German Word Formation and the Organization of the Mental Lexicon


  • Kimberly Brosche University of Vienna


Human language abilities depend on the mental lexicon (ML). However, it is still a matter of debate what exactly is listed in this collection of words and their building blocks [1]. Studies conducted for English [2], [3] suggest that not only affixes, but also affix combinations (-less-ness in rest-less- ness) are represented in the ML, since speakers do not need words or other semantic cues to judge whether an affix combination exists in the given language. The present study investigates whether the same is also true for a more inflectional (e.g. case endings) and composition friendly (long sequences of words are written together and used as a single word) Germanic language than English, namely German.

Methods: 32 native and 29 advanced non- native German speakers were visually presented with 60 suffix combinations without bases (e.g. -erschaft) via an online questionnaire. 30 of these combinations exist in standard German, while the other 30 are non-existing, the latter served as distractors. Of the existing combinations, 15 were productive and 15 unproductive, i.e. deriving less than 10 lemmata (words). Since we were interested in speakers’ intuitions about word formation, the task was to intuitively identify existing suffix combinations

Hypotheses: Based on the results for English [3], native and advanced non-native speakers were expected to perform on a par. Both groups should also recognize productive combinations more accurately than unproductive ones [3], thus supporting the idea that the former are memorized as such, whereas the latter are learned in whole words.

Results: On average, native speakers identified 80% of the existing and 74% of the non-existing combinations correctly (p=.05). Accuracy for productive combinations (90%) was significantly higher than for unproductive ones (70%), p=.02. Non-native speakers’ accuracy was around 71% for both existing and non- existing combinations, but the difference between productive (80%) and unproductive (63%) combinations was statistically significant, p=.01. Overall, natives’ (77%) accuracy was significantly above non-natives’ (71%), p=.01. Discussion: We confirm previous research for English (and other languages): suffix combinations are listed in the ML and productive suffix combinations are more easily recognizable than unproductive ones. However, unlike for English, German native speakers significantly outperformed advanced non-native speakers. Possible reasons, such as inadequate self- assessment (as “advanced”), should be addressed by future studies targeting the organization of the ML as an indicator of language proficiency.


[1] M. Taft, “The Nature of Lexical Representation in Visual Word Recognition,” in The Oxford Handbook of Reading, A. Pollatsek and R. Treiman, Eds. Oxford University Press, pp. 99–113, 2015.

[2] S. Manova, “Ordering restrictions between affixes,” in The Wiley Blackwell Companion to Morphology, Wiley Blackwell, 2021.
[3] S. Manova and G. Knell, “Two-suffix combinations in native and non-native English,” in All Things Morphology: Its independence and its interfaces, Moradi et al., Eds. Benjamins, pp. 305–323, 2021.