Mutual intelligibility (MI) among related languages is a gradient phenomenon shaped by lexical, grammatical, and phonetic-phonological similarity. This study proposes a neural language modeling approach to quantifying MI patterns within the Turkic language family. Using IPA-transcribed naturalistic text from six Turkic languages, we train character-level LSTM models on a source language and fine-tune them on target languages that vary in genealogical distance. Cross-lingual transfer is evaluated using character-level cross-entropy (CE) loss, Area Under the Curve (AUC), and Rate of Change (ROC), which together capture model generalization, learning dynamics, and early-stage adaptation. We further examine whether model performance is predicted by cophenetic distance, lexical similarity, weighted trigram frequency overlap, and differences in vowel harmony index. Overall, the results suggest that character-level language models can approximate MI gradients across Turkic languages: closely related pairs generally show lower CE loss and smaller AUC, while more distant pairs show greater early-stage change. Lexical similarity, local phonotactic overlap, and genealogical distance appear to be the most informative predictors of model convergence. These findings provide preliminary evidence that neural language models trained on naturalistic text can offer a scalable way to model MI patterns, including directional asymmetries, across closely related languages.

Speech errors are a natural part of communication, yet they rarely lead to complete communicative failure because both speakers and comprehenders can detect and correct errors. Although prior research has examined error monitoring and correction in production and comprehension separately, integrated investigation of both systems has been impeded by the scarcity of parallel data. In this study, we present SPACER, a parallel dataset that captures how naturalistic speech errors are corrected by both speakers and comprehenders. We focus on single-word substitution errors extracted from the Switchboard speech corpus, accompanied by speaker’s self-repairs and comprehenders’ responses from an offline text-editing experiment. Our exploratory analysis suggests asymmetries in error correction strategies: speakers are more likely to repair errors that introduce greater semantic and phonemic deviations, whereas comprehenders tend to correct errors that are phonemically similar to more plausible alternatives or do not fit into prior contexts. Our dataset enables future research on the integrated approach of language production and comprehension.

Whereas there is a growing literature that probes neural language models to assess the degree to which they have latently acquired grammatical knowledge, little if any research has investigated their acquisition of discourse modeling ability. We address this question by drawing on a rich psycholinguistic literature that has established how different contexts affect referential biases concerning who is likely to be referred to next. The results reveal that, for the most part, the prediction behavior of neural language models does not resemble that of human language users.