Translation counts

Main analysis

The heat map below presents the number of senses with at least one translation to each target language in the main analysis. The post-processed query output used to create the heatmap is available at output/count_tr_sense_top_langs_pp.tsv.

Truncated counts

The table below provides the untruncated values for the counts truncated in the above heatmap.

Language	Target Language	Count (thousands) [a]
Bokmål	Nynorsk	33
Nynorsk	Bokmål	32
English	Italian	12
Italian	English	12
English	French	11
Bokmål	English	11

[a] N senses with >= 1 translation

Sensitivity analysis: count matches using reverse translation path

The source- by target-language combination with the greatest number of senses that have a 'translation' property path from the target language and not to the target language is English to Italian (n=219). Note that this does not imply the overall translation counts would increase by this amount if we modified the criteria to include paths from the target language because the target senses might be counted as translations through other properties already.

The post-processed file with these counts is the same file that has the main analysis results: output/count_tr_sense_top_langs_pp.tsv.

Sensitivity analysis: count matches from synonyms of source sense

The table below gives the source- by target-language combinations with the greatest increase in counts in the sensitivity analysis that also considers matches from synonyms of the source sense.

Language	Target Language	Count - Sensitivity Analysis	Count - Increase from Main Analysis
Japanese	English	9364	344
Japanese	German	4922	248
Japanese	Swedish	4218	239

The post-processed query output used to generate the table is available at output/count_tr_synsense_top_langs_pp.tsv.

Sensitivity analysis: count sense glosses as translations

The heat map below presents the number of senses with at least one translation to each target language in the sensitivity analysis that counts sense glosses as translations. The changes from the main analysis are best visualized by downloading both figures and toggling between them in a graphics program. The post-processed query output used to create the heatmap is available at output/count_tr_glsense_top_langs_pp.tsv.

We see in this figure compared to the main analysis, there are no longer source language rows near the top of the table with no translations. In addition, there are now 21 counts that exceed 10,000. In many cases, the increased counts are for 'English' as the target language (source languages for selected cells: Nynorsk, Persian, Egyptian, Akkadian, Dagbanli, Malayalam). The counts for Northern Sami also increased from negligible to above the truncation threshold for target languages of Bokmål, Nynorsk, and English. Aragonese now has many translations to Spanish and New Persian has many to Persian. However, for the case of New Persian, we should note that the language item for 'New Persian' (Q56356571) has no IETF language tag assigned, and we presume based on this result that the standard for this language is to write the glosses with a language tag of 'fa' which is the IETF language tag of the 'Persian' language item (Q9168). Besides 'New Persian', 'Hindustani' is the other language presented in the table that had no IETF language tag on its language item (Q11051).

Untruncated counts and English Wiktionary counts

The table below provides the untruncated values for the counts truncated in the above heatmap. The corresponding counts from English-language Wiktionary are also provided. Note that there are 69 other languages not listed in the table where the translation counts to English as the target language are more than 10,000 in the English Wiktionary.

Language	Target Language	Wikidata Count (thousands) [a]	enWK Count (thousands) [b]
Bokmål	English	50	30
Bokmål	Nynorsk	33	n/a
Nynorsk	English	33	28
Aragonese	Spanish	33	n/a
Nynorsk	Bokmål	32	n/a
French	English	30	120
Egyptian	English	18	8
Sumerian	English	17	2
Akkadian	English	16	1
Danish	English	15	34
German	English	15	118
Japanese	English	14	138
English	Italian	12	n/a
Italian	English	12	164
Malayalam	English	12	16
Dagbanli	English	11	0
English	French	11	n/a
Persian	English	11	25
Northern Sami	Bokmål	10	n/a
Northern Sami	Nynorsk	10	n/a
Northern Sami	English	10	5

enWK = English-language Wiktionary

[a] N senses with >= 1 translation

[b] For rows where English is the target language, this reports the number of gloss definitions on 4 October 2026 from https://en.wiktionary.org/wiki/Wiktionary:Statistics.

Resource use

The Wikidata lexicographical data dump file was 0.7G zipped (bz2) and 7.8G unzipped. The Apache Jena database after loading the lexicographical data was 35G. The data presented in the table are for queries run using the tdb2.tdbquery command except where the Fuseki server is indicated.

Analysis	Run time (minutes)	Memory use (G) [a]
Main analysis and sensitivity analysis using reverse translation path	58.5	9.9
Main analysis and sensitivity analysis using reverse translation path (Fuseki)	57.5	similar to above
Sensitivity analysis counting matches from synonyms of source sense	67.9	similar to above
Sensitivity analysis counting sense glosses as translations	72.4	similar to above

[a] From inspecting memory use while the program was running.