# Morphology alignment report

Generated by `pipeline.morphology`. Maps every Leeds Quranic Arabic Corpus word (mustafa0x fork of QAC v0.4) to a quranbench token id, comparing surface forms. Nothing is dropped silently: every divergence is listed here.

## Totals

- our tokens: **77881**
- Leeds words: **77429**
- aligned Leeds words: **77429** (100.0000%)
  - exact surface match: 77106
  - matched after normalisation (hamza/letter fold): 317
  - matched after extended fold (dagger-alef/maksura): 2
  - Leeds word spanning two of our whitespace tokens (merged): 4 words -> 8 tokens
- separated-basmala tokens annotated by copy from 1:1: **448**
- unresolved failures: **0**
- our tokens left unaligned: **0**

## Token count reconciliation vs Leeds

Our corpus counts 77,881 tokens including the 112 separated basmalas (112 × 4 = 448). Excluding them leaves **77,433**, which still includes al-Fatiha's basmala (its verse 1:1). Leeds/this fork counts **77,429** words. The difference of **4** is exactly four places where Tanzil's whitespace segmentation yields two tokens where Leeds keeps one word:

- `2:181:3` Leeds `بَعْدَمَا` = our `بَعْدَ` + `مَا`
- `8:6:4` Leeds `بَعْدَمَا` = our `بَعْدَ` + `مَا`
- `13:37:8` Leeds `بَعْدَمَا` = our `بَعْدَ` + `مَا`
- `37:130:3` Leeds `إِلْ يَاسِينَ` = our `إِلْ` + `يَاسِينَ`

## Tokens aligned only after the extended fold

The same word in each edition, differing only in how a medial long ā is seated (Leeds: alef-maksura + dagger alef; Tanzil: dagger alef alone). Aligned deliberately — a documented judgement call.

- `quran:tanzil-uthmani:12:39:1` Leeds `يَٰصَىٰحِبَىِ` vs Tanzil `يَٰصَٰحِبَىِ`
- `quran:tanzil-uthmani:12:41:1` Leeds `يَٰصَىٰحِبَىِ` vs Tanzil `يَٰصَٰحِبَىِ`

## Words with more than one root (judgement call)

A compound word whose segments carry different roots. The token's head `root` is taken from its first root-bearing stem.

- `quran:tanzil-uthmani:20:94:2` roots: أ م م, ب ن ي

## Roots

- distinct roots: **1651**
- tokens with a root: **50716**
- tokens with no root (particles, pronouns, proper nouns, muqaṭṭaʿāt): **27165** (34.9%)

### Top 10 roots by occurrence

| root | slug | occurrences |
| --- | --- | --- |
| أ ل ه | e-l-h | 2963 |
| ق و ل | q-w-l | 1722 |
| ك و ن | k-w-n | 1390 |
| ر ب ب | r-b-b | 980 |
| أ م ن | e-m-n | 879 |
| ع ل م | aa-l-m | 854 |
| ق و م | q-w-m | 660 |
| أ ي ي | e-y-y | 597 |
| ر ح م | r-hh-m | 563 |
| أ ت ي | e-t-y | 549 |