Discuss results - Irish and Celtic languages

Posted by: Clár on May 26, 2015, 18:36
Irish & Scottish comparison
Very interesting project!
There are a number of issues with the Irish language examples (Death/Marb, Four/Ceithre, Sun/Ghrian, Two/Do, Water/Dobhar). I can offer more specific feedback via email if they would be helpful!
Name: Clár
Posted by: Vincent on May 26, 2015, 21:16
Hi Clár,
Good evening,
Thank you for your reaction and proposal. I am very interested on feedbacks for Celtic languages and if anything is wrong with Irish, I would appreciate the correct version.
One thing is important: I need the right word and transcription but also which part of the word is the root - so I need to know which are the consonants of the mere lexical morphemes because they are the ones the system processes. For example the infinitive marks in Germanic languages "n", Slavic "t" and Romance "r" at the end of the verbs are not processed.
I will really appreciate since I had no feedbacks on Irish so far.
Kind regards
Posted by: Clár on May 27, 2015, 15:15
Hi Vincent,
I am happy to help. Please see my suggestions below.nbsp;
English English Irish Corrections Root
Death -D-TH- Death -M-R-B- Marb Death as an abstract concept = Bás Marbh (never 'marb') is an adjective meaning 'dead' or can be a noun as in 'the dead' Bás
Ear -R- Ear -K-L-S- Cluas Cluas
Eye -J- Eye -S-L- Súil Súl, maybe? The 'i' feels like a lenition to me because the adjective for 'eye' is be 'súl'
Four -F-R- Four -K-TH-R- Ceithre Four = A ceathair The root would be 'ceathair' or 'ceath' as the name of the number is 'a ceathair', the adjective is 'ceithre' and counting people is 'ceathrar'
Hand -H-N-D- Hand -L-M-H- Lámh Lámh
I -J- I -M-
Name -N-M- Name -N-M- Ainm Ainm is right but the similarity to English is a coincidence Ainm
Night -N-T- Night -KH- Oíche Oíche
Nose -N-S- Nose -S-R-N- Srón Srón
Sun -S-N- Sun -G-R-N- Ghrian Sun = Grian. The 'h' lenition (séimhiú) in 'an ghrian' only applies when used with the definite article 'an'. Grian
Three -TH-R- Three -T-R- Tríy Three = A trí Trí
Tongue -T-N-G- Tongue -T-N-G- Teangay Exact consonant match -T-/-T- Exact consonant match -N-/-N- Exact consonant match -G-/-G- Teanga/ 'Teang' as declensions and other forms include 'teangacha', 'theanga', 'teangaigh' etc.
Tooth -T-TH- Tooth -F-K-L- Fiacail I would feel like 'Fiacl' is the root somehow as plural is 'fiacla'
Two -T- Two -D- Do Two = A dó
Water -W-T-R- Water -D-B-R- Dobhar Water = Uisce I have only ever heard the word 'dobhar' to mean 'floods' or 'torrents' and is far from being a translation of 'water'. The Irish/Scots Gaelic 'uisce'/'uisge' incidentally is the etymological root of the English 'whiskey' from 'uisce beatha' Uisce
Who -W- Who -K-
Wind -W-N-D- Wind -G-TH- Gaoth Related consonant match -D-/-TH- Depending on how you are treating the 'h', the root could be 'gaot'
You (thou) -TH- Thou [1] -T- Related consonant match -TH-/-T-
Kind regards,
Posted by: Vincent on May 29, 2015, 11:52
Thank you very much Clara,
I really appreciate this. I will make the change this week end and send you the link when it is updated... I will probably bring the Celtic languages a bit closer together in the tree...
I have a question regarding "Marbh" - How do yo say "to die"?
"M-R","M-T" and "M-R-T" is found in almost all Indo-European languages in words which have to do with "death" so even English in "Murder", but also in Semitic languages (Arabic "Mut") and Polynesian "Mot". There is semantic shift between "Die", "Death", "Murder", etc but it seems the "M-T" stem is quite universal...
Thanks again
Have a nice weekend

Posted by: Seanán on Nov 24, 2016, 15:23
Irish & Breton comparison
Regarding Irish and Breton, they are closer than you indicate.
For Death, there is Marv but the word Marbh in Irish was not used. Bh in Irish has a V sound so Marv in Breton and Marbh in Irish have the same sound. These words are identical.
For Eye, Breton has Lagad and Irish has Súil. There is the word log in Irish wish means a hallow. Log súile mean the hollow of the eye in Irish. I think that Breton might be using a cognate with log in Irish.
The major difference between the Gaelic Celtic languages and the Brythonnic languages are that the Gaelic languages use a C where the Brythonnic languages use a P. So for the Breton Pevar for number 4, Irish has ceathair. If one changed pevar to cevar, the link between the numbers are clearer.
For Hand, Breton has Dorn and Irish has Lámh. However, the word dorn exists in Irish and it means a fist. Dorn means hand in different shapes. Therefore if you compared dorn in Breton and Irish, you would have an exact match.
For Night, Breton has Noz and Irish oíche. The word for tonight in Irish is anocht. The middle of the word is the same aNOCHt as noz. It would be better to use these words than oíche.
For tooth, Breton has Dant which is probably of French origin, dent. Irish has another word for tooth which is déad. Déad is related to dant, dent etc.
Regarding the word for water, all the Gaelic languages have two words for water. Irish has uisce and dobhar. Breton has Dour for water, so when Dour is compared with dobhar, it is clear that they are the same words. As mentioned above with Marv and Marbh, bh in Irish is pronounced like a V or W sound so dobhar in Irish is pronounced dower, very similar to Dour in Breton.
Another example not included is the Breton word for a man, Den, and the Irish word for a person (but which originally meant a man), duine. Den and duine are the same word.
I think that if you compare the words that I mentioned above, you will find that Irish and Breton are closer.
Posted by: Vincent on Nov 24, 2016, 18:10
Hello Seanán,
Thank you very much for these very interesting remarks. Most points you mention challenge the limits of my system: semantic shift, universal sound change rules vs. local ones, etc.
I had already noticed that the Celtic languages sometimes bring "frustrating" results in eLinguistics.net, connecting them to actually unrelated languages better than to some Indo-European ones, so there is much room for improvement and your remarks might fill the gap. When very few words are identified as cognates, we get a result with a higher probability to be due to chance than to actual relatedness.
I can (and I will) implement some of your remarks, but I have to keep some improvements off the system even if it is frustrating: as a completely computerized system from syllable to syllable comparison up to tree design, I have to use the same rules for all comparisons. So I can't apply special sound correspondences for some languages and not for other. This is very frustrating in many cases - like here with "C and P" and in many other cases like "B and M" in the Turkic languages -> if I "allow" the system to use these correspondences, I get better results locally but I increase the exposure to chance in all other cases.
Re. semantic shift "eye - hollow", "hand - fist", "man - person" - for now, I have to keep it as it is to respect the rules that I do not interfere in the results. I mean, if I start to extend the roots I use in some languages, I have to do it in all languages, otherwise I give some "local advantage".
However, the points you make here are very interesting and there are many other examples in other languages. I am considering using semantic clusters rather than single words, allowing the system to use several roots. Of course, it will also increase the exposure to chance, but I am sure that this will not be to a problematic level, and we can measure it anyway. The concept of "death" for instance seems to be "packed" in "-M-R-T-" (or "-M-R-" / "-M-T-") in most IE, Afro Asiatic and Autronesian languages - sometimes as "passive death", sometimes as "active death" ("Murder", which relates Germanic languages to this stem...)
Points I will change in the next days:
- BH - V correspondence in how the words are coded.
  - "Tooth" in Irish
- "Night" in Irish - i will include it with a footnote: as the root actually matters, I think it is OK to do so in this case. I have a similar issue with French "nose" - "Nez" which is pronounced "Né" but we also say "nasale" for the adjective and the unpronounced "z" is left from a time when it used to be pronounced.
- "Water" -> I will check it in all Celtic languages.

We will see the impact on the results but I think it will bring about 4 to 5 points to most Irish to IE languages comparisons and improve the Celtic classification.
I will contact you again when I will have changed it - probably by the middle of next week.
Thanks again for your time and invaluable contribution!
Kind regards
Posted by: Seanán on Nov 24, 2016, 19:58

Cher Vincent,
Super! Thanks for your comprehensive reply.
Here is a table of similar words in the Celtic languages http://www.omniglot.com/language/celtic/connections/index.php
You will see immediately the similarities. There is an astericks beside the Welsh word for a woman. Contemporary Welsh uses the word beginning with the letter M. The Irish word for a woman is bean, but the plural word is mná. Welsh has a similar word.
Here are some dictionary entries for words that I mentioned:
http://www.teanglann.ie/ga/fgb/dobhar - Dover on the south of England comes from this. It basically means a place beside water.
See more about dorn here http://www.omniglot.com/language/celtic/connections/body.htm#pb
I could go on but as you will see from the Omniglot pages, there a lot of similarities between the Celtic languages.
See the same word for house. In Irish it is teach (pronounced chock). It is related to the Spanish word for a roof, techo. In old Irish, teach was spelt tech. I guess in ancient times, roofs were very large and walls were very low on simple wooden houses so people talked about their house and roof as the same thing. As architecture changed over the centuries and walls got higher, for the Irish tech meant the whole structure, whereas for the Spanish, techo meant only the top of the house.
Posted by: Vincent on Nov 30, 2016, 21:14
Good evening Seanán,
I have made changes in the samples. It brings the Celtic languages even closer to each other and all of them closer to other IE languages, I will have a closer look at it when I have more time. I will also update the trees soon as I have included 30+ languages since the last update. I will inform you when the trees are updated.
Thanks again for your remarks.
Posted by: Seanán on Dec 1, 2016, 12:21

Good morning Vincent,
Thanks for your replies. I see that you have made the changes and that the Celtic languages are closer to each other however an unfortunate effect of these changes is that Scottish Gaelic and Irish are now farther from each other when really they are very close like perhaps Portuguese and Spanish or Czech and Slovak.
Is there anyway to rectify this? The original score for Irish and Scottish Gaelic was 10 whereas now it is 23. I think that it should be 10.
For example, the words Oíche / anocht in Irish are not seen to have any relation to Oidhche in Scottish Gaelic. This is crazy the original spelling of oíche in Irish was oidhche. The dh in Irish is like a "ee" sound so the dh was changed to í. Oíche and oidhche are the same. en.wiktionary.org/wiki/oidhche
The same goes for the words for tooth. They are the same in Irish and Scottish Gaelic. Fiacal in Irish and Fiacaill in Scottish Gaelic. As I wrote in my previous e-mail, there are two words in Irish for tooth - fiacal and déad. Fiacal is by far the most commonly used. (There is even an Irish toothpaste brand called Fiacla which just means teeth). See here c1.staticflickr.com/5/4067/4483256670_b612a13b6e_b.jpg
The same applies for the words for water. In Irish it is uisce and in Scottish Gaelic it uisge. The C and G are basically the same sound. Again like fiacal, uisce is by far the most commonly used term for water in Irish.
In fact Scottish Gaelic only uses uisge
I know it would be a pain the orifice, but could these terms be added to show the similarity and proximity of Irish and Scottish Gaelic? They are the languages that are closest to each other and Gaelic speakers in Ireland and Scotland hold each other in high regard as they speak variations of a common Gaelic language.
Perhaps there might be a way to show similarities between languages using more than one word for a term e.g. uisce/dobhar for water, lámh/dorn for hand, fiacal/déad for tooth. It would make the results more accurate.
By the way it is very interesting to learn about languages that I never heard of before such as Sariqoli.
Posted by: Vincent on Dec 3, 2016, 22:46
Hello Seanán,
Very interesting issue... I have changed again night, water and tooth. I think it is right to do so because we have to take the words/roots which are being used now. Using older roots too is certainly OK when studying a specific relationship, but for automated mass comparison it is certainly the right way to keep the same rule for all comparisons, taking the most common word if there is more than one available. Of course, this principle limits the system.
I have also reviewed the words and transcription in the other Celtic languages and now we have a clearer picture:
Brittonic languages:
Breton - Cornish: 13
Breton - Welsh: 19
Welsh - Cornish: 26
Goidelic languages:
Old Irish - Irish: 8
Old Irish - Scottish Gaelic: 10
Irish - Scottish Gaelic: 7
Gaulish is a "stand alone", distant from 30 to 37 to the Brittonic languages and 40 to 46 fo the Goidelic ones. Goidelic to Brittonic comparisons range from 47 to 58. I agree with you that 47-58 is too high. How much of Breton or Welsh can an Irish speaker understand?
Thanks again for your help.
Kind regards
Blog author: