Sentence Ordering (SO) is a linguistic task which requires re-ordering of shuffled sentences into a coherent paragraph. SO has downstream applications, but also serves as a semantic probe for computational models as this capability is essential for understanding narrative structures, causal and temporal relations within texts. Despite its importance, prior research has been limited to predictable English language structures and has not thoroughly addressed the complexities of multilingual and varied narrative contexts. To fill this gap, we introduce a novel and comprehensive Multilingual Sentence Ordering task that extends SO to diverse narratives across 12 languages, including challenging code-switched texts. We have developed MultiSO, a new benchmark dataset that represents these challenges. Our findings reveal that both specialized sentence ordering models and advanced Large Language Models like GPT-4 face significant challenges with this task.

The positive effect of adding subword information to word embeddings has been demonstrated for predictive models. In this paper we investigate whether similar benefits can also be derived from incorporating subwords into counting models. We evaluate the impact of different types of subwords (n-grams and unsupervised morphemes), with results confirming the importance of subword information in learning representations of rare and out-of-vocabulary words.

Increasing the capacity of recurrent neural networks (RNN) usually involves augmenting the size of the hidden layer, with significant increase of computational cost. Recurrent neural tensor networks (RNTN) increase capacity using distinct hidden layer weights for each word, but with greater costs in memory usage. In this paper, we introduce restricted recurrent neural tensor networks (r-RNTN) which reserve distinct hidden layer weights for frequent vocabulary words while sharing a single set of weights for infrequent words. Perplexity evaluations show that for fixed hidden layer sizes, r-RNTNs improve language model performance over RNNs using only a small fraction of the parameters of unrestricted RNTNs. These results hold for r-RNTNs using Gated Recurrent Units and Long Short-Term Memory.