Monday 25 April 2022

Going places in Welsh and English

I'm learning Welsh on Duolingo. Most people in our department are doing a language on it, in some cases for practical purposes but for many of us, for family reasons. In my case, my grandmother was Welsh and spoke it as a first language, so I feel a connection to it for that reason even though I'm not Welsh myself. My other grandparents also lived in Pembrokeshire for a long time, though they weren't Welsh, and so when we visited we'd hear it spoken, especially if we went along the coast a bit to Carmarthenshire. 

I knew a fair amount about Welsh grammar before I began, because I'm a linguist so I happen to know quite a few things about quite a few languages that I don't speak, and Welsh is a language that has been studied and written about reasonably often. It's frequently studied alongside others as well as in its own right because you can compare it to other Celtic languages for within-family comparative work, or with languages from other families where it's useful as an example of a language that has VSO word order. 

Backtrack! VSO? V stands for Verb, S for Subject and O for object, and these three letters are how we talk about the basic or default order of the elements of sentences in a language. English is SVO: [S The British public] [V despise] [O the current government]. This is one of the two most common orders globally, the other being SOV (e.g. Korean). VSO is less common and it can be useful, when examining linguistic theories, to illustrate how they work in languages with this order. The same VSO order is found in languages all over the world, including Berber, the Mayan languages, and Te Reo Māori. 

There are lots of other interesting things about Welsh grammar, obviously, it's just that this is the one I happen to have read most about because word order is the area I've worked in most of all. And of course VSO is an abstraction and there is widespread use of auxiliary verbs (the English equivalent would be something like The British public is despising the government) which is still VSO but it's not quite so straightforward as just classifying languages and being done with it. Luckily, I suppose, or I'd be out of a job. 

So all this is to say that as I begin to work out what the heck is going on with this language, I might write some posts about it. So far I'm floundering in the dark as many things are very different from English. The app starts you off with things where you can make comparisons, but I'm honestly finding some of it baffling. Of course you don't get explicit grammar instruction on Duolingo, so I've got myself a grammar book as well to help me work out the things that aren't easily deducible. 

I assume they deliberately choose constructions to teach you that are similar to your language of instruction (English, in my case), to make things easier at the beginning. For me, a linguist, some of this was actually really surprising to me. For instance, all of the prepositions and articles that I've had to use so far have worked just like in English. Maybe this doesn't sound so strange but I know very well that prepositions and articles are infamously unpredictable and variable across languages. So when I'm getting sentences to translate that include a direct word-for-word equivalent of going to the office or whatever, I'm surprised. The main thing that has been different in this respect is the absence of an indefinite article (a/an in English), so here I felt back on more solid ground as I know this is a feature common to lots of languages and fully expected it. 

Another, more specific, thing that seems weirdly the same is the verb(-type thing) mynd, which apparently means 'going'. English uses go for a travelling or motion event, like going to the office: I'm travelling from one place to another. Welsh also appears to do this: mynd i'r swyddfa is the direct equivalent, and both languages need to use it with a subject and an auxiliary verb: I'm going to the office or Dw i'n mynd i'r swyddfa. Here's the weird bit though: English also uses go for expressing the future, which is not totally bizarre but also not universal by any means, and Welsh seems to do the same. So just as in English you can say I'm going to wait here, you can say Dw i'n mynd i aros yma in Welsh, and they both express future with no motion involved. This use of go for expressing the future isn't unique to English and Welsh (French and Spanish have it, and Jamal Ouhalla describes it for Moroccan Arabic here) but it's also not that common. Perhaps someone can tell me whether this similarity is a result of the very close contact between the languages, a shared familial feature, or simply coincidence that both languages have travelled this grammaticalisation path, just like Moroccan Arabic. 

Monday 18 April 2022

It wasn't then, but it was now

Linguists are liable to stop mid-conversation to write down something they or their interlocutor said, if it was linguistically interesting. Out for drinks with colleagues, the following happened. 
David, picking up a nearly-empty beer: Is this mine? 
*David downs the last of the beer*
Me: Well, it was now! 
And at this point we stopped to reflect on the fact that you really can't say that, unless you're a linguist mucking about with language for funnies. 

Now is a flexible sort of thing. It doesn't only mean this precise moment (which can be measured in infinitesimally smaller moments anyway). It can be the present time give or take a few minutes, or a few hours: 
I'm busy right now; I'll call you back when I'm finished. 
Or the present time give or take a few days: 
The delivery is due around now, I think. 
It can be the present time, probably extending into the future, since some past causal event or just compared to before: 
We have to submit form 3A now to claim expenses, not form 3B (since the new manager updated the policy). 

We generally have better living conditions now (compared to in the last century), though there are still too many people living in poverty.  

My now was, I think, a present situation resulting from a causal event, where since David had downed the pint, it was now (and forever, I suppose), his. If the beer wasn't his before, It is now

But as I said my line, I mentally switched it from It is now. Because the beer no longer existed, I changed the tense from present (is) to past (was). I left the temporal adverbial now untouched, however, thereby creating a weird mismatch between past time (the tense of the verb expressing the existence of the beer) and present time (the adverbial expressing the time that the state of affairs described pertains to). 

This is why we have things like present perfect and past perfect, that you might have studied at school: I have drunk the beer is present (state of affairs) and perfect (action completed in the past), so although the beer is already drunk, I'm describing the present situation. I had drunk the beer is about a completed action and a past state of affairs – the situation that I'm describing, in which the beer had already been drunk, was at some past time. 

I was, in effect, trying to make now cover both the present time (our current situation) and a past time (when the beer existed). Now can easily include some time in the past, like all of the past time since the new manager changed the policy in the example above, but crucially it has to be conceptualised as not being in a distinct past time. It has to be a single now period compared to a before period (or then, or whatever). So it has to be present tense for a current state of affairs. 

And lastly, the reason my line was a dorky linguist joke rather than an incomprehensible failure of tenses is that the phrase It is now! has become a set phrase or 'chunk' of language. I wasn't consciously referencing the commentary of the 1966 football world cup final, but this is an extremely well-known use of the phrase and I think it has got into people's vocabulary, including mine, as an idiomatic expression. It highlights the inevitability of the current situation, the sense of having pinched a victory, and the impossibility of having the outcome reversed. 

Monday 11 April 2022

At last, someone has written about Wordle!

I've held off on blogging about Wordle, because everyone else did it, and because I didn't have anything particular to say. People tend to assume that if you're a linguist, you like word games, but I don't think that's any more true for us than for normal people. Some of us do, others don't. I happen to love crosswords (because there is a quiz or a puzzle element) and dislike Scrabble (because I'm not good at anagrams). I do, as it happens, love Wordle. I love logic puzzles like sudoku, and this is basically just a logic puzzle with an added constraint. 

There is, or used to be, a board game called Mastermind which was a pure logic version of Wordle. (If you don't know what Wordle is, by the way, I don't know where you've been. It's what we all spent the early part of 2022 doing.) There, the thing you had to guess was the sequence of coloured pegs. There were only a few colours, and only a sequence of four, so much fewer than the 26 letters and five slots that Wordle involves. And you needed like ten goes to get it, rather than the six that you get with Wordle. The rules were the same: you got told if you'd got one right and in the right place, or right but in the wrong place, or wrong. You weren't told which one, though, which did make it harder in that respect (otherwise it would have been incredibly easy). I loved this game and I'm not sure why I never had my own copy (maybe no one else liked playing it with me, or maybe I never mentioned that I liked it?) but I played it when I was at other people's houses. 

So yes, I do love Wordle, because of the logic puzzle aspect. The word part of it does add something interesting for me, though. I like the constraint it puts on the possible answers. It's not the case, as in Mastermind, that every combination is equally possible. Some just aren't, or are much less likely, and that's due to the rules of either languages in general, or English in particular. So an example of a language-in-general thing is that there are going to be some vowels in the word, and some consonants. An example of an English-in-particular thing is that the last letter is probably an 'e' or a consonant, because we don't have so many words that end in 'a', 'i', 'o' or 'u' (though we do have some, so it's not absolutely ruled out). Another English-in-particular thing is that if you know you've got an 'h' in there somewhere, it's possibly the first letter but if it's not, you've likely also got a 't' for 'th' or a 'g' for 'gh' in there. Not always; ahead would have stumped me in that case. 

Screenshot of my Wordle stats showing a normal distribution with most words taking me four guesses to get.

I've been paying attention to how I solve them, and I usually get the answer on the fourth go. I imagine this is true for most people, as we'd expect a 'normal distribution' with very few right on the first or second go (that's a lucky guess) and few taking six (that's some bad luck or a word that has many very similar to it). 

I'm not sharing any new insights on how to solve them – I just do the same as you all do and rule out the most common letters first until I can see what it's likely to be. But what interests me is how quickly you get to the point where it can only realistically be one word. This is normally where I am by guess four. 

Here are a couple of recent ones, where the answers were epoxy and lowly. Just coincidence that they both end in a 'y', I think. I vary my starting words but always try to include some common letters. Sometimes I just use things I see nearby like the dogs' names. In both these cases, by the time I'd had three guesses I didn't have many right, but I had ruled out nearly all the possibilities, and there was only one possible word that I could think of in each case that could fit what I knew. 


Screenshot of Wordle with the word 'lowly', correct on the fourth go with few correct letters on the previous three. The previous image shows the same but for 'epoxy', but I can't edit the alt text for some reason.

This is the most satisfying way of playing the game, I think. If you end up with only one letter to get and several possibilities, it becomes chance and annoying, and if you get it right with some lucky guesses you don't feel like you earnt it, whereas this way you feel happy that you worked it out. 

I also saw Lesley Jeffries talking about doing it in other languages, and noting that her guess distribution was much more spread, presumably because her vocabulary is not a large in those languages and so she is likely to need more goes to get it right than the average speaker of that language would (and she noted that she is relying on phonotactics, which is those rules of the language that I mentioned earlier). 

Monday 4 April 2022

Foreshadowing via conjunction

There is a film called Willy's Wonderland, starring Nicolas Cage, and it is bizarre and wonderful. You should watch it. Cage has to spend the night in a disused family amusement animatronic kids' party type place, working as a cleaner in order to pay off the bill for fixing his car. The man who makes this deal with him says something along the lines of, 

'You walk out of that building in the morning and I'll have your car waiting for you.'

I've written before about conjunctions ('and') working as a conditional, in my post about the instruction(?) Don't touch the actors and they won't touch you. There, I noted that you could interpret the whole sentence as a statement of fact, that you shouldn't touch the actors and, independently, they also will not touch you, but that in fact we interpret it as a conditional: if you don't touch the actors, they won't touch you. 

In the line from this film, the same is true. Either the man was saying to Nicolas Cage that he'd be able to leave in the morning and when he did so his car would be ready, or, very much foreshadowing what was to come, that there is a possibility that he might not walk out in the morning but that if he did, his car would be waiting. 

I felt like the intonation gives us a clue as to which it is, but I lack the technical skills to show you how it goes on this blog. It's something like this, though: in the conditional interpretation, we get rising intonation to the end of the first clause and then falling on the second one, whereas on the statement of facts interpretation, they're both falling, and similar to each other. 

(Yes, this could also be a threat, as noted in that other post as well. Context would have made that really weird – who offers to fix your car as a threat?)