Hebrew is one of the many languages whose text corpora are included in Sketch Engine, a tool for discovering how language works. Sketch Engine is designed for linguists, lexicologists, lexicographers, researchers, translators, terminologists, teachers and students working with Hebrew to easily discover what is typical and frequent in the language and to notice phenomena which would go unnoticed without a large sample of Hebrew text.
Sketch Engine has tools to identify and analyse collocations, synonyms and antonyms, examples of use in context, keywords or terms. Frequency word lists of Hebrew single-word or multi-word expressions of various types can be generated. Even users without any technical knowledge can create their own Hebrew corpus using the Sketch Engine's intuitive built-in tool.
Tools to work with Hebrew text corpora
To work with the Hebrew language, Sketch Engine offers the following tools:
Hebrew concordance
The concordancer included in Sketch Engine can be used to display a list of examples (called concordance) of the search word or phrase as it appears in Hebrew language text corpora. The search will display the keyword with some context to the right and context to the left of the keyword (KWIC concordance). more»
Hebrew word lists
The word list feature will generate a frequency list of all words that appear in a text or corpus. A very large corpus can be used to generate a list of all words that exist in Hebrew or all words that start, contain or end with specific characters. Advanced options can be used to generate lists of grammatical categories or parts of speech used in a corpus together with their frequencies. more»
N-grams in Hebrew
Generating a list of N-grams contained in a text makes it possible to identify and study patterns and notice phenomena related to multi-word units (MWU) in Hebrew that cannot be detected by other tools. more»
List of available Hebrew corpora
- trial – available to both trial users as well as paying subscribers
- main – only available to paying subscribers
- on demand – access to the corpus is subject to specific terms, click the corpus name for details
| Corpus | Access policy | Size in words |
|---|---|---|
| CHILDES Hebrew Corpus | main | 807,657 |
| ELEXIS Hebrew Web 2021 | main | 1,043,504,840 |
| ELEXIS Hebrew Web 2021 (heTenTen21) WSD sample | main | 2,017,821 |
| Gutenberg Hebrew 2020 | main | 158,212 |
| Hebrew Drama Corpus | main | 954,359 |
| Hebrew General Corpus (web crawled, mostly newspapers) | main | 157,947,728 |
| Hebrew Translation Corpus | trial | 1,180,003 |
| Hebrew Web (HebWaC) | main | 47,832,254 |
| Hebrew Web 2014 (heTenTen14, Meni/Alon tagged + lempos) | ondemand | 895,876,116 |
| Hebrew Web 2014 (heTenTen14, no POS tagging) | main | 890,282,843 |
| Hebrew Web 2021 (heTenTen21) | trial | 2,775,686,699 |
| Open Parallel Corpus (OPUS) – Hebrew | main | 130,972,343 |
| OpenSubtitles 2018 parallel – Hebrew | main | 371,473,205 |
| Timestamped JSI web corpus 2014-2016 Hebrew | trial | 111,339,363 |
| Timestamped JSI web corpus 2014-2021 Hebrew | main | 466,851,576 |
| Timestamped JSI web corpus 2021-03 Hebrew | main | 8,450,710 |
| Timestamped JSI web corpus 2021-04 Hebrew | main | 6,544,178 |
