Self-supervised objectives have driven major advances in NLP by leveraging large-scale unlabeled data, but such resources are scarce for many of the world's languages. Surprisingly, they have not been explored much for character-level tasks, where smaller amounts of data have the potential to be beneficial. We investigate the effectiveness of self-supervised auxiliary tasks for morphological inflection -- a character-level task highly relevant for language documentation -- in extremely low-resource settings, training encoder-decoder transformers for 19 languages and 13 auxiliary objectives. Autoencoding yields the best performance when unlabeled data is very limited, while character masked language modeling (CMLM) becomes more effective as data availability increases. Though objectives with stronger inductive biases influence model predictions intuitively, they rarely outperform standard CMLM. However, sampling masks based on known morpheme boundaries consistently improves performance, highlighting a promising direction for low-resource morphological modeling.

Large language models, particularly multilingual ones, are designed, claimed, and expected to cater to native speakers of varied languages. We hypothesise that the current practices of fine-tuning and evaluating these models may not perfectly align with this objective owing to a heavy reliance on translation, which can introduce translation artefacts and defects. It remains unknown whether the nature of the instruction data has an impact on the model output; conversely, it is questionable whether translated test sets can capture such nuances. Due to the often coupled practices of using translated data in both stages, such imperfections could have been overlooked. This work investigates these issues using controlled native or translated data during instruction tuning and evaluation stages. Experiments on eight base models and eight different benchmarks show that native or generation benchmarks reveal a notable difference between native and translated instruction data especially when model performance is high, whereas other types of test sets cannot. The comparison between round-trip and single-pass translations reflects the importance of knowledge from language-native resources. Finally, we demonstrate that regularization is beneficial to bridging this gap on structured but not generative tasks.

Humans, on the other hand, need none of this. On the basis of very limited or incomplete data, we nonetheless come to the right conclusion about many things (yes, we are fallible, but the miracle is that we are right so often). Noam Chomsky's entire claim to fame in linguistics really amounts to exploring this underdetermination problem, which he referred to as "the poverty of the stimulus." Humans pick up language despite very varied experiences with other human language speakers. Babies born in abusive and sensory deprived environments pick up language.