[HN Gopher] Why We Need Arabic Language Models
       ___________________________________________________________________
        
       Why We Need Arabic Language Models
        
       Author : thinkingemote
       Score  : 24 points
       Date   : 2025-10-19 19:50 UTC (3 hours ago)
        
 (HTM) web link (www.natureasia.com)
 (TXT) w3m dump (www.natureasia.com)
        
       | sarabande wrote:
       | Does anyone know if they published the dataset?
        
       | nakamoto_damacy wrote:
       | I wonder of you pre train on Hebrew and Arabic if it will find
       | the similarities between the RTL writing direction. So many
       | similar words. I guess both came from Aramaic? If so, how about
       | the trifecta of ancient languages with Aramaic then Hebrew the
       | Arabic.
        
         | ch4s3 wrote:
         | They don't come from Aramaic, Arabic is a Southwestern Semitic
         | language and Aramaic and Hebrew are Northwestern Semitic
         | languages. Aramaic and Hebrew tree are sort of cousins with
         | Hebrew splitting off from southern Canaanite which was sort of
         | a siblings language with an older form of Aramaic.
        
           | nakamoto_damacy wrote:
           | They = Hebrew, Aramaic and Arabic
           | 
           | ---
           | 
           | ## 1. "They all kept the triconsonantal root system -- where
           | word meaning is based on three core consonants (like K-T-B =
           | "write" - Hebrew katav, Arabic kataba, Aramaic ktav)."
           | 
           | *Source evidence:*
           | 
           | * The article "Triliteral Roots / Consonantal Roots" states
           | that many Semitic languages (including Arabic, Hebrew) have
           | roots typically made of three consonants (triliteral) and
           | that words are formed by inserting vowels, etc. ([Transparent
           | Blogs][1]) * A source says: "Both Hebrew and Arabic rely on a
           | triliteral root system, meaning words are formed from three
           | core consonants. Example of the root K-T-B..." ([Biblical
           | Hebrew][2]) * Another general description: "The roots of
           | verbs and most nouns in the Semitic languages are
           | characterized as a sequence of consonants ... such abstract
           | consonantal roots are used..." ([Wikipedia][3]) So this claim
           | is well supported.
           | 
           | *Arabic translation of the claim:*
           | 
           | > HtfZt jmy`h bnZm ljdhr lthlthy lHrwf -- Hyth y`tmd m`n~
           | lklm@ `l~ thlth@ Hrwf Smt@ 'ssy@ (mthl k-t-b = "ktb/yktb" -
           | l`bry@ ktb (katav), l`rby@ ktb (kataba), larmy@ ktab (ktav)).
           | 
           | *Hebrew translation of the claim:*
           | 
           | > kvln shmrv `l shytt hshvrsh htlt-`bry -- shbh mshm`vt hmylh
           | mbvsst `l shlvshh `ytsvrym l(lmshl ktb = "ktb" - `bryt ktb (
           | katav ), `rbyt ktb ( kataba ), rmyt ktb ( ktav )).
           | 
           | *Citations (for this claim):*
           | 
           | * Semitic linguistics: "The roots of verbs and most nouns in
           | the Semitic languages are characterized as a sequence of
           | consonants ..." ([Wikipedia][3])
           | 
           | * "Both Hebrew and Arabic rely on a triliteral root system,
           | meaning words are formed from three core consonants."
           | ([Biblical Hebrew][2])
           | 
           | * Description of the K-T-B root being used in both Arabic and
           | Hebrew. ([Wikipedia][4])
           | 
           | ---
           | 
           | ## 2. "They share similar grammar and sound systems, just
           | evolved differently."
           | 
           | *Source evidence:*
           | 
           | * A blog post on Duolingo says: "Because Arabic and Hebrew
           | are part of the same large language family, their grammars
           | often 'work' in similar ways." ([Duolingo Blog][5]) * A site
           | "Arabic and Hebrew Compared" states: "Arabic and Hebrew
           | morphology ... is based on the consonant root system. ..."
           | ([Google Sites][6]) * The Wikipedia article on Semitic
           | languages states that the Semitic languages share many
           | grammatical features (word order, non-concatenative
           | morphology, etc.) ([Wikipedia][7]) So yes, there is support
           | for similar grammar and sound (phonological) systems.
           | 
           | *Arabic translation of the claim:*
           | 
           | > nhm tshtrkn fy nHw wSwtyt mtshbh@, rq tTwrt bshkl mkhtlf.
           | 
           | *Hebrew translation of the claim:*
           | 
           | > hn khvlqvt dqdvq vm`rkvt tslylym dvmvt, rq shhtptkhv bvpn
           | shvnh.
           | 
           | *Citations (for this claim):*
           | 
           | * "Because Arabic and Hebrew ... their grammars often 'work'
           | in similar ways." ([Duolingo Blog][5]) * "Arabic and Hebrew
           | morphology ... is based on the consonant root system."
           | ([Google Sites][6]) * "Semitic languages share a number of
           | grammatical features ..." ([Wikipedia][7])
           | 
           | ---
           | 
           | ## 3. "Many religious and cultural interactions over
           | millennia reinforced overlap (borrowed or re-borrowed
           | vocabulary)."
           | 
           | *Source evidence:*
           | 
           | * The article "Similarities Between Hebrew and Arabic"
           | mentions: "Many Hebrew and Arabic words are cognates,
           | retaining similar meanings and sounds." ([Biblical
           | Hebrew][2]) * A blog "Halal, Hillul, and the Shared Meanings
           | of Hebrew and Arabic" discusses relationships between similar
           | sounding words (cognates) due to shared roots. ([Hebrew
           | College][8]) * Comparative grammar sources mention that
           | because Hebrew, Arabic and Aramaic are closely related, there
           | has been lexical borrowing and shared vocabulary.
           | ([semiticroots.net][9]) So your statement about
           | religious/cultural interaction reinforcing overlap
           | (vocabulary) is broadly supported.
           | 
           | *Arabic translation of the claim:*
           | 
           | > l`dyd mn ltf`lt ldyny@ wlthqfy@ `br l'lfyt `zWazt ltdkhl
           | (st`rt 'w '`dt st`r@ mfrdt).
           | 
           | *Hebrew translation of the claim:*
           | 
           | > ynspvr yntrqtsyvt dtyvt vtrbvtyvt lvrk lpy hshnym khyzqv t
           | hhshtlbvt (hshlh v hshlh mkhdsh shl vtsr mylym).
           | 
           | *Citations (for this claim):*
           | 
           | * "Many Hebrew and Arabic words are cognates ..." ([Biblical
           | Hebrew][2]) * "The relationships between similar-sounding
           | words ... in the case of the Semitic languages, similar
           | roots." ([Hebrew College][8]) * "Hebrew, Arabic, and Aramaic
           | ... than between Hebrew and any other language ..."
           | ([semiticroots.net][9])
           | 
           | ---
           | 
           | [1]: https://blogs.transparent.com/hebrew/hebrew-grammar-
           | consonan... "Hebrew Grammar: Consonantal Roots - Transparent
           | Language Blog" [2]: https://biblicalhebrew.org/similarities-
           | between-hebrew-and-a... "Similarities Between Hebrew and
           | Arabic" [3]: https://en.wikipedia.org/wiki/Semitic_root?utm_s
           | ource=chatgp... "Semitic root - Wikipedia" [4]:
           | https://en.wikipedia.org/wiki/K-T-B?utm_source=chatgpt.com
           | "K-T-B" [5]: https://blog.duolingo.com/are-arabic-hebrew-
           | persian-related/... "Dear Duolingo: Are Arabic, Hebrew, and
           | Persian related?" [6]:
           | https://sites.google.com/site/mopclanguages/arabic-and-
           | hebre... "MOPC Languages - Arabic and Hebrew Compared" [7]: h
           | ttps://en.wikipedia.org/wiki/Semitic_languages?utm_source=c..
           | . "Semitic languages" [8]:
           | https://hebrewcollege.edu/blog/halal-hillul-and-the-
           | shared-m... "Halal, Hillul, and the Shared Meanings of Hebrew
           | and Arabic" [9]: https://www.semiticroots.net/downloads/Compa
           | rative%20Grammar... "Comparative Grammar of the Semitic
           | Languages"
        
       | binarymax wrote:
       | Do we need language specific LLMs? I can't vouch for the data
       | coverage or accuracy of Arabic in the leading models today, but I
       | do know them to be highly cross-lingual capable.
        
         | readthenotes1 wrote:
         | Makes me wonder who did the translation
         | 
         | "This is a translation of the Arabic article published on 3rd
         | August 2025"
         | 
         | Full irony would be from an LLM
        
           | tokai wrote:
           | Such declarations have become pretty useless without any
           | indicator of the translation method.
        
       | nzeid wrote:
       | > Clear examples emerge when global language models address
       | culturally sensitive issues, such as social relationships or
       | political debates. They often adopt ambiguous positions that
       | overlook the Arab cultural context, creating a gap between these
       | digital tools and the values and lived experiences of Arab users.
       | 
       | Well I have bad news, my friend. English language models are also
       | terrible at this.
       | 
       | This whole article seems to stem from the premise that it's
       | important for LLMs to engage cultural issues competently. But...
       | should they even?
        
         | Fade_Dance wrote:
         | >But... should they even?
         | 
         | I don't see why not?
         | 
         | Also, while I don't have access to this perspective myself, I'd
         | imagine this is an unending annoyance in many areas of the
         | world, since they are consuming often quite America-centric
         | offerings where localization is an after-thought and contracted
         | out.
        
       ___________________________________________________________________
       (page generated 2025-10-19 23:01 UTC)