Gr ETEL 4 Jan Odijk LREC Miyazaki 2018

Gr. ETEL 4 Jan Odijk LREC Miyazaki 2018 -05 -10 1

Overview • Gr. ETEL 1, 2, 3 • Gr. ETEL 4 – Developers: Martijn van der Klis, Sheean Spoel, Gerson Foks (DH Lab) • Illustration 2

Gr. ETEL 1, 2, 3 • Gr. ETEL: KU Leuven – Cooperation CLARIN-NL and CLARIN Flanders • Gr. ETEL 2, 3: extensions, improvements in other Flemish projects • Application for searching in a treebank – Treebank = text corpus in which each sentence has been assigned a syntactic structure – Syntactic structure is usually a tree • Core feature: example based querying 3

Gr. ETEL 1, 2, 3 • Treebanks: – LASSY-Small (1 m tokens, written language) – CGN (1 m tokens, spoken language) – (V 3) So. Na. R Treebank (>500 m tokens) • V 1: http: //nederbooms. ccl. kuleuven. be/eng/gretel/ • V 2: http: //gretel. ccl. kuleuven. be/gretel-2. 0/ • V 3: http: //gretel. ccl. kuleuven. be/gretel 3/index. php 4

Gr. ETEL 4 • Gr. ETEL 4: UU Utrecht – In CLARIAH and UU-internal Ann. Cor project • New functionality that KU Leuven could not add: – Upload a user’s own corpus incl. metadata – Search in the user’s own automatically parsed corpus – Analysis of search results combined with metadata – Better support for Xpath Queries – Improved interface functionality • V 4 (alpha!) http: //gretel. hum. uu. nl/gretel 4/ 5

Illustration • Upload Corpus – Plain text or CHILDES CHAT – TEI and Fo. LIA to follow • CHAT Utterances are cleaned and metadata uploaded: – knor [!= pigsound], ik heb honger – knor, ik heb honger 6

Corpus Upload 7

Corpus Overview 8

Corpus Details 9

Query Example • Constructions with 3 bare verbs in the Dutch CHILDES Van Kampen Laura Corpus • Example sentence: – Hij zal dat willen doen 10

Example Sentence 11

Parse Tree 12

Select Parts 13

Query Tree 14

Select Treebank 15

Query //node[@cat and node[@pt="ww" and @rel="hd"] and node[@cat="inf" and @rel="vc" and node[@rel="hd" and @pt="ww"] and node[@rel="vc" and @cat="inf" and node[@pt="ww" and @rel="hd"]]]] 16

Example: Query Output 17

Utterance Details 18

Result Statistics 19

Analysis 20

Some Results • 3 verbs: – 335 hits found – 313 by adults, 12 by child – 4 by child do not occur among adults – 8 others are not in most frequent of adults – Child examples as of month 43 (3; 7) • 2 verbs: – 6, 645 in total, 1, 363 uttered by child – as of month 23 (1; 11). 21

Concluding remarks • Gr. ETEL is a very user-friendly search engine – Enables searching for constructions – Enables search for disambiguated words • Utrecht extensions – Enable searching in your own research corpus – Enable detailed analysis of search results 22

Concluding remarks • User-friendliness – Also implies limitations! • Automatic parsing – Is not flawless – Requires additional checks before conclusions can be reliably drawn • Try it out! http: //gretel. hum. uu. nl/gretel 4/index. php – Even if it is still under development 23

Thanks for your attention 24

More information • http: //portal. clarin. nl, http: //www. clariah. nl • • • Recorded lecture on Gr. ETEL: http: //lecturenet. uu. nl/Site 1/Catalog/Full/c 9 f 887 bc 45154 af 5 bd 7 cdb 218216816621 Educational Package: http: //dev. clarin. nl/sites/default/files/Educational. Module-v 4 b. pdf Augustinus, L, Vandeghinste, V, Schuurman, I and Van Eynde, F. 2017. Gr. ETEL: A Tool for Example-Based Treebank Mining. In: Odijk, J and van Hessen, A. (eds. ) 2017. CLARIN in the Low Countries, Pp. 269– 280. London: Ubiquity Press. DOI: https: //doi. org/10. 5334/bbi. 22 License: CC-BY 4. 0 Odijk, J. , van der Klis, M. , and Spoel, S. (2018). Extensions to the Gr. ETEL treebank query application. Proceedings of the 16 th International Workshop on Treebanks and Linguistic Theories (TLT 16) pp 46 -55, Prague. http: //aclweb. org/anthology/W/W 17 -7608. pdf Odijk & Van Hessen (eds. ) 2017. CLARIN in the Low Countries. London: Ubiquity Press. (Open Access). DOI: http: //dx. doi. org/10. 5334/bbi 25