I just found myself caught in a bit of a grammatical confusion (it happens sometimes).
For example: “Did you watch the football on Saturday afternoon?” Wnest ti wylio’r bêl-droed bnawn Sadwrn? (I’m unsure how to add the accent on the “e”).
Whenever I translate “you watched,” my instinct is to use “Wnest ti wylio”, but “Wyliest ti” is often suggested as the translation.
After doing some research, I found this online:
Short-form past tense: Used for common or irregular verbs that have their own built-in past tense (e.g., Es i for “I went”).
Periphrastic construction: Used for most other verbs, where you use “wnes i” + the verb-noun (e.g., Wnes i wylio for “I watched”).
However, aside from the SSiW course, I haven’t come across “Wyliest ti” being used anywhere else.
Is this short-form version of the verb in the past tense for “you watched” considered very colloquial, or is it just my bad Googling?
I don’t know where you did your online research, but every verb has a short form past-tense “built-in”. However, in speech you’ll mainly hear the short forms for the most common irregular verbs (such as mynd, dod and cael) and for monosyllabic everyday verbs like gweld and dweud.
But there is no difference in meaning between the two forms, and there is no set rule when to use which, so it is always a matter of speaker’s preference whether they want to use wyliest ti or wnest ti wylio.
Gweiadur.com has a section where you can look up verb conjugations, the “berfiadur”. The site is free, but you need to register.
On android phones there is the “Ap Geiriaduron” with a nifty feature: when you search for a word in a conjugated form, it will give you the root word (although it won’t say which form the entered word is). So if you enter “wyliest”, the app will show you “gwylio”. (It will also account for mutations, comparatives and plurals and all that which would usually make it harder to find a word in a printed dictionary)
I did the same; and it’s true that there were more hits for “wnest ti wylio”. However, almost all of them that I could see were language learning or teaching resources, which tells us what forms get taught, but doesn’t necessarily reflect actual usage - courses might tend to teach it in the context of “wnest ti + soft mutation + verb-noun is a good way to ask a question - try it with all these verbs” - and that seems to be borne out by the results I saw.
The thing is, that it’s quite a conversational thing to say, but might therefore be much more often said than written, and harder to find online. (If you’re into corpus linguistics, I believe there’s a corpus of modern spoken Welsh somewhere, but that does feel a bit heavy-duty to be honest!)
Short forms are apparently favoured more in the South than the North, but you have now (indirectly) earwormed me with a song by the indubitably Northern Y Cledrau, that has a line about “a bad channel that you have never watched” (sianel wael na wyliest ti erioed)!
Thank you Hendrik. I have checkled out those resources you suggested and they are very good.
I try not to overthink the grammer too much, but sometimes I cant move on until I really understand a thing. The principle of: “speak it, make mistakes, and move on”, is defitenly good advice, I can see how a person can bury their mind in unimportant (at first) details.
(I’ve only dabbled, but I do find corpora really interesting…)
I haven’t worked out how to search on collocations yet (so I can’t compare with the periphrastic), but a search on “wyliais” returned 6 results and “gwyliais” 20, which suggests that use of the inflected version isn’t terribly widespread… (In contrast, searching on “es” returned 1,719 entries.)
You are completely correct! Why would a mostly spoken version of a language, be written in many places? Lol ! When you put it like that, I feel a bit silly for having asked the question to begin with
And then there is literary Welsh . . . Maybe when I am doing my PHD
This is very interesting. From the kind of technical terms they use, it seems like there might be an intent to implement this into some kind of large language model.
This for example:
‘The CorCenCC dataset includes 14,338,149 tokens (circa 11.2-million-words). The data in CorCenCC represents a wide range of contexts, genres and topics. This data has, as far as possible, been anonymised using a combination of manual and automated techniques, and has been fully tagged in terms of part-of-speech (POS) and semantic categories. The POS and semantic tagging was carried out using CyTag and SemCyTag tools, available from CorCenCC’s GitHub website: CorCenCC · GitHub’
Or maybe not, I am no expert.
It would be cool if something like ChatGPT could actually speak everyday spoken Welsh, without making too many mistakes.
Hmmm… I just had a go (choose “advanced” on the radio buttons by the search box, and it’ll let you type in a phrase). I got 6 hits for “wyliais i”, plus 3 for “gwyliais i”, making only 9 for the short form – but I got only 4 in total for “wnes i wylio”, suggesting that in their data the short form is at least as current as the periphrastic one. (But also that there’s not a lot of data.) Probably, if I knew what I was doing, I could formulate some sort of wildcard query that would draw in more verbs for a more statistically meaningful result… but, alas, I don’t.
All of the different areas of research that are going on feed into each other (so this team, the team at Canolfan Bedwyr who are doing all the work around voice recognition in Welsh, and so on). At the moment, ChatGPT will operate in Welsh to a certain degree. The VR has now got to the point that it can cope if you talk slowly and clearly (it was demonstrated to us on Teams by one of the government’s team), but as soon as you speed up / go a bit too dialectical / start code-switching it can’t cope.
When you put it like that, I feel a bit silly for having asked the question to begin with 
