middle line:84% If you ever study semantics in my department, 00:58:19.890 --> 00:58:22.940 align:middle line:84% I have several German semantics colleagues, 00:58:22.940 --> 00:58:24.690 align:middle line:84% colleagues whose native language is German 00:58:24.690 --> 00:58:26.520 align:middle line:90% and they are semanticists. 00:58:26.520 --> 00:58:31.787 align:middle line:84% And there were two people who invented the kind of functions 00:58:31.787 --> 00:58:33.870 align:middle line:84% that he's talking about, where you have a function 00:58:33.870 --> 00:58:36.120 align:middle line:84% and it changes the function into another function that 00:58:36.120 --> 00:58:38.200 align:middle line:90% takes another kind of thing. 00:58:38.200 --> 00:58:40.440 align:middle line:84% One of them had the surname Curry, 00:58:40.440 --> 00:58:42.865 align:middle line:84% and the other had the surname Schönfinkel. 00:58:42.865 --> 00:58:45.720 align:middle line:90% 00:58:45.720 --> 00:58:47.340 align:middle line:90% This guy was, I think, British. 00:58:47.340 --> 00:58:49.170 align:middle line:90% And this person was German. 00:58:49.170 --> 00:58:52.660 align:middle line:84% So most English speakers refer to this as currying in. 00:58:52.660 --> 00:58:55.950 align:middle line:84% But if you are German, you refer to it as Schönfinkelization. 00:58:55.950 --> 00:58:56.910 align:middle line:90% [LAUGHTER] 00:58:56.910 --> 00:59:00.360 align:middle line:84% And apparently Schönfinkel discovered it before Curry, 00:59:00.360 --> 00:59:03.420 align:middle line:84% so the Germans are annoyed by the fact that we call it 00:59:03.420 --> 00:59:04.502 align:middle line:90% currying. 00:59:04.502 --> 00:59:06.210 align:middle line:84% I'm not German, so I'm not annoyed by you 00:59:06.210 --> 00:59:08.580 align:middle line:90% calling it currying in. 00:59:08.580 --> 00:59:10.950 align:middle line:84% Indeed, we clearly want something more structured 00:59:10.950 --> 00:59:12.600 align:middle line:90% than what I'm giving us. 00:59:12.600 --> 00:59:15.645 align:middle line:84% And I'm just going to leave your point right there. 00:59:15.645 --> 00:59:17.400 align:middle line:84% We need to do something more sophisticated 00:59:17.400 --> 00:59:19.578 align:middle line:90% than what I'm doing here. 00:59:19.578 --> 00:59:21.120 align:middle line:84% I keep hinting at the fact that we're 00:59:21.120 --> 00:59:24.120 align:middle line:84% going to want to investigate these trees, the trees that I'm 00:59:24.120 --> 00:59:26.160 align:middle line:84% showing you, with other techniques 00:59:26.160 --> 00:59:30.390 align:middle line:84% to find out whether these are the right constituency 00:59:30.390 --> 00:59:31.270 align:middle line:90% structures. 00:59:31.270 --> 00:59:33.660 align:middle line:84% And when we do that, I think we'll 00:59:33.660 --> 00:59:36.743 align:middle line:84% have another reason to revise what I'm telling you now. 00:59:36.743 --> 00:59:38.910 align:middle line:84% So there are several reasons to be uneasy about what 00:59:38.910 --> 00:59:40.920 align:middle line:90% I'm telling you now. 00:59:40.920 --> 00:59:43.770 align:middle line:84% I've shown you a theory which works fine as long as 00:59:43.770 --> 00:59:46.032 align:middle line:90% heads only have one argument. 00:59:46.032 --> 00:59:47.490 align:middle line:84% As soon as they have two arguments, 00:59:47.490 --> 00:59:50.538 align:middle line:84% these kinds of unsettling questions arise. 00:59:50.538 --> 00:59:52.080 align:middle line:84% And we're going to come back to them. 00:59:52.080 --> 00:59:53.630 align:middle line:90% OK? 00:59:53.630 --> 00:59:54.960 align:middle line:90% Good question. 00:59:54.960 --> 00:59:56.790 align:middle line:90% Other questions about this? 00:59:56.790 --> 00:59:57.870 align:middle line:90% This is great. 00:59:57.870 --> 01:00:01.140 align:middle line:84% I feel like I'm in a graduate seminar. 01:00:01.140 --> 01:00:02.150 align:middle line:90% OK? 01:00:02.150 --> 01:00:02.810 align:middle line:90% All right. 01:00:02.810 --> 01:00:05.210 align:middle line:90% So this is where we are then. 01:00:05.210 --> 01:00:07.145 align:middle line:84% Trees are constructed by binary Merge. 01:00:07.145 --> 01:00:09.020 align:middle line:84% We have the Projection Principle, which says, 01:00:09.020 --> 01:00:10.220 align:middle line:84% if there's a selection relation, you 01:00:10.220 --> 01:00:11.780 align:middle line:84% need to do that Merge relation first. 01:00:11.780 --> 01:00:14.690 align:middle line:90% 01:00:14.690 --> 01:00:17.570 align:middle line:84% And that's going to get us contrasts like "Mary wrote 01:00:17.570 --> 01:00:19.785 align:middle line:84% the novel on a typewriter" versus "Mary 01:00:19.785 --> 01:00:22.160 align:middle line:84% wrote on a typewriter the novel," where the second one is 01:00:22.160 --> 01:00:26.270 align:middle line:84% no good because "write" has the option of being 01:00:26.270 --> 01:00:27.560 align:middle line:90% a transitive verb. 01:00:27.560 --> 01:00:32.000 align:middle line:84% And when it's a transitive verb, it selects for a direct object. 01:00:32.000 --> 01:00:33.590 align:middle line:90% And that's the novel. 01:00:33.590 --> 01:00:35.990 align:middle line:84% "On a typewriter" is like "with a feather." 01:00:35.990 --> 01:00:37.520 align:middle line:84% You can do anything on a typewriter. 01:00:37.520 --> 01:00:38.880 align:middle line:90% You can combine with anything. 01:00:38.880 --> 01:00:39.830 align:middle line:90% So it's an adjunct. 01:00:39.830 --> 01:00:42.110 align:middle line:84% And so there's no hurry to merge it. 01:00:42.110 --> 01:00:47.072 align:middle line:90% 01:00:47.072 --> 01:00:48.780 align:middle line:84% So we've worked our way to the conclusion 01:00:48.780 --> 01:00:51.150 align:middle line:84% that "write," when it's transitive, absolutely, 01:00:51.150 --> 01:00:53.792 align:middle line:84% absolutely has to select an object. 01:00:53.792 --> 01:00:56.250 align:middle line:84% And the object has to be the first thing you merge together 01:00:56.250 --> 01:00:57.360 align:middle line:90% with "write." 01:00:57.360 --> 01:00:59.772 align:middle line:84% So it has to be right after "write" in English. 01:00:59.772 --> 01:01:01.230 align:middle line:84% That's why you can say, "Mary wrote 01:01:01.230 --> 01:01:02.355 align:middle line:90% the novel on a typewriter." 01:01:02.355 --> 01:01:05.112 align:middle line:84% You cannot say "Mary wrote on a typewriter the novel." 01:01:05.112 --> 01:01:06.570 align:middle line:84% The novel, the thing that's getting 01:01:06.570 --> 01:01:08.895 align:middle line:84% written, it absolutely has to be right next to "write." 01:01:08.895 --> 01:01:11.850 align:middle line:90% 01:01:11.850 --> 01:01:13.770 align:middle line:90% Darn! 01:01:13.770 --> 01:01:18.270 align:middle line:84% That last example looks problematic, right? 01:01:18.270 --> 01:01:21.510 align:middle line:84% If I ask you, "What did Mary write on a typewriter," 01:01:21.510 --> 01:01:24.300 align:middle line:84% we've just worked our way to the conclusion that "write," 01:01:24.300 --> 01:01:26.308 align:middle line:84% when it takes a direct object, the direct object 01:01:26.308 --> 01:01:28.350 align:middle line:84% has to be the first thing you Merge with "write." 01:01:28.350 --> 01:01:29.808 align:middle line:84% It has to be right next to "write." 01:01:29.808 --> 01:01:32.010 align:middle line:84% There can't be anything intervening between "write" 01:01:32.010 --> 01:01:36.840 align:middle line:84% and it's complement, the thing that's getting written. 01:01:36.840 --> 01:01:39.960 align:middle line:84% What's the thing that's getting written in that example? 01:01:39.960 --> 01:01:41.820 align:middle line:84% What did Mary write on the typewriter? 01:01:41.820 --> 01:01:42.510 align:middle line:90% Joseph? 01:01:42.510 --> 01:01:42.780 align:middle line:90% AUDIENCE: "What." 01:01:42.780 --> 01:01:43.780 align:middle line:90% NORVIN RICHARDS: "What." 01:01:43.780 --> 01:01:45.000 align:middle line:90% Yeah. 01:01:45.000 --> 01:01:47.648 align:middle line:84% This sounds like an Abbott and Costello routine. 01:01:47.648 --> 01:01:48.690 align:middle line:90% "What's getting written." 01:01:48.690 --> 01:01:50.070 align:middle line:90% Yes. 01:01:50.070 --> 01:01:50.820 align:middle line:90% [LAUGHTER] 01:01:50.820 --> 01:01:52.740 align:middle line:90% 01:01:52.740 --> 01:01:55.320 align:middle line:84% But that's not the sister of "write." 01:01:55.320 --> 01:01:56.520 align:middle line:90% That's nowhere near "write." 01:01:56.520 --> 01:01:58.680 align:middle line:84% It's way the heck at the beginning of the sentence. 01:01:58.680 --> 01:02:02.000 align:middle line:90% What the heck is going on? 01:02:02.000 --> 01:02:03.942 align:middle line:84% Are you all emotionally distraught? 01:02:03.942 --> 01:02:05.150 align:middle line:90% That's what I'm trying to do. 01:02:05.150 --> 01:02:05.650 align:middle line:90% Yeah? 01:02:05.650 --> 01:02:07.670 align:middle line:90% OK, good. 01:02:07.670 --> 01:02:11.450 align:middle line:84% So two possible responses-- oh, well, we 01:02:11.450 --> 01:02:13.730 align:middle line:84% didn't have time to get too attached to the Projection 01:02:13.730 --> 01:02:14.438 align:middle line:90% Principle anyway. 01:02:14.438 --> 01:02:16.590 align:middle line:84% It was just a couple of slides ago. 01:02:16.590 --> 01:02:18.200 align:middle line:90% So much for that idea. 01:02:18.200 --> 01:02:19.420 align:middle line:90% That's one possibility. 01:02:19.420 --> 01:02:21.698 align:middle line:84% So apparently heads can select for things, 01:02:21.698 --> 01:02:23.490 align:middle line:84% and they don't have to be anywhere near it. 01:02:23.490 --> 01:02:25.020 align:middle line:90% So, well. 01:02:25.020 --> 01:02:25.520 align:middle line:90% But, look. 01:02:25.520 --> 01:02:26.645 align:middle line:90% Here's another possibility. 01:02:26.645 --> 01:02:30.860 align:middle line:84% We could say, no, the Projection Principle is right-- 01:02:30.860 --> 01:02:32.840 align:middle line:90% is correct. 01:02:32.840 --> 01:02:35.360 align:middle line:84% I should have picked a different verb. 01:02:35.360 --> 01:02:37.140 align:middle line:84% The Projection Principle is correct. 01:02:37.140 --> 01:02:38.660 align:middle line:84% It is true that a verb like "write" 01:02:38.660 --> 01:02:40.400 align:middle line:84% that selects for an object absolutely has 01:02:40.400 --> 01:02:41.692 align:middle line:90% to merge with the object first. 01:02:41.692 --> 01:02:45.010 align:middle line:84% And so, really what you're doing when 01:02:45.010 --> 01:02:47.350 align:middle line:84% you start to ask a question like, "What did Mary 01:02:47.350 --> 01:02:48.820 align:middle line:84% write on a typewriter?" is you are 01:02:48.820 --> 01:02:50.635 align:middle line:90% merging "write" with "what." 01:02:50.635 --> 01:02:53.020 align:middle line:84% "What" is starting off as the sister of "write." 01:02:53.020 --> 01:02:55.810 align:middle line:84% So you start off with something more like "Mary wrote what 01:02:55.810 --> 01:02:59.110 align:middle line:90% on a typewriter." 01:02:59.110 --> 01:03:01.630 align:middle line:84% And then, apparently, there is some operation 01:03:01.630 --> 01:03:04.260 align:middle line:84% that takes "what" and puts it somewhere else, 01:03:04.260 --> 01:03:05.635 align:middle line:84% at the beginning of the sentence. 01:03:05.635 --> 01:03:10.780 align:middle line:90% 01:03:10.780 --> 01:03:14.590 align:middle line:90% So-- this was a while ago-- 01:03:14.590 --> 01:03:17.050 align:middle line:84% Kateryna asked me, what about languages 01:03:17.050 --> 01:03:18.730 align:middle line:90% where word order is freer? 01:03:18.730 --> 01:03:21.640 align:middle line:84% And I said, things are about to get more complicated. 01:03:21.640 --> 01:03:24.110 align:middle line:84% Things have now gotten more complicated. 01:03:24.110 --> 01:03:25.900 align:middle line:90% So, yeah, there's selection. 01:03:25.900 --> 01:03:27.070 align:middle line:90% Things select for things. 01:03:27.070 --> 01:03:29.920 align:middle line:84% And when x selects for y, then y needs 01:03:29.920 --> 01:03:31.780 align:middle line:90% to merge with x right away. 01:03:31.780 --> 01:03:34.330 align:middle line:84% And there are adjuncts that can merge wherever they want. 01:03:34.330 --> 01:03:36.460 align:middle line:90% And there are things like this-- 01:03:36.460 --> 01:03:38.440 align:middle line:90% movement operations. 01:03:38.440 --> 01:03:41.150 align:middle line:84% I can spend some time trying to convince you that this is true. 01:03:41.150 --> 01:03:44.350 align:middle line:84% But right now let me just assert that it's true-- cases where, 01:03:44.350 --> 01:03:48.008 align:middle line:84% indeed, things do merge just the way we've convinced ourselves 01:03:48.008 --> 01:03:48.550 align:middle line:90% that they do. 01:03:48.550 --> 01:03:49.758 align:middle line:90% They merge where they should. 01:03:49.758 --> 01:03:51.400 align:middle line:90% They merge as sisters of verbs. 01:03:51.400 --> 01:03:56.080 align:middle line:84% But then something else happens, and they end up somewhere else. 01:03:56.080 --> 01:03:58.000 align:middle line:84% So there's this operation of movement 01:03:58.000 --> 01:04:00.370 align:middle line:84% where you take the word "what," and you move it 01:04:00.370 --> 01:04:03.160 align:middle line:84% to the beginning of the sentence in order to form questions 01:04:03.160 --> 01:04:04.480 align:middle line:90% like this one. 01:04:04.480 --> 01:04:07.450 align:middle line:84% Notice that in English, and indeed in many languages-- 01:04:07.450 --> 01:04:08.290 align:middle line:90% most languages. 01:04:08.290 --> 01:04:11.290 align:middle line:84% Not all languages, but in many languages, 01:04:11.290 --> 01:04:15.580 align:middle line:84% if I'm amazed by something you've 01:04:15.580 --> 01:04:20.920 align:middle line:84% said-- if you say, "Mary wrote a proof of Fermat's Last Theorem 01:04:20.920 --> 01:04:23.710 align:middle line:84% on a typewriter," I can say, "Mary 01:04:23.710 --> 01:04:25.660 align:middle line:90% wrote what on the typewriter?" 01:04:25.660 --> 01:04:28.330 align:middle line:84% So it actually is possible to say these words in this order. 01:04:28.330 --> 01:04:31.180 align:middle line:84% I just have to be in a particular emotional state. 01:04:31.180 --> 01:04:32.995 align:middle line:90% I have to be astonished. 01:04:32.995 --> 01:04:35.170 align:middle line:84% This is actually cross-linguistically extremely 01:04:35.170 --> 01:04:36.910 align:middle line:90% common for-- 01:04:36.910 --> 01:04:39.340 align:middle line:84% they're called echo questions, where, 01:04:39.340 --> 01:04:41.933 align:middle line:84% if I can't believe what you just said 01:04:41.933 --> 01:04:44.350 align:middle line:84% or if I couldn't quite hear what you just said-- if you're 01:04:44.350 --> 01:04:47.710 align:middle line:84% telling me on the T, and you say, 01:04:47.710 --> 01:04:50.047 align:middle line:84% "Mary wrote [GARGLING NOISE] on a typewriter," 01:04:50.047 --> 01:04:51.880 align:middle line:84% and there's lots of noise in the background, 01:04:51.880 --> 01:04:53.960 align:middle line:84% and I missed the crucial thing, I can say, 01:04:53.960 --> 01:04:55.997 align:middle line:84% "Mary wrote what on a typewriter?" 01:04:55.997 --> 01:04:57.580 align:middle line:84% These are called echo questions, where 01:04:57.580 --> 01:05:01.300 align:middle line:84% I'm repeating most of what you just said 01:05:01.300 --> 01:05:04.150 align:middle line:84% and replacing part of it with a wh-word. 01:05:04.150 --> 01:05:06.100 align:middle line:84% And it's extremely common cross-linguistically 01:05:06.100 --> 01:05:09.100 align:middle line:84% for it to be possible to leave these question 01:05:09.100 --> 01:05:12.077 align:middle line:84% words like "what" in the positions 01:05:12.077 --> 01:05:14.660 align:middle line:84% where you would expect them to be in those kinds of questions, 01:05:14.660 --> 01:05:16.577 align:middle line:84% in the positions where they would be selected. 01:05:16.577 --> 01:05:18.062 align:middle line:90% It's not universal, actually. 01:05:18.062 --> 01:05:20.020 align:middle line:84% There are languages that don't let you do that. 01:05:20.020 --> 01:05:22.640 align:middle line:84% If you're working on a language, it 01:05:22.640 --> 01:05:24.760 align:middle line:84% could be interesting to do some finding out 01:05:24.760 --> 01:05:29.430 align:middle line:84% about how it forms these kinds of questions. 01:05:29.430 --> 01:05:32.450 align:middle line:84% OK, so lots of questions about what's 01:05:32.450 --> 01:05:34.910 align:middle line:90% going on in this last example. 01:05:34.910 --> 01:05:36.920 align:middle line:84% There are several reasons to think 01:05:36.920 --> 01:05:40.670 align:middle line:90% that "what" is ending up in CP. 01:05:40.670 --> 01:05:41.750 align:middle line:90% Remember CP? 01:05:41.750 --> 01:05:47.300 align:middle line:84% It was the thing that had heads like "whether" and "that." 01:05:47.300 --> 01:05:50.420 align:middle line:84% And we said verbs like "know" and "think"-- 01:05:50.420 --> 01:05:52.700 align:middle line:84% and "wonder" I think was the example I used before-- 01:05:52.700 --> 01:05:54.740 align:middle line:90% they select for clauses. 01:05:54.740 --> 01:05:56.900 align:middle line:84% And in particular, they select for properties 01:05:56.900 --> 01:05:58.200 align:middle line:90% of the head of the clause. 01:05:58.200 --> 01:06:02.525 align:middle line:84% So "think" can select for a clause that's headed by "that." 01:06:02.525 --> 01:06:03.800 align:middle line:90% "That" is the complementizer. 01:06:03.800 --> 01:06:06.410 align:middle line:84% That's our word for it in syntax. 01:06:06.410 --> 01:06:08.300 align:middle line:84% "That" is the complementizer that introduces 01:06:08.300 --> 01:06:10.185 align:middle line:90% that embedded clause. 01:06:10.185 --> 01:06:11.810 align:middle line:84% "I don't know whether he ate the ants." 01:06:11.810 --> 01:06:15.050 align:middle line:84% "Whether" is a complementizer that can be selected by "know." 01:06:15.050 --> 01:06:17.840 align:middle line:84% You can't "think whether he ate the ants." 01:06:17.840 --> 01:06:20.270 align:middle line:84% And actually there are verbs like "know" that 01:06:20.270 --> 01:06:23.510 align:middle line:84% can have either "whether" or "that." 01:06:23.510 --> 01:06:25.700 align:middle line:84% With a verb like "know" that can have "whether," 01:06:25.700 --> 01:06:29.670 align:middle line:84% it can also have "what" immediately following it. 01:06:29.670 --> 01:06:33.590 align:middle line:84% So you can say, "I don't know what he ate." 01:06:33.590 --> 01:06:36.320 align:middle line:84% And notice that if it has "what," 01:06:36.320 --> 01:06:38.420 align:middle line:84% it cannot have either "that" or "whether." 01:06:38.420 --> 01:06:43.550 align:middle line:84% So you can say, "I know that he ate the ants." 01:06:43.550 --> 01:06:46.910 align:middle line:84% You can say, "I know whether he ate the ants." 01:06:46.910 --> 01:06:48.683 align:middle line:84% And you can say, "I know what he ate." 01:06:48.683 --> 01:06:50.600 align:middle line:84% But you cannot say, "I know what that he ate," 01:06:50.600 --> 01:06:51.890 align:middle line:90% or "what whether he ate." 01:06:51.890 --> 01:06:54.020 align:middle line:84% So "that" and "whether" and "what"-- they all 01:06:54.020 --> 01:06:55.940 align:middle line:84% seem to be sort of in the same slot there. 01:06:55.940 --> 01:06:58.945 align:middle line:90% They're in the CP somehow. 01:06:58.945 --> 01:07:00.820 align:middle line:84% We'll see they're not quite in the same slot. 01:07:00.820 --> 01:07:02.710 align:middle line:84% But the idea that "what" ends up somewhere 01:07:02.710 --> 01:07:07.330 align:middle line:84% in CP, that seems to be part of what's going on here. 01:07:07.330 --> 01:07:10.040 align:middle line:90% 01:07:10.040 --> 01:07:13.010 align:middle line:84% And then, as I said, I promised you there 01:07:13.010 --> 01:07:16.190 align:middle line:84% will be various reasons to think that "what" starts off 01:07:16.190 --> 01:07:19.040 align:middle line:84% as the sister of "write" and moves into CP. 01:07:19.040 --> 01:07:21.860 align:middle line:84% But one of them is the one I just went through. 01:07:21.860 --> 01:07:24.680 align:middle line:84% Under certain circumstances, communicative circumstances, 01:07:24.680 --> 01:07:26.845 align:middle line:84% I can say, "Mary wrote what on a typewriter?" 01:07:26.845 --> 01:07:28.220 align:middle line:84% like if I couldn't quite hear you 01:07:28.220 --> 01:07:31.610 align:middle line:84% or if I don't believe what you just said. 01:07:31.610 --> 01:07:33.830 align:middle line:84% So here's a tree that we're going 01:07:33.830 --> 01:07:36.860 align:middle line:84% to draw for "What will Mary write?" 01:07:36.860 --> 01:07:39.200 align:middle line:84% It's going to have "what" starting off 01:07:39.200 --> 01:07:41.780 align:middle line:84% as the sister of "write," right where the Projection 01:07:41.780 --> 01:07:43.250 align:middle line:90% Principle wants it to be. 01:07:43.250 --> 01:07:46.310 align:middle line:84% "Write" selects for a direct object, and it gets one. 01:07:46.310 --> 01:07:47.840 align:middle line:90% That's the sister of "write." 01:07:47.840 --> 01:07:52.790 align:middle line:84% And then it moves to the edge of the CP. 01:07:52.790 --> 01:07:56.480 align:middle line:84% And so that's the kind of tree that we're going to draw. 01:07:56.480 --> 01:07:59.400 align:middle line:84% This phenomenon-- there's a name for it. 01:07:59.400 --> 01:08:01.950 align:middle line:90% It's called wh-movement. 01:08:01.950 --> 01:08:04.850 align:middle line:84% It's called wh-movement because these question words-- 01:08:04.850 --> 01:08:10.640 align:middle line:84% these words like "what" and "where" and "why" and "when"-- 01:08:10.640 --> 01:08:15.350 align:middle line:84% tend to start with the letters W and H in English. 01:08:15.350 --> 01:08:17.840 align:middle line:84% "Who" and "what" and "where" and "when" and "why"-- 01:08:17.840 --> 01:08:22.430 align:middle line:84% and "how," which doesn't start with the letters W and H, 01:08:22.430 --> 01:08:24.560 align:middle line:84% but it does contain the letters W and H, 01:08:24.560 --> 01:08:26.130 align:middle line:90% just not in the right order. 01:08:26.130 --> 01:08:28.939 align:middle line:84% So these words are called wh-words. 01:08:28.939 --> 01:08:30.050 align:middle line:90% And they're called that. 01:08:30.050 --> 01:08:32.029 align:middle line:84% It's just a technical term in syntax. 01:08:32.029 --> 01:08:35.300 align:middle line:84% We call these words wh-words, regardless of the language 01:08:35.300 --> 01:08:36.229 align:middle line:90% we're working in. 01:08:36.229 --> 01:08:39.367 align:middle line:84% It's a sort of embarrassingly Anglocentric term. 01:08:39.367 --> 01:08:41.450 align:middle line:84% So they're called wh-words because in English they 01:08:41.450 --> 01:08:42.979 align:middle line:90% start with W and H. 01:08:42.979 --> 01:08:44.479 align:middle line:84% The language that you're working on, 01:08:44.479 --> 01:08:46.370 align:middle line:84% they surely will not start with W and H. 01:08:46.370 --> 01:08:49.160 align:middle line:84% But we're still going to want to call them wh-words. 01:08:49.160 --> 01:08:52.399 align:middle line:84% And this phenomenon is called wh-movement. 01:08:52.399 --> 01:08:55.310 align:middle line:84% And it's a cross-linguistically very common phenomenon. 01:08:55.310 --> 01:08:57.904 align:middle line:90% 01:08:57.904 --> 01:08:59.779 align:middle line:84% There's an English example of it there again. 01:08:59.779 --> 01:09:01.208 align:middle line:90% "What did you put on the table?" 01:09:01.208 --> 01:09:03.500 align:middle line:84% Here's the Tagalog for "What did you put on the table?" 01:09:03.500 --> 01:09:06.800 align:middle line:84% You can see the Tagalog word for "what" doesn't begin with W 01:09:06.800 --> 01:09:10.520 align:middle line:84% and H. Tagalog does have W and H, 01:09:10.520 --> 01:09:13.220 align:middle line:84% but words can't start with W and H. 01:09:13.220 --> 01:09:17.213 align:middle line:84% And there's the Finnish for "where did I put my clothes." 01:09:17.213 --> 01:09:19.130 align:middle line:84% I should really learn how to say "what did you 01:09:19.130 --> 01:09:20.660 align:middle line:90% put on the table" in Finnish. 01:09:20.660 --> 01:09:23.069 align:middle line:84% I don't know why that's my Finnish example. 01:09:23.069 --> 01:09:26.750 align:middle line:84% So various unrelated languages that have wh-movement. 01:09:26.750 --> 01:09:30.540 align:middle line:84% It's a cross-linguistically very common phenomenon. 01:09:30.540 --> 01:09:32.062 align:middle line:90% It's not universal. 01:09:32.062 --> 01:09:33.770 align:middle line:84% Some of you, in fact, are native speakers 01:09:33.770 --> 01:09:37.640 align:middle line:84% of languages in which wh-words don't move 01:09:37.640 --> 01:09:39.859 align:middle line:90% to the left edge of the clause. 01:09:39.859 --> 01:09:41.910 align:middle line:90% They just stay where they are. 01:09:41.910 --> 01:09:44.314 align:middle line:84% So in Chinese, for example, if you 01:09:44.314 --> 01:09:45.814 align:middle line:84% want to say "What did Zhangsan buy," 01:09:45.814 --> 01:09:48.120 align:middle line:84% you don't say, "What did Zhangsan buy," 01:09:48.120 --> 01:09:49.370 align:middle line:90% or at least you don't have to. 01:09:49.370 --> 01:09:51.260 align:middle line:84% The standard thing to say is, literally, 01:09:51.260 --> 01:09:53.930 align:middle line:84% "Zhangsan bought what," which, as I said, 01:09:53.930 --> 01:09:57.230 align:middle line:84% you can say in English, but only under special circumstances. 01:09:57.230 --> 01:09:58.940 align:middle line:84% In Chinese, that's just the natural way 01:09:58.940 --> 01:10:01.520 align:middle line:84% to ask that question, is to leave [INAUDIBLE] right 01:10:01.520 --> 01:10:04.580 align:middle line:84% where it would be as the object of the verb. 01:10:04.580 --> 01:10:06.680 align:middle line:84% And here are a couple more examples-- 01:10:06.680 --> 01:10:08.690 align:middle line:84% Bafut, which is a language of Cameroon, 01:10:08.690 --> 01:10:12.200 align:middle line:84% and Hopi, which is a language of the American Southwest. 01:10:12.200 --> 01:10:15.100 align:middle line:90% 01:10:15.100 --> 01:10:18.160 align:middle line:84% So there's cross-linguistic variation 01:10:18.160 --> 01:10:19.635 align:middle line:90% in how you form wh-questions. 01:10:19.635 --> 01:10:21.010 align:middle line:84% There are languages, like English 01:10:21.010 --> 01:10:23.350 align:middle line:84% and Finnish and Tagalog and many other languages, where 01:10:23.350 --> 01:10:25.240 align:middle line:84% you take your wh-words and you move them 01:10:25.240 --> 01:10:27.600 align:middle line:90% to the left edge of CP. 01:10:27.600 --> 01:10:29.350 align:middle line:84% And then there are languages, like Chinese 01:10:29.350 --> 01:10:32.600 align:middle line:84% and Bafut and Hopi, where the wh-words just don't move. 01:10:32.600 --> 01:10:36.060 align:middle line:84% They just stay where they would normally go, 01:10:36.060 --> 01:10:38.060 align:middle line:84% wherever they would normally go in the sentence. 01:10:38.060 --> 01:10:39.380 align:middle line:90% There's no movement going on. 01:10:39.380 --> 01:10:42.820 align:middle line:84% So in Chinese, if you wanted to say, 01:10:42.820 --> 01:10:44.950 align:middle line:84% "Zhangsan bought a book," instead of "what," 01:10:44.950 --> 01:10:48.160 align:middle line:84% you would just put the word for "book." 01:10:48.160 --> 01:10:50.270 align:middle line:84% Chinese normal word order is subject, 01:10:50.270 --> 01:10:51.940 align:middle line:84% verb, object, kind of like English. 01:10:51.940 --> 01:10:53.380 align:middle line:90% Same deal for Bafut. 01:10:53.380 --> 01:10:55.270 align:middle line:84% Hopi has the verb at the end of the sentence. 01:10:55.270 --> 01:10:58.702 align:middle line:84% So these wh-words, these words that mean things 01:10:58.702 --> 01:11:00.160 align:middle line:84% like "what" and "who," they're just 01:11:00.160 --> 01:11:02.050 align:middle line:84% going where the object would normally go 01:11:02.050 --> 01:11:04.910 align:middle line:90% in these wh-in-situ languages. 01:11:04.910 --> 01:11:08.660 align:middle line:84% And those are almost the only two options. 01:11:08.660 --> 01:11:11.690 align:middle line:84% Take your wh-word and move it to the left edge of CP, 01:11:11.690 --> 01:11:14.720 align:middle line:84% or leave it where it would normally go. 01:11:14.720 --> 01:11:16.550 align:middle line:90% You can imagine others, right? 01:11:16.550 --> 01:11:20.240 align:middle line:84% So, for example, it's not clear that there 01:11:20.240 --> 01:11:22.400 align:middle line:84% are any languages in which you take the wh-word 01:11:22.400 --> 01:11:24.945 align:middle line:84% and you move it to the end of the clause. 01:11:24.945 --> 01:11:26.570 align:middle line:84% You could imagine a language like that, 01:11:26.570 --> 01:11:27.890 align:middle line:84% but it's not clear that there are 01:11:27.890 --> 01:11:30.390 align:middle line:84% any languages in which you say, "You put on the table what?" 01:11:30.390 --> 01:11:33.110 align:middle line:90% 01:11:33.110 --> 01:11:37.400 align:middle line:84% I think my next slides make that observation more complicated. 01:11:37.400 --> 01:11:39.687 align:middle line:84% And I think I want to skip them, actually. 01:11:39.687 --> 01:11:41.270 align:middle line:84% So to make this fact more interesting, 01:11:41.270 --> 01:11:44.288 align:middle line:84% let me show you one other kind of cross-linguistic variation. 01:11:44.288 --> 01:11:45.330 align:middle line:90% I'm not going to do that. 01:11:45.330 --> 01:11:46.580 align:middle line:90% So let me do that fast. 01:11:46.580 --> 01:11:49.940 align:middle line:84% We'll come back to this because we're running low on time. 01:11:49.940 --> 01:11:51.830 align:middle line:84% And I want to show you one other quick thing. 01:11:51.830 --> 01:11:56.053 align:middle line:90% 01:11:56.053 --> 01:11:57.470 align:middle line:84% In the remaining five minutes, I'm 01:11:57.470 --> 01:11:59.090 align:middle line:90% going to set up another thing. 01:11:59.090 --> 01:12:03.230 align:middle line:84% And then we'll come back, and we'll start here next time. 01:12:03.230 --> 01:12:06.290 align:middle line:84% I've been concentrating on wh-questions where you're only 01:12:06.290 --> 01:12:08.150 align:middle line:90% asking about one thing. 01:12:08.150 --> 01:12:10.410 align:middle line:84% In many languages-- not all languages, 01:12:10.410 --> 01:12:12.260 align:middle line:84% but in many languages, it's possible to ask 01:12:12.260 --> 01:12:14.280 align:middle line:84% about more than one thing at the same time. 01:12:14.280 --> 01:12:16.943 align:middle line:84% So you can ask questions like, "What did you give to whom," 01:12:16.943 --> 01:12:18.860 align:middle line:84% where, if I ask you a question like that, what 01:12:18.860 --> 01:12:23.870 align:middle line:84% I want is a list of people and things such that person 01:12:23.870 --> 01:12:28.052 align:middle line:84% x got thing x and person y got thing y. 01:12:28.052 --> 01:12:30.635 align:middle line:84% That's what I'm asking you if I ask you that kind of question. 01:12:30.635 --> 01:12:36.060 align:middle line:90% 01:12:36.060 --> 01:12:37.630 align:middle line:90% Again, there's some variation. 01:12:37.630 --> 01:12:40.650 align:middle line:84% So English, as I said, is a language with wh-movement. 01:12:40.650 --> 01:12:44.340 align:middle line:84% But it's specifically a language with wh-movement of one WH. 01:12:44.340 --> 01:12:46.800 align:middle line:84% So if you want to ask a multiple wh-question, 01:12:46.800 --> 01:12:49.230 align:middle line:84% you move one of your wh-words, but you leave 01:12:49.230 --> 01:12:50.640 align:middle line:90% the other ones where they are. 01:12:50.640 --> 01:12:52.445 align:middle line:84% So you say, "What did you give to whom?" 01:12:52.445 --> 01:12:55.860 align:middle line:84% or "What did you give to whom when, where, why, how?" 01:12:55.860 --> 01:12:57.720 align:middle line:84% So all your other wh-words just stay 01:12:57.720 --> 01:12:59.880 align:middle line:90% where they would normally go. 01:12:59.880 --> 01:13:03.120 align:middle line:84% There are other languages in which all of the wh-words 01:13:03.120 --> 01:13:07.870 align:middle line:84% move in this big, moving herd to the beginning of the clause. 01:13:07.870 --> 01:13:10.140 align:middle line:84% So in Bulgarian, for example, you 01:13:10.140 --> 01:13:12.570 align:middle line:84% don't ask literally, "What did you give to whom?" 01:13:12.570 --> 01:13:16.390 align:middle line:84% You ask literally, "What to whom did he give?" 01:13:16.390 --> 01:13:18.900 align:middle line:84% So "what" and "to whom" are both at the beginning 01:13:18.900 --> 01:13:19.710 align:middle line:90% of the sentence. 01:13:19.710 --> 01:13:21.335 align:middle line:84% There are bunch of languages like this. 01:13:21.335 --> 01:13:23.070 align:middle line:90% Mohawk is another one. 01:13:23.070 --> 01:13:27.390 align:middle line:84% Mohawk's a language spoken not too far from here, 01:13:27.390 --> 01:13:30.620 align:middle line:90% in parts of America and Canada. 01:13:30.620 --> 01:13:33.400 align:middle line:84% These are both languages in which the wh-words all 01:13:33.400 --> 01:13:35.290 align:middle line:84% have to move in a multiple wh-question. 01:13:35.290 --> 01:13:38.930 align:middle line:90% 01:13:38.930 --> 01:13:42.350 align:middle line:84% And again, there's some variation between languages. 01:13:42.350 --> 01:13:47.690 align:middle line:84% There are languages that move one wh-word, like English. 01:13:47.690 --> 01:13:50.330 align:middle line:84% There are languages that move no wh-words, like Chinese. 01:13:50.330 --> 01:13:52.460 align:middle line:84% There are languages that move all the wh-words, 01:13:52.460 --> 01:13:54.980 align:middle line:90% like Bulgarian or Mohawk. 01:13:54.980 --> 01:13:57.660 align:middle line:84% And again, it's not too hard to imagine other options. 01:13:57.660 --> 01:13:59.870 align:middle line:84% So for example, you could imagine a language 01:13:59.870 --> 01:14:02.810 align:middle line:90% that only moves two wh-words. 01:14:02.810 --> 01:14:04.460 align:middle line:84% That would be a language where you 01:14:04.460 --> 01:14:07.940 align:middle line:84% would say, "Who what gave to whom?" 01:14:07.940 --> 01:14:11.750 align:middle line:90% to mean who gave what to whom. 01:14:11.750 --> 01:14:13.358 align:middle line:84% There aren't any languages like that. 01:14:13.358 --> 01:14:14.900 align:middle line:84% The only kinds of languages there are 01:14:14.900 --> 01:14:16.700 align:middle line:90% are the English kind-- 01:14:16.700 --> 01:14:18.620 align:middle line:84% so first, the Mandarin kind, the Chinese kind, 01:14:18.620 --> 01:14:19.790 align:middle line:84% where you don't move any of them; 01:14:19.790 --> 01:14:22.310 align:middle line:84% the English kind, where you move one, where you say, "Who 01:14:22.310 --> 01:14:25.370 align:middle line:84% gave what to whom?"; and the Bulgarian/Mohawk kind, where 01:14:25.370 --> 01:14:26.120 align:middle line:90% you move them all. 01:14:26.120 --> 01:14:28.460 align:middle line:84% You say, "Who what to whom gave?" 01:14:28.460 --> 01:14:31.130 align:middle line:84% There aren't any languages like this that just 01:14:31.130 --> 01:14:33.390 align:middle line:90% move two or move up to two. 01:14:33.390 --> 01:14:37.380 align:middle line:84% You can imagine a language like that, but there aren't any. 01:14:37.380 --> 01:14:39.960 align:middle line:84% And let me just leave you with this last point, 01:14:39.960 --> 01:14:42.465 align:middle line:84% and we'll probably take it up here again next time. 01:14:42.465 --> 01:14:46.020 align:middle line:90% 01:14:46.020 --> 01:14:49.470 align:middle line:84% Logical problem of language acquisition-- 01:14:49.470 --> 01:14:50.820 align:middle line:90% here's a game we can play. 01:14:50.820 --> 01:14:52.060 align:middle line:90% Here's a function. 01:14:52.060 --> 01:14:54.450 align:middle line:84% It's a function that if you give it 1, the answer is 1. 01:14:54.450 --> 01:14:55.530 align:middle line:84% If you give it 2, the answer is 2. 01:14:55.530 --> 01:14:56.947 align:middle line:84% If you give it 3, the answer is 3. 01:14:56.947 --> 01:14:58.410 align:middle line:84% If you give it 4, the answer is 4. 01:14:58.410 --> 01:15:01.020 align:middle line:84% What do you think the answer is if you give it 5? 01:15:01.020 --> 01:15:01.520 align:middle line:90% AUDIENCE: 5. 01:15:01.520 --> 01:15:02.990 align:middle line:90% NORVIN RICHARDS: 5? 01:15:02.990 --> 01:15:04.830 align:middle line:84% Can anybody think of any other options? 01:15:04.830 --> 01:15:05.330 align:middle line:90% Ha! 01:15:05.330 --> 01:15:06.740 align:middle line:90% You've fallen into my trap. 01:15:06.740 --> 01:15:09.957 align:middle line:84% The function that I actually had in mind was this one. 01:15:09.957 --> 01:15:11.420 align:middle line:90% [LAUGHTER] 01:15:11.420 --> 01:15:12.800 align:middle line:90% Such that the answer is 29. 01:15:12.800 --> 01:15:16.190 align:middle line:84% In fact, the answer could be anything at all. 01:15:16.190 --> 01:15:18.830 align:middle line:84% So this is the kind of question you sometimes get 01:15:18.830 --> 01:15:20.300 align:middle line:90% asked on standardized tests. 01:15:20.300 --> 01:15:23.820 align:middle line:84% And it's always unfair because the answer could be anything 01:15:23.820 --> 01:15:24.320 align:middle line:90% at all. 01:15:24.320 --> 01:15:26.150 align:middle line:84% That first thing which I carefully 01:15:26.150 --> 01:15:29.360 align:middle line:84% devised to be 0 as long as the thing I was giving the function 01:15:29.360 --> 01:15:32.880 align:middle line:84% was 1, 2, 3, or 4, you could multiply that by anything. 01:15:32.880 --> 01:15:36.590 align:middle line:90% And so 5 could be any number. 01:15:36.590 --> 01:15:37.855 align:middle line:90% Life is like that a lot. 01:15:37.855 --> 01:15:39.980 align:middle line:84% There are lots of places where you go through life, 01:15:39.980 --> 01:15:42.710 align:middle line:84% and you get a certain number of finite observations 01:15:42.710 --> 01:15:44.630 align:middle line:90% of how life works. 01:15:44.630 --> 01:15:46.160 align:middle line:84% And then you have to make a decision 01:15:46.160 --> 01:15:47.690 align:middle line:90% about what the basic rule is. 01:15:47.690 --> 01:15:49.335 align:middle line:84% So you're doing science, let's say. 01:15:49.335 --> 01:15:50.210 align:middle line:90% And you go out there. 01:15:50.210 --> 01:15:54.253 align:middle line:90% You see many white swans. 01:15:54.253 --> 01:15:55.670 align:middle line:84% At some point, you have to decide, 01:15:55.670 --> 01:15:57.500 align:middle line:84% have I seen enough white swans that I get 01:15:57.500 --> 01:15:59.360 align:middle line:90% to decide that swans are white? 01:15:59.360 --> 01:16:01.250 align:middle line:84% Have I seen enough black crows that I get 01:16:01.250 --> 01:16:03.260 align:middle line:90% to say that crows are black? 01:16:03.260 --> 01:16:04.040 align:middle line:90% You never know. 01:16:04.040 --> 01:16:06.940 align:middle line:84% Maybe the next one is going to be purple or green. 01:16:06.940 --> 01:16:08.090 align:middle line:90% It's just hard to know. 01:16:08.090 --> 01:16:12.470 align:middle line:84% And this function is sort of a dramatic example of that. 01:16:12.470 --> 01:16:15.080 align:middle line:84% Imagine, just in the remaining seconds, 01:16:15.080 --> 01:16:17.930 align:middle line:90% that you are a Bulgarian child. 01:16:17.930 --> 01:16:20.550 align:middle line:84% Were any of you Bulgarian children? 01:16:20.550 --> 01:16:22.520 align:middle line:84% OK, so just imagine counterfactually 01:16:22.520 --> 01:16:24.950 align:middle line:90% that you were a Bulgarian child. 01:16:24.950 --> 01:16:27.560 align:middle line:84% Here you are, growing up, hearing 01:16:27.560 --> 01:16:30.620 align:middle line:90% your parents ask wh-questions. 01:16:30.620 --> 01:16:34.520 align:middle line:84% And there's got to be some number, n, 01:16:34.520 --> 01:16:37.790 align:middle line:84% that is the largest number of wh-words 01:16:37.790 --> 01:16:40.370 align:middle line:90% you ever heard in a question. 01:16:40.370 --> 01:16:44.660 align:middle line:84% Maybe you heard your parents say two wh-words in a question. 01:16:44.660 --> 01:16:46.520 align:middle line:90% Maybe you heard them say three. 01:16:46.520 --> 01:16:48.110 align:middle line:84% It's kind of unlikely that you ever 01:16:48.110 --> 01:16:50.240 align:middle line:90% heard them say four or five. 01:16:50.240 --> 01:16:53.030 align:middle line:84% I mean, it depends, I guess, on what your parents' line of work 01:16:53.030 --> 01:16:54.080 align:middle line:90% was. 01:16:54.080 --> 01:16:56.270 align:middle line:84% But there's some number that's the largest number. 01:16:56.270 --> 01:16:59.360 align:middle line:84% And it's probably different from Bulgarian to Bulgarian. 01:16:59.360 --> 01:17:00.780 align:middle line:90% So what we might expect-- 01:17:00.780 --> 01:17:03.140 align:middle line:84% so Bulgarians are sort of in the position 01:17:03.140 --> 01:17:06.290 align:middle line:84% you guys were in with that function. 01:17:06.290 --> 01:17:09.260 align:middle line:84% If you just hear this, let's say, "What to whom 01:17:09.260 --> 01:17:10.910 align:middle line:90% did he give?" 01:17:10.910 --> 01:17:13.130 align:middle line:84% you've got to ask yourself, is the rule move 01:17:13.130 --> 01:17:14.540 align:middle line:90% all the wh-phrases? 01:17:14.540 --> 01:17:16.670 align:middle line:84% Is it move two of the wh-phrases? 01:17:16.670 --> 01:17:20.300 align:middle line:84% Is it move maximally three, maximally four, whatever? 01:17:20.300 --> 01:17:23.120 align:middle line:84% There are literally infinitely many rules 01:17:23.120 --> 01:17:26.370 align:middle line:84% that are compatible with this data point. 01:17:26.370 --> 01:17:28.800 align:middle line:84% And so you might have expected that Bulgarians, depending 01:17:28.800 --> 01:17:31.320 align:middle line:84% on how weird their parents were, would grow up 01:17:31.320 --> 01:17:32.700 align:middle line:84% with slightly different grammars, 01:17:32.700 --> 01:17:34.860 align:middle line:84% that when you took Bulgarians and you put them 01:17:34.860 --> 01:17:39.600 align:middle line:84% into a hyperbaric chamber and you gradually added wh-words 01:17:39.600 --> 01:17:41.640 align:middle line:84% to the questions, that they would diverge 01:17:41.640 --> 01:17:43.650 align:middle line:84% as the questions got more and more complicated, 01:17:43.650 --> 01:17:46.500 align:middle line:84% that there would be the two-wh Bulgarians 01:17:46.500 --> 01:17:49.305 align:middle line:84% and the three-wh Bulgarians and the four and the five, 01:17:49.305 --> 01:17:50.890 align:middle line:90% that you would find that out. 01:17:50.890 --> 01:17:52.290 align:middle line:90% That's not what we find out. 01:17:52.290 --> 01:17:54.330 align:middle line:90% The experiment has been done. 01:17:54.330 --> 01:17:56.740 align:middle line:84% What you find out is the Bulgarians are all the same. 01:17:56.740 --> 01:18:00.420 align:middle line:84% They all move all their wh-phrases. 01:18:00.420 --> 01:18:04.890 align:middle line:84% What we think is going on really is that being a human being 01:18:04.890 --> 01:18:08.040 align:middle line:84% means having the kind of mind that can put language together 01:18:08.040 --> 01:18:10.960 align:middle line:90% in some ways but not others. 01:18:10.960 --> 01:18:14.040 align:middle line:84% There was never any danger of the Bulgarian children 01:18:14.040 --> 01:18:16.140 align:middle line:84% entertaining the possibility that the grammar 01:18:16.140 --> 01:18:18.240 align:middle line:90% was move two wh-phrases. 01:18:18.240 --> 01:18:21.030 align:middle line:84% They knew that there aren't languages like that. 01:18:21.030 --> 01:18:23.050 align:middle line:84% Human languages don't work that way. 01:18:23.050 --> 01:18:25.980 align:middle line:84% This is what we mean when we talk about universal grammar. 01:18:25.980 --> 01:18:28.350 align:middle line:84% We'll review this next time because we're out of time. 01:18:28.350 --> 01:18:31.220 align:middle line:84% But we'll talk more about this next time. 01:18:31.220 --> 01:18:37.000 align:middle line:90%