Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Willkommens-Merkel, Chaos-Johnson, and Tore-Klose: Modeling the Evaluative Meaning of German Personal Name Compounds

Annerose Eichel, Tana Deeg, André Blessing, Milena Belosevic, Sabine Arndt-Lappe, Sabine Schulte im Walde

TL;DR

This study investigates whether German personal name compounds (PNCs) carry evaluative meaning beyond their base names by analyzing 321 PNCs and their full names in discourse. It combines valence-norm based context analysis and sentiment-PLMs to assess relative positivity/negativity, and enriches data with personal, domain-specific, and extralinguistic features. Domain differences emerge, with politics tending toward negative PNC evaluations and sports/show business tending toward positive ones, while modifiers can drive extreme valence shifts. A regression framework shows that incorporating extra-linguistic information alongside PNC valence yields high explained variance ($\text{Adj. }R^{2}$ up to $0.96$), highlighting how domain, party membership, and modifier valence shape evaluative interpretation in German PNCs and informing future discourse-analytic and NLP applications.

Abstract

We present a comprehensive computational study of the under-investigated phenomenon of personal name compounds (PNCs) in German such as Willkommens-Merkel ('Welcome-Merkel'). Prevalent in news, social media, and political discourse, PNCs are hypothesized to exhibit an evaluative function that is reflected in a more positive or negative perception as compared to the respective personal full name (such as Angela Merkel). We model 321 PNCs and their corresponding full names at discourse level, and show that PNCs bear an evaluative nature that can be captured through a variety of computational methods. Specifically, we assess through valence information whether a PNC is more positively or negatively evaluative than the person's name, by applying and comparing two approaches using (i) valence norms and (ii) pretrained language models (PLMs). We further enrich our data with personal, domain-specific, and extra-linguistic information and perform a range of regression analyses revealing that factors including compound and modifier valence, domain, and political party membership influence how a PNC is evaluated.

Willkommens-Merkel, Chaos-Johnson, and Tore-Klose: Modeling the Evaluative Meaning of German Personal Name Compounds

TL;DR

This study investigates whether German personal name compounds (PNCs) carry evaluative meaning beyond their base names by analyzing 321 PNCs and their full names in discourse. It combines valence-norm based context analysis and sentiment-PLMs to assess relative positivity/negativity, and enriches data with personal, domain-specific, and extralinguistic features. Domain differences emerge, with politics tending toward negative PNC evaluations and sports/show business tending toward positive ones, while modifiers can drive extreme valence shifts. A regression framework shows that incorporating extra-linguistic information alongside PNC valence yields high explained variance ( up to ), highlighting how domain, party membership, and modifier valence shape evaluative interpretation in German PNCs and informing future discourse-analytic and NLP applications.

Abstract

We present a comprehensive computational study of the under-investigated phenomenon of personal name compounds (PNCs) in German such as Willkommens-Merkel ('Welcome-Merkel'). Prevalent in news, social media, and political discourse, PNCs are hypothesized to exhibit an evaluative function that is reflected in a more positive or negative perception as compared to the respective personal full name (such as Angela Merkel). We model 321 PNCs and their corresponding full names at discourse level, and show that PNCs bear an evaluative nature that can be captured through a variety of computational methods. Specifically, we assess through valence information whether a PNC is more positively or negatively evaluative than the person's name, by applying and comparing two approaches using (i) valence norms and (ii) pretrained language models (PLMs). We further enrich our data with personal, domain-specific, and extra-linguistic information and perform a range of regression analyses revealing that factors including compound and modifier valence, domain, and political party membership influence how a PNC is evaluated.
Paper Structure (41 sections, 2 equations, 3 figures, 6 tables)

This paper contains 41 sections, 2 equations, 3 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Personal Name Compounds (PNCs)
  4. Approaches to Modelling PNCs
  5. Data
  6. Targets
  7. Personal Name Compounds (PNCs)
  8. Full Names
  9. Context Corpora
  10. Twitter
  11. Deutscher Wortschatz
  12. Evaluating PNCs via Valence
  13. Assessing Context-Level Valence
  14. Valence Norms
  15. Valence Exploration
  16. ...and 26 more sections

Figures (3)

  • Figure 1: Overview of name (blue triangles) vs. PNC (orange dots) valence with PNC frequency visualized by size of orange dots (minimum frequency = 5). One ore more PNCs can relate to one name, e.g., Willkommens-Merkel ('Welcome-Merkel') and Migranten-Merkel ('Migrant-Merkel') both referring to Angela Merkel and bearing a more negatively evaluative character than the name itself.
  • Figure 2: Domain-specific (politics: pol, sports, others) and cross-domain valence comparison for PNCs (p) and names (n). Green triangles and orange lines illustrate arithmetic mean and median values, respectively. Min. PNC freq. = 5.
  • Figure 3: Comparison of PNC valence at discourse level determined by humans (solid green line), valence norms (dashed orange line, x-markers), XLM-based PLMs (dashed pink and violet lines, triangle markers), and German BERT-based PLMs (dashed blue and cyan lines, rhombus markers).