a wolf angel is not a good angel

I hate when USPS doesn’t actually update the tracking information. It was sent three days ago, and they still haven’t bothered to update anything. I like being able to check up on where my package is.

Also, I hate the smell of aftershave. If you are dating me (which I hope you’re not, because as far as *I* know, I’m not dating anyone) please do not ever wear any.

A boyfriend once put on aftershave — I don’t remember why, we were just at his place, not planning on going anywhere. And when I smelled it,I made him go away and wash it off and change shirts. One of my roommates wears aftershave, so periodically the whole house smells of it. Ugh. But not as bad as when he comes in from smoking (which irritates me — we had said no smokers).

I was discussing linguistics-as-a-field with N, another student in my department, today. He’s fairly anti-syntax; he feels that as it’s structured (even OT syntax) things aren’t defined enough (true) and no real predictions are ever made (mostly true). I enjoy finding things; I enjoy finding correlations between these things and other things in syntax. But I think much of the explanation is . . . lacking, a bit, in being more than descriptive. Positing a feature that appears just where we need it to is not so much a leap forward.
That’s why I like doing syntax of less studied languages: I don’t have to be the 57 millionth person to study tough constructions in English. I can do more basic work — work I tend to find more valuable — and still find good information, valuable work.

Of course it’s hard, because syntax is the most crowded of the subfields. So even less-studied languages are not exactly unstudied. Unless you’re doing fieldwork, which is a huge goal of mine. ‘This beach has bigger waves than that beach.’ ‘His tan is darker than my tan.’ Why not do research in the sun, I ask you.

I think I’d like to be a semanticist, only I think I’m totally incapable of it. But almost nobody does crosslinguistic semantics (except in my department![1] Yay! No one works on only English in either syntax or semantics.[2] I am very good with that), and I feel it’s so important to do. You can’t claim generality based on English, or English-and-[insert IE language here]. I even know of non-native speakers who work almost solely on English semantics. Are they insane? (More than the rest of us, that is?)

I have some fun ideas about semantics, using the all-and-only-overt-morphemes-matter approach. Also using the movement-free approach, which . . . I still like movement. It just makes so much sense for questions and for relative clauses.

I went into linguistics because I love how languages work, and I still love neat little pieces of data in the odd language. Data matters.

I mentioned to N how when I was working on my honours thesis in undergrad, I was asking one of the profs (not my advisor) about how I would explain this, given the word order and the meaning and whatever. And she said ‘Oh, word order doesn’t matter.’ What? What is the point of syntax if we drop word order entirely? The sad thing is, she was serious.

I don’t want to be like that, so into the newest latest coolest technologies that I lose sight of reality.

[1] Now any of the linguists who might be reading this can be absolutely sure about what school I’m at, on the off chance that (a) they cared and (b) they weren’t already. (a) seems pretty unlikely, though.

[2] Of course no one works on English in phonology. Why bother? Either you explain the byzantine stress system or you figure out which words take which affixes or you snore.

This just seems appropriate for today.

I also like http://www.despair.com/power.html. But my all-time favourite (and the one I am hoping gets made into a lithograph real soon) is http://www.despair.com/loneliness.html.

I’m not sure how I feel about this whole grad school thing today. I don’t know why, really. Oh, I suppose I have reasons:no clue what’s going on in semantics or how I could possibly find a topic in one; lacking vital data in syntax; unable to program, making learnability very difficult (and I screwed up on my tableaux — again — I need to start all over and see where I went wrong). And I’m tired and I’m still a bit lonely and it’s cold cold cold here.

And I read a post at Invisible Adjunct and I just feel . . . silenced. Not by what she wrote. But by my utter inability to know how to respond. I still don’t. I’ve been thinking about it.
I’ve written this paragraph time after time, and it never comes out to what I’m thinking. Worried and scared and enjoying my work and so much tangled together that I wonder if I shouldn’t just let it be, tiptoe around the questions and keep on going.

And since I totally forgot — a link: http://www.amritas.com/

Now we’re getting a little more detailed. The example on Amritas is fairly extreme — you’d need far more evidence than a form [pa] and one [panak] to assume the rules and representations there. You’d need more examples of the two deletion rules, more examples of the -k being its own suffix and not allomorphic with -nak, and you’d need to have good reason to associate [pa] as being the base of [panak]. This is possible, of course, and I have no idea how you’d determine if it were probable or not. You’d need to look at the specific language — discussions like these can make anything sound bad or good.

I’m still waiting to hear if he thinks you can ever differ from your underlying representation. If you can’t, you really throw out a great deal of modern phonology. Sometimes I think that would be ideal (I’m still not a phonologist, though I play one on the internet), though.

My landlords are still trying to finish construction here. Except that it seems to involve people doing loud stuff right next to my bedroom window every single morning at 8. I don’t need to wake up at 8, and I’m resenting that I have to do so quite so often.

If this continues tomorrow I will need to phone them and say that I absolutely need them to take some days off — or start later, whatever — because I cannot deal with this. I am so, so, so tired. I want to cry.

And I had such an exciting moment Friday night. I got a hit playing Jewish geography. Since I don’t remember names so well, that’s pretty impressive for me — especially with the whole different country thing. (We recognised on description of what someone is doing now and not names, which is probably how it happened. We weren’t really playing, but I’m going to count it as a win anyways.)

Okay, a little more context about phonology. The question appears to be not “can your UR be different from your output?” but rather “can your UR be different from all your outputs?” As in, could you have something like /pana/ -> [pa] and never ever see the /na/ part, in any context?

At first pass, I’m inclined to say no. There’s no reason you would posit something different from what you hear — at least not without overwhelming evidence. Morphophonemic alternations are good evidence, but then usually one of the things appear, just not in all contexts.

This is a debate in modern phonology. Let’s say you’re in a language that uses schwa to break up consonant clusters, but nowhere else. So there will (probably) be words that always always have schwa in them, just exactly where you’d predict based on the phonotactics. Are they there in your underlying representations? It used to be thought that the answer was no; we want to make our URs as simple as we possibly can. Then with OT and the whole Richness of the Base thing, it sort of changed to the answer being yes. You store what you hear, barring information to the contrary. Now I know there is some discussion on how divergent an underlying form can be from the output, but this is usually about one segment each in a few languages, as far as I know. (Look, I’m not a phonologist.)

I can think of a way that you could have a UR that wasn’t identical to any output you have, but shared each part with at least one output. Let’s imagine a language that adds glottal stops to beginnings of words if they start with vowels, and deletes final vowels.
Look at an input /ata/. You need to add a glottal stop, so you get {?ata}. Then you delete the final vowel (order here doesn’t matter) and you get [?at].

Now let’s imagine a prefix, /n-/. Add the prefix, you get /n-ata/. Delete the final vowel, you get [nat].

Why would you supply that final /a/? Well, if you only had prefixes, you wouldn’t. But — hey! — let’s imagine a suffix /-k/. Add that and you get /ata-k/, and finally [?atak]. Assuming you had further data that showed your suffix was /-k/ and not /-ak/ (imaginea form /iti-k/ -> [?itik] etc), you would, once you’re trying to get your copies of each morpheme down to a single form, end up with /ata/, even though you never hear it.

(I lie, a bit. You could hear it in /n-ata-k/, of course. There are more complex examples where you wouldn’t hear it, but this is good enough.)

This is tough work. There is research being done on how learnable it is — but we *do* have morphophonemic alternations, so it must be learnable somehow, even if we don’t know how. A carefully ordered set of rules does not work so well. This is why few people believe in just a carefully ordered set of rules — though as a way to see things, it is not the worst way to begin learning phonology.

Why lie in an intro phonology course? Should we simplify? Yes and no. It’s nice to know the truth, but the reality is sometimes you need a little bit of practice in the basics ofa field before actually learn how things really are. Courses in science are courses in successively smaller lies. Electrons do not circle the atom in nice orbits, no matter how useful a picture that is. But it’s a lot easier to learn that than to learn that a 3p electron has a 95% probability of being in this particular area, at first.

This isn’t to say that linguistics is like chemistry in that; it’s not. Or that linguistics, or MIT and satellite linguistics departments (which do not comprise all the linguistics done in the US, by any means) do not have problems. However, having marginally unfaithful outputs to your input is not one of them. (You are welcome to say that we memorise all the different forms. But you have to say something to explain it.)

I am intrigued by the incomplete discussion on Amaravati about the difference between underlying forms and output forms (or whatever terminology he prefers). Ignoring SPE, which was unusually happy about the idea of Very Different underlying forms, how does he want to get morphophonemic alternation without some set of rules (constraints, now[1])? Do we just store all the different forms in our head, with a list of when to use what? Because unless you have alll the different forms in your head, you need to have some rules to get from one to the other. One of those two has to be true. And sure, we could (for example) remember the three ways to pronounce the standard plural (-s, -z, -ez, where the e is actually a schwa) and the three ways to pronounce the past tense (-t, -d, -ed) and note that the rules are pretty much the same for both of them . .. it just seems painful. (Of course there are all sorts of other options. Seamless morphology is one I’ve heard of; it probably works just fine for English, but I’m suspicious of how it would work for a nice synthetic language.)

Perhaps it’s unfair to argue before I’ve seen his response, but not that unfair. Look, we know that we pronounce words a little different depending on the suffixes (sane vs. sanity) and we pronounce suffixes differently depending on the word (as above, dogs, cats, foxes) and we pronounce prefixes differently depending on the word (impossible, indefensible, incorrigeable). All languages do this, to some extent (some greater than English, some less). So either we have two or more ways to pronounce lots of words or sometimes words come out differently than we store them in our head. (Actually, it could be both, but let’s ignore that for the moment.) Both have pluses and minuses — you can pick your poison. But it’s a little disingenuous to show an example of rule-based phonology with entirely no context and say “look! Look how silly!” Unless it’s SPE.

[1] And why are we using rules? Ease of exposition, perhaps? Most people have gone on to constraint-based systems, I believe.

A remarkably productive afternoon. I headed off to Borders. It’s usually a second choice for a bookstore, but it has better coffee and the movie my mother wanted me to buy for her. So Borders it was. Also it’s closer. I had intended to try to rent a movie after, but decided I wasn’t in the mood.

I got a fair amount of my handout for Wednesday done; the rest is empty because I don’t have the data, yet. I thought I knew wht they would be, but then something has made me wary, so it’ll be a great excitement when I find out. Or something. Linguistics by email? Not a good idea.

I also started pulling together my 4-constraint system. I need to go and see what entirely mutually exclusive languages I have that each take parts of a different, coherent language. Then I need to see how disentaglable that is. I should be doing this, say now, but I am not. I should also be having dinner but am not. All those shoulds.

Maybe I should be sleeping. They seem to be doing loud construction here in the mornings again. Weekends only, of course. I am tired and cold.