Cuneiform recorded typologically unrelated languages for more than three millennia, raising a natural question: which of its statistical regularities are properties of the shared writing system, and which belong to individual languages? We answer this with two probe pairs targeting the same script-versus-language decomposition. The first compares two unsupervised, dictionary-free segmentation methods: transitional probability (TP), grounded in infant statistical-learning research, and Morfessor. To our knowledge, this is the first systematic evaluation of dictionary-free segmentation on cuneiform; the only prior published baseline relies on a hand-built dictionary (Max-Match, F1 = 0.74). Both our methods reach F1 > 0.97 within language (Morfessor reaches F1 > 0.99 on Akkadian and Sumerian), exceeding MaxMatch by more than 23 F1 points, but diverge sharply in cross-language transfer: TP holds F1 > 0.95 across all six language pairs, while Morfessor collapses to F1 = 0.27 from the 13K-token Elamite source. A learning curve and a Hittite stress test (34.7% foreign-script logograms, where TP drops 35 points and Morfessor only 8) further isolate the two underlying mechanisms. The second probe pair extends the same decomposition to representations: concatenating Unicode and Latin features improves classification in 11 of 12 morphosyntactic tasks (p < 0.01). We release CUNEI-TOOLS and a three-language benchmark.
Figures & tables
AKK
SUX
ELX
HIT
Tokens
1,255,669
146,143
13,383
170,496
Texts
13,720
394
85
7,099
Vocabulary
123,001
31,906
13,189
—
Lemmas
28,953
3,765
748
—
POS classes
24
14
7
—
Entity types
11
8
3
—
Table 1: Dataset statistics. AKK = Akkadian, SUX = Sumerian, ELX = Elamite, HIT = Hittite Yavasan and Gordin (2024) . The ELX token count comprises 13K lemma-dictionary entries plus the Utu-našu running corpus; segmentation experiments train on running text only (§ 3 ), and an expanded 871-document Elamite corpus is analyzed in Appendix A.14 . Hittite serves exclusively as the substrate-violation stress test (§ 4.1.3 ), so annotation-dependent rows do not apply.
Method
Lang
F1
P
R
TP
AKK
.971
.946
.997
SUX
.972
.946
.999
ELX
.989
.993
.984
Morfessor
AKK
.997
.997
.996
SUX
.993
.997
.990
ELX
.985
.999
.971
Table 2: Word boundary inference (5-fold held-out CV). Both dictionary-free methods exceed the prior dictionary-based MaxMatch baseline Homburg and Chiarcos (2016) (F1 = 0.74 on Middle Babylonian) by more than 23 F1 points. Morfessor’s slight in-language edge does not generalize to cross-language transfer (Table 3 ).
TP (test F1)
Morfessor (test F1)
Train
AKK
SUX
ELX
AKK
SUX
ELX
Train tokens
AKK
.971 †
.967
.992
.997 †
.981
.984
1.26M
SUX
.960
.972 †
.993
.859
.995 †
.953
146K
ELX
.959
.969
.989 †
.304
.267
.985 †
13K
Table 3: Cross-language word boundary transfer (F1). † = in-language (held-out CV). Off-diagonal cells = zero-shot transfer with no test-language hyperparameter tuning. Bold = > 10 F1 point gap from TP. Train tokens are Table 1 word counts; in segmentation-stream units the corpora are 2.44M / 338K / 8.8K (ELX segmentation trains on the Utu-našu running text only, § 3 ). 95% bootstrap confidence intervals (1,000 document resamples) are within ± 0.006 across all transfer cells; the cross-language dissociation between TP and Morfessor (65+ F1 points at ELX-as-source) is therefore well outside any reasonable noise band. Visualized in Figure 1 .
Morfessor source
→ AKK
→ ELX
SUX @ 13K stream tokens
.595 ± .031
.710 ± .091
SUX @ 8,831 (ELX budget)
.542 ± .051
.637 ± .077
SUX full (Table 3 )
.857
.953
ELX full (Table 3 )
.304
—
Table 4: Matched-size transfer control. Shrinking the Sumerian source to the Elamite budget costs 32 points ( → AKK), yet at the same budget the Elamite source sits a further 24 points lower: size and composition both matter, and neither alone explains the collapse.
Figure 1: Cross-language transfer F1 by source language and method. The dissociation is corpus-size mediated: TP (blue) maintains F1 > 0.95 across all source corpora, while Morfessor (orange) tracks TP at the largest source (AKK, 1.26M tokens), trails by approximately 10 points at the medium source (SUX, 146K tokens), and collapses to below 0.31 at the smallest source (ELX, 13K tokens). The pattern is exactly what would be expected if TP requires only stable bigram statistics (which converge from approximately 104 tokens) while Morfessor requires a lexicon that covers the target corpus; coverage, not language identity per se, predicts its transfer (§ 4.1.1 ).
Figure 2: In-language learning curve for TP (blue) and Morfessor (orange) on Akkadian, evaluated on a fixed held-out test set. TP is essentially flat at F1 ≈ 0.96 across all training-corpus sizes from 1K to 1.26M tokens, confirming that the script-level substrate property is recoverable from minimal data. Morfessor starts at F1 = 0.67 at 1K tokens, climbs to F1 ≈ 0.92 at 5K, and reaches TP’s level by approximately 10K tokens; above the crossover, Morfessor slightly exceeds TP. The in-language curve isolates property-recoverability (the focus of this experiment) from cross-language vocabulary overlap (the focus of Table 3 ); together they account for the full cross-language dissociation.
Task
Repr.
AKK
SUX
ELX
Unified
Latin
.693
.847
.636
Unicode
.621
.750
.659 ∗∗
Concat
.712 ∗∗∗
.870 ∗∗∗
.666 ∗∗∗
Table 5: POS classification (F1-macro, n-gram LR). Significance markers are paired-bootstrap p -values comparing Concat against the better of Latin and Unicode within each column: ∗∗∗p<0.01 ; ∗∗p<0.02 . Concatenation wins across all three languages. Separated grammatical-vs-entity results in Appendix A.2 .
Appendix figures & tables13 assets
Supplementary material from the paper’s appendix.
Appendix
Task
Repr.
AKK
SUX
ELX
Gram.
Latin
.669
.872
.615
Unicode
.657 ∗
.832
.634 ∗∗
Concat
.682 ∗∗∗
.888 ∗∗∗
.642 ∗∗∗
Ent. det.
Latin
.937
.947
.873
Unicode
.839
.896
.876 ns
Concat
.939 ∗∗∗
.954 ∗∗∗
.882 ∗∗∗
Appendix
Table 6: Separated grammatical POS, entity detection, and entity typing (F1-macro). Significance markers are paired-bootstrap p -values comparing Concat against the better of Latin and Unicode within each column: ∗∗∗p<0.01 ; ∗∗p<0.02 ; ∗p<0.05 . Concatenation wins all 9 comparisons; 8 are significant.
Task
Lang
Best
Cat.
Δ
Uni.
AKK
.373
.412
+.039
SUX
.428
.451
+.023
ELX
.471
.489
+.018
Gram.
AKK
.516
.513
− .003
SUX
.630
.654
+.024
ELX
.465
.469
+.004
Appendix
Table 7: Character Transformer: best single representation vs. concatenation. Concat wins 5/6 Transformer comparisons; combined with logistic regression (Table 5 ): concat wins 17/18 across both architectures.
Figure 3: Unicode feature weight share by POS category. Elamite is Unicode-dominant; Akkadian and Sumerian show balanced contributions for open-class words but Latin dominance for entities and closed-class items.
Both
Both
Latin
Uni.
correct
wrong
only
only
AKK
63.1%
18.5%
12.4%
6.0%
SUX
85.2%
5.2%
7.5%
2.1%
ELX
66.9%
20.2%
5.5%
7.3%
Appendix
Table 8: Error outcome distribution. “Latin only” and “Unicode only” columns represent the complementary pool (9.6–18.4%).
Rare ( ≤ 26)
Frequent ( > 26)
AKK
28.9%
19.1%
SUX
16.3%
8.5%
Appendix
Table 9: Percentage of tokens correctly classified by exactly one representation, split by token frequency.
Metric
Repr.
AKK
SUX
ELX
Silhouette
Latin
− .100
− .071
− .037
Unicode
− .093
− .147
− .063
Morph. coh.
Latin
.764
.791
.639
Unicode
.796
.839
.644
Graph agr.
Latin
.645
.741
.460
Unicode
.512
.638
.452
Appendix
Table 10: Embedding metrics (cosine similarity).
Task
Repr.
AKK
SUX
ELX
Unified
Latin
.571
.671
.564
Unicode
.334
.422
.543
Concat
.548
.650
.571
Gram.
Latin
.542
.728
.529
Unicode
.401
.508
.516
Concat
.529
.705
.534
Appendix
Table 11: ByT5-small POS classification (F1-macro). Latin strongly outperforms Unicode, consistent with Latin-script pretraining bias.
θ
AKK
SUX
ELX
0.20
.963
.957
.591
0.30
.967
.966
.685
0.50
.968
.971
.897
0.70
.971
.971
.954
0.90
.961
.956
.984
0.95
.949
.943
.989
Appendix
Table 12: TP word boundary F1 at selected thresholds.
corpusweight
ELX
SUX
AKK
0.25
.998
.994
.996
0.5
.999
.994
.997
1.0 (paper)
.985
.993
.996
2.0
.063
.076
.122
4.0
.057
.065
.065
Appendix
Table 13: In-language 5-fold CV F1 by corpusweight. Bold = source-side selected value.
Src
Tgt
cw=1.0
cw=0.5
TP
AKK
SUX
.981
.982
.967
AKK
ELX
.984
.997
.992
SUX
AKK
.857
.977
.960
SUX
ELX
.953
.993
.993
ELX
AKK
.304
.765
.959
ELX
SUX
.267
.713
.969
Appendix
Table 14: Zero-shot Morfessor transfer at the published default vs. the source-selected corpusweight, against untuned TP. Bold = TP’s advantage from the smallest source survives tuning.
Source
Size
→ AKK
→ SUX
→ ELX
AKK
5K
—
.615
.826
AKK
13K
—
.743
.955
AKK
25K
—
.826
.973
AKK
full
—
.981
.984
SUX
5K
.482
—
.554
SUX
13K
.595
—
.710
Appendix
Table 15: Morfessor zero-shot transfer across the source-size grid (mean over 5 seeds; seed std ≤.091 , typically ≤.03 ). ELX-exp = the 871-document expanded corpus (Appendix A.14 ).
Training source
TP
Morf.
Tokens
Dictionary only
.994
.727
68,251
Utu-našu only (paper)
.989
.985
6,759
Combined
.995
.991
75,010
Appendix
Table 16: Elamite segmentation by training source, evaluated on identical held-out running-text folds.
Tokens
BiLSTM (sup.)
TP
Morf.
1,000
.957
.961
.637
5,000
.957
.964
.908
13,000
.962
.968
.963
50,000
.994
.968
.991
Appendix
Table 17: Supervised vs. unsupervised boundary detection on identical training subsamples (seed std ≤.063 ; ≤.01 except Morfessor at 1K). For calibration, the all-boundaries trivial baseline reaches F1 =0.961 on this boundary-dense test set; the substantive comparisons are therefore precision-driven and lie above that floor.
School of CS and Engineering, Vellore Institute of Technology, Vellore, India · Technology Innovation Hub, Indian Institute of Technology, India · College of Technology and Engineering, WestCliff University, USA +2