Recent work investigates whether LMs learn human-like linguistic generalizations and representations from developmentally plausible amounts of data. Yet, the basic linguistic units processed in these LMs are determined by subword-based tokenization, which limits their validity as models of learning at and below the word level. In this paper, we explore the potential of tokenization-free, phoneme- and grapheme-based language models. We demonstrate that small models based on the Llama architecture can achieve strong linguistic performance on standard syntactic and novel lexical/phonetic benchmarks when trained with character-level vocabularies. We further show that phoneme-based models almost match grapheme-based models in standard tasks and novel evaluations. Our findings suggest a promising direction for creating more linguistically plausible language models that are better suited for computational studies of language acquisition and processing.

We analyze the influence of utterance-level construction distributions in German child-directed/child-available speech on the resulting word-level, syntactic and semantic competence (and their underlying learning trajectories) in small LMs, which we train on a novel collection of developmentally plausible language data for German. We find that trajectories are surprisingly robust for markedly different distributions of constructions in the training data, which have little effect on final accuracies and almost no effect on global learning trajectories. While syntax learning benefits from more complex utterances, word-level learning culminates in better scores with more fragmentary utterances. We argue that LMs trained on developmentally plausible data can contribute to debates on how conducive different kinds of linguistic stimuli are to language learning.

We present grapheme-llama and phoneme-llama, character-based language models trained for the 2024 BabyLM challenge. Through these models, we explore an under-researched approach to downsizing: replacing subword-based tokenization with character-level tokenization, drastically reducing the vocabulary size. The grapheme model is trained on a standard BabyLM dataset, while the phoneme model uses a phoneme-converted version of this dataset. Results show that grapheme-based models perform better overall, achieving scores comparable to subword-based models on grammatical benchmarks. Despite lower performance, phoneme models also demonstrate promising grammatical learning. We argue that our results challenge conventional wisdom on language modeling techniques and open up novel research questions with character- and phoneme-based models as objects of inquiry.

Increasing efforts are put into gamification of experimentation software in psychology and educational applications and the development of serious games. Computer-based experiments with game-like features have been developed previously for research on cognitive skills, cognitive processing speed, working memory, attention, learning, problem solving, group behavior and other phenomena. It has been argued that computer game experiments are superior to traditional computerized experimentation methods in laboratory tasks in that they represent holistic, meaningful, and natural human activity. We present a novel experimental framework for forced choice categorization tasks or speech perception studies in the form of a computer game, based on the Unity Engine – the Gamified Discrimination Experiments engine (GDX). The setting is that of a first person shooter game with the narrative background of an alien invasion on earth. We demonstrate the utility of our game as a research tool with an application focusing on attention to fine phonetic detail in natural speech perception. The game-based framework is additionally compared against a traditional experimental setup in an auditory discrimination task. Applications of this novel game-based framework are multifarious within studies on all aspects of spoken language perception.