Signal in Noise: Exploring Meaning Encoded in Random Character Sequences with Character-Aware Language Models
This work provides insights into the intrinsic link between meaning and primitive information in language models, which could benefit researchers in NLP and cognitive science, though it is incremental as it builds on existing distributional hypothesis studies.
The researchers tackled the problem of understanding how meaning is encoded in language models by analyzing embeddings of random character sequences (garble), extant language, and pseudowords using CharacterBERT, and they identified an axis in the embedding space that separates these classes and relates to linguistic structures like part-of-speech and concreteness.
Natural language processing models learn word representations based on the distributional hypothesis, which asserts that word context (e.g., co-occurrence) correlates with meaning. We propose that $n$-grams composed of random character sequences, or $garble$, provide a novel context for studying word meaning both within and beyond extant language. In particular, randomly generated character $n$-grams lack meaning but contain primitive information based on the distribution of characters they contain. By studying the embeddings of a large corpus of garble, extant language, and pseudowords using CharacterBERT, we identify an axis in the model's high-dimensional embedding space that separates these classes of $n$-grams. Furthermore, we show that this axis relates to structure within extant language, including word part-of-speech, morphology, and concept concreteness. Thus, in contrast to studies that are mainly limited to extant language, our work reveals that meaning and primitive information are intrinsically linked.