According to this newspaper article from January the 3rd, a research team in Malta (Dr. Colin Layfield and Dr. John Abela ) is busy with research on the VMS. Concrete results are not announced yet. Maybe it is worth to keep an eye on this project.
Dr Colin Layfield and Dr John Abela are both senior lecturers in the Department of Computer Information Systems in the Faculty of Information and Communication Technology ( University of Malta ). Their research areas include Artificial Intelligence and Natural Language Processing.
To the article:
You are not allowed to view links.
Register or
Login to view.
edit: I found this older paper (May 2020):
You are not allowed to view links.
Register or
Login to view.
This is very interesting news to me, thanks for sharing bi3mw!
I do wish they had included Maltese in the comparison languages for that May 2020 paper.
As I understand it though, there are no surviving Maltese texts from before the 1485 Kantilena so perhaps they were only going with languages for which they had enough early 15th century material.
(07-01-2021, 12:24 AM)VViews Wrote: You are not allowed to view links. Register or Login to view.This is very interesting news to me, thanks for sharing bi3mw!
I do wish they had included Maltese in the comparison languages for that May 2020 paper.
As I understand it though, there are no surviving Maltese texts from before the 1485 Kantilena so perhaps they were only going with languages for which they had enough early 15th century material.
Unfortunately that can’t be the reason, as all the languages they used were “contemporary” (see p.77, 2nd col., 2nd para.). And given their location, a corpus of contemporary Maltese would seem as possible as it could be. But l’m still interested in what they produce, especially if they manage more age appropriate texts.
Oh I missed that! Quite the caveat indeed. I hope they can course correct in their future research. I can think of a few linguists on this forum they might want to get in touch with...
Their May 2020 paper was already discussed briefly You are not allowed to view links.
Register or
Login to view. .
(07-01-2021, 09:11 AM)ReneZ Wrote: You are not allowed to view links. Register or Login to view.Their May 2020 paper was already discussed briefly You are not allowed to view links. Register or Login to view. .
Sorry Rene, I have not seen that.
I have written Colin Layfield an email in which I asked how the status of current research is and whether in the foreseeable future a new paper will be published. Let's wait and see if there will be a response.
Out of fairness, let me add that I am reasonably aware of what this group is doing, but it is not up to me to report on that. Let's see what feedback you get.
I received feedback from Dr. Layfield today regarding my inquiry. There is definitely something in the works:
Quote:....
I do have several ongoing projects that I have students working on (3 MSc’s and 2 undergrad). These involve image processing tasks as well as additional analysis of the text and Stylometric analysis. I’ll say more about this when the research gets to the point we have something report.
....
So in any case, it is worthwhile to follow up on the projects.
I must admit I had to look up stylometric analysis (after having done so, I wonder if he had to look up the spelling? The fact that it's capitalised makes me suspect he copied and pasted it. And that's my stylometric analysis for today
)
Anyway, it seems like a hint for an interesting project. I look forwards to reading more about it.
The hyphotesis under they work are quite restrictive.
The main weakiness of this study is the method and the data used. They work under the hypothesis that the number of characters is that of EVA. If they are not, if one carachter is splitted in several ones (as a consecuence of different caligraphy for the same character) or several characters are united in a single one the results of the analisys will be wrong. The wider the differences, the more probable the divergence.
They havent taken in consideration the posibility of null characters, or letters or words represented in diffentent ways, they assume in fact that there arent any of them.
They compare what they call segments with word probability to find similar patters with another similar lenguages. I dont understand how they do it, are they taking the segments as words? As syllabes? As chains of letters?