Our research mostly focuses on how the properties of our social network (e.g., size, density) influence how good we are at understanding others, at processing information, and at learning and updating our representations . We examine these questions using a combination of individual differences, experimental, and computational methods and across different linguistic levels (phonological, lexical, semantic). We also aim at understanding how social network effects at the individual level cascade to effects at the community level and influence language evolution and change.
Check out our latest papers:
- Iacozza, S. , Meyer, A.S. & Lev-Ari, S. (in press) How in-group bias influences the level of detail of speaker-specific information encoded in novel lexical representations
Journal of Experimental Psychology: Learning, Memory, and Cognition
TLDR: We process and represent the speech of ingorup and outgroup members differently. The greater individuals’ implicit ingroup bias, the more they store speech from ingroup speakers in more detail than speech from outgroup speakers. We show this by looking at participants’ source memory for novel words they learn from (speakers presented as) ingroup and outgroup members.
- Raviv, L., Meyer, A.S., & Lev-Ari, S. (2019) Larger communities create more systematic languages. Proceedings of The Royal Society B 286, 20191262.
TLDR: By having small and large groups play a communication game in the lab using only words that they invent we find that larger communities create more structured languages, suggesting that cross-linguistic differences in grammatical and morphological complexity could be (partially) accounted for by community size. We also find that this effect is driven by the greater input variability.
- Lev-Ari, S., Dodsworth, R., Mielke, J. & Peperkamp, S. (2019) The different roles
of expectations in phonetic and lexical processing. In Proceedings of the 20th Annual Conference of the International Speech Communication Association (INTERSPEECH)
TLDR: Previous research investigates how the expectations listeners have of speakers influence language processing using either a lexical or a phonetic task, and the two are treated as interchangeable. We show that the two are not equivalent and have different underlying mechanisms that are sensitive to different individual differences. Performance on the phonetic task was sensitive to working memory whereas performance on the lexical task was sensitive to implicit bias.
- Raviv, L., Meyer, A.S., & Lev-Ari, S. (2019) Compositional structure can emerge without generational transmission. Cognition, 182, 151-164.
TLDR: An experimental study showing that structure can emerge within a language during the first generation of users, and that one of the main pressures for its emergence is having multiple interaction partners.
- Lev-Ari, S. (2019) People with larger social networks are better at predicting what someone will say but not how they will say it. Language, Cognition, and Neuroscience, 34,1, 101-114. doi:10.1080/23273798.2018.1508733
TLDR: People with larger social networks are better than those with smaller social networks at predicting the meaning that someone will express (e.g., car or bicycle) but not the form they would use (e.g., bicycle or bike).
Lev-Ari, S. (2018) The development of linguistic skills from a social network perspective. Keynote lecture at Socially Situated Language Processing, Berlin, Germany, September 3-4, 2018.
Iacozza, S., Meyer, A.S., & Lev-Ari, S. (2018) Novel lexical representations are shaped by speakers’ in-group status and learners’ in-group biases. Talk presented at Socially Situated Language Processing, Berlin, Germany, September 3-4, 2018.
Lev-Ari, S. (2018) People with smaller social networks are better at talker Identification. Poster presented at the 24th AMLaP conference, Berlin, Germany, September 6-8, 2018.