Skip to main content
NAWA classifies Arabic text into four major dialect groups with 100% accuracy, powered by HUMAIN’s ALLaM large language model - the most advanced Arabic AI model available.

​
Supported dialects

DialectCodeRegionsKey markers
GulfgulfSaudi Arabia, UAE, Kuwait, Qatar, Bahrain, Omanشلون، وايد، زين، حلو، يالله
EgyptianegyptianEgyptأوي، ده، دي، عايز، ازاي، كده
LevantinelevantineLebanon, Syria, Jordan, Palestineهلق، كتير، شو، هيك، منيح
MSAmsaPan-Arab (formal)إن، الذي، بالإضافة، يجب، لذلك
Dialect detection accuracy: 100% - validated against a 100-comment test corpus spanning all 4 dialects (Gulf, Egyptian, Levantine, MSA) during Phase 0.5 evaluation.

​
Example classifications

Comment: “وش رايكم بالمحتوى الجديد؟ أنا شايف إنه وايد حلو”Translation: “What do you think of the new content? I think it’s very nice”Markers: وش (what - Gulf), وايد (very - Gulf), حلو (nice - Gulf)
{
  "dialect": "gulf",
  "dialect_confidence": 0.97
}
Comment: “الفيديو ده جامد أوي، عايز تاني كده”Translation: “This video is amazing, I want more like this”Markers: ده (this - Egyptian), أوي (very - Egyptian), عايز (I want - Egyptian), كده (like this - Egyptian)
{
  "dialect": "egyptian",
  "dialect_confidence": 0.98
}
Comment: “كتير حلو الفيديو، بس شو القصة ورا الأغنية؟”Translation: “Very nice video, but what’s the story behind the song?”Markers: كتير (very - Levantine), شو (what - Levantine)
{
  "dialect": "levantine",
  "dialect_confidence": 0.95
}
Comment: “يجب أن نشجع هذا النوع من المحتوى الهادف”Translation: “We should encourage this type of meaningful content”Markers: يجب أن (must - MSA formal), هذا النوع من (this type of - MSA structure)
{
  "dialect": "msa",
  "dialect_confidence": 0.93
}

​
How it works

​
NAGL + ALLaM pipeline

NAWA uses a two-stage pipeline called NAGL (NAWA Augmented Generation Layer):
  1. Language detection: Identifies the input language. Arabic text is routed to the ALLaM model via HUMAIN’s API.
  2. Dialect classification: ALLaM analyzes morphological patterns, vocabulary, and syntactic structures to determine the dialect.
  3. Confidence scoring: A calibrated confidence score (0–1) indicates how certain the model is about the dialect classification.
ALLaM is developed by HUMAIN (formerly SDAIA) and is the most advanced large language model purpose-built for Arabic. NAWA is an official HUMAIN partner.

​
The dialect_confidence field

The dialect_confidence score ranges from 0 to 1:
RangeInterpretation
0.90–1.00High confidence - strong dialectal markers present
0.70–0.89Medium confidence - some dialectal features detected
0.50–0.69Low confidence - text may be code-switched or ambiguous
< 0.50Very low - text may be too short or use minimal dialectal features
Short comments (under 10 words) often have lower dialect confidence because there are fewer linguistic markers. For critical applications, consider filtering on dialect_confidence > 0.7.

​
Improving accuracy with feedback

If NAWA misclassifies a dialect, submit feedback to improve the model: 
curl -X POST https://api.trynawa.com/v1/feedback \
  -H "Authorization: Bearer nawa_test_sk_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "request_id": "req_abc123",
    "field": "dialect",
    "expected_value": "levantine",
    "comment": "This is Lebanese Arabic, uses هلق and كتير"
  }'
RLHF feedback is incorporated into model fine-tuning cycles, continuously improving accuracy across dialects.