Header

UZH-Logo

Maintenance Infos

Semantische Äquivalenz in Geburtserzählungen: Anwendung von Word Embeddings


Bubenhofer, Noah (2020). Semantische Äquivalenz in Geburtserzählungen: Anwendung von Word Embeddings. Zeitschrift für germanistische Linguistik, 48(3):562-589.

Abstract

The present study focuses on serially occurring narrations of ‘everyday’ life, more specifically on birthing as narrated by mothers on online forums; the underlying idea being that these narrations happen against the background of cultural narratives.The present paper uses word embedding models to detect typical topics and actors in these narrations. The calculation of word embeddings automatically constructs semantic spaces, where semantic relations (synonymy in particular) can be modeled. This method offers a way to think of synonymy as ‘functional equivalence in discourse’.The present study relies on previous work with n-grams (Bubenhofer, 2018). N-grams are sequences of words that often appear together; their sequential order in different narrations gives insight in narrative patterns. A further step in the analysis is the construction of ‘narrative topoi’, which is achieved through clustering neighboring vectors. The emerging clusters can in turn be grouped into five narrative elements of ‘telling a birth story’: 1) disruption of daily life, 2) personnel, 3) body, 4) fear, 5) joy. While it seems obvious that certain themes ‘belong’ into the narration of a delivery, it is less obvious with what vocabulary these themes are expressed.The presented method of clustering word-embedding-profiles adds tremendously to the modelling of a narrative. Its advantages lie in its potential to show lexical variation, as it also includes rare, non-conformative orthographical variants. Furthermore, it allows for a discourse-specific (and usage-based) view on semantic relations. The same applies to relations between semantic clusters. Seen from a discourse linguistics or cultural analysis perspective, word embeddings renew our understanding of semantics. This shows particularly fruitful if used to analyze (discourse dependent) derivations between semantic spaces.

Abstract

The present study focuses on serially occurring narrations of ‘everyday’ life, more specifically on birthing as narrated by mothers on online forums; the underlying idea being that these narrations happen against the background of cultural narratives.The present paper uses word embedding models to detect typical topics and actors in these narrations. The calculation of word embeddings automatically constructs semantic spaces, where semantic relations (synonymy in particular) can be modeled. This method offers a way to think of synonymy as ‘functional equivalence in discourse’.The present study relies on previous work with n-grams (Bubenhofer, 2018). N-grams are sequences of words that often appear together; their sequential order in different narrations gives insight in narrative patterns. A further step in the analysis is the construction of ‘narrative topoi’, which is achieved through clustering neighboring vectors. The emerging clusters can in turn be grouped into five narrative elements of ‘telling a birth story’: 1) disruption of daily life, 2) personnel, 3) body, 4) fear, 5) joy. While it seems obvious that certain themes ‘belong’ into the narration of a delivery, it is less obvious with what vocabulary these themes are expressed.The presented method of clustering word-embedding-profiles adds tremendously to the modelling of a narrative. Its advantages lie in its potential to show lexical variation, as it also includes rare, non-conformative orthographical variants. Furthermore, it allows for a discourse-specific (and usage-based) view on semantic relations. The same applies to relations between semantic clusters. Seen from a discourse linguistics or cultural analysis perspective, word embeddings renew our understanding of semantics. This shows particularly fruitful if used to analyze (discourse dependent) derivations between semantic spaces.

Statistics

Citations

Dimensions.ai Metrics
1 citation in Web of Science®
1 citation in Scopus®
Google Scholar™

Altmetrics

Downloads

96 downloads since deposited on 07 Jan 2021
57 downloads since 12 months
Detailed statistics

Additional indexing

Other titles:Semantic equivalence in birth stories: application of word embeddings
Item Type:Journal Article, refereed, original work
Communities & Collections:06 Faculty of Arts > Institute of German Studies
Dewey Decimal Classification:000 Computer science, knowledge & systems
410 Linguistics
Scopus Subject Areas:Social Sciences & Humanities > Language and Linguistics
Social Sciences & Humanities > Developmental and Educational Psychology
Social Sciences & Humanities > Linguistics and Language
Uncontrolled Keywords:Linguistics and Language, Developmental and Educational Psychology, Language and Linguistics
Language:German
Date:25 November 2020
Deposited On:07 Jan 2021 13:47
Last Modified:24 Apr 2024 01:50
Publisher:De Gruyter
ISSN:0301-3294
OA Status:Green
Publisher DOI:https://doi.org/10.1515/zgl-2020-2014
Related URLs:https://uzb.swisscovery.slsp.ch/permalink/41SLSP_UZB/rloemb/alma990009798220205508 (Library Catalogue)
  • Content: Published Version
  • Language: German