EXP-0004 — Root rank-frequency distribution
completed Deterministic computation
Question
What does the rank-frequency distribution of triliteral roots look like across the corpus?
Hypothesis
Descriptive baseline; no directional hypothesis. Provides the frequency table behind the rare-root graph threshold (D-028) and future lexical-trajectory work (Program C).
Counting policy
Root tags from morphology dataset quran-morphology-8f38b39; words without a root tag (particles, some pronouns) are excluded by construction.
Results — root rank-frequency
1651 distinct roots over 50,268 root-tagged words. Log-log slope -1.3958 (Ordinary least squares on log(frequency) vs log(rank), roots with frequency >= 2. Descriptive statistic only; no inferential claim.)
| Rank | Root | Words | Ayahs |
|---|---|---|---|
| 1 | أله | 2,851 | 1,879 |
| 2 | قول | 1,722 | 1,383 |
| 3 | كون | 1,390 | 1,176 |
| 4 | ربب | 980 | 871 |
| 5 | أمن | 879 | 723 |
| 6 | علم | 854 | 728 |
| 7 | قوم | 660 | 597 |
| 8 | أيي | 597 | 562 |
| 9 | أتي | 549 | 486 |
| 10 | كفر | 525 | 465 |
| 11 | بين | 523 | 454 |
| 12 | شيأ | 519 | 449 |
| 13 | رسل | 513 | 429 |
| 14 | يوم | 475 | 437 |
| 15 | أرض | 461 | 440 |
| 16 | سمو | 381 | 352 |
| 17 | كلل | 377 | 355 |
| 18 | عذب | 373 | 336 |
| 19 | عمل | 360 | 313 |
| 20 | جعل | 346 | 311 |
| 21 | رحم | 339 | 313 |
| 22 | أنس | 338 | 317 |
| 23 | رأي | 328 | 297 |
| 24 | كتب | 319 | 279 |
| 25 | هدي | 316 | 268 |
| 26 | ظلم | 315 | 290 |
| 27 | نفس | 298 | 270 |
| 28 | قبل | 294 | 282 |
| 29 | نزل | 293 | 257 |
| 30 | ذكر | 292 | 264 |
Reproducibility record
- Corpus
- quran:hafs-kufan:v1
- Method version
- root-frequency-v1
- Results sha256
- bf03c335105f31ad2f4fb83c48d512b1…
- Environment
- {"platform":"Linux-6.18.5-x86_64-with-glibc2.39","python":"3.11.15"}
- Ran
- 2026-07-31T02:31:56Z → 2026-07-31T02:31:56Z
Re-run with python3 services/research-worker/scripts/run_experiment.py data/research/specs/EXP-0004.json. Input checksums are recorded in the result file.