Independent Component Analysis (ICA) is an algorithm originally developed for finding separate sources in a mixed signal, such as a recording of multiple people in the same room speaking at the same time. Unlike Principal Component Analysis (PCA), ICA permits the representation of a word as an unstructured set of features, without any particular feature being deemed more significant than the others. In this paper, we used ICA to analyze word embeddings. We have found that ICA can be used to find semantic features of the words and these features can easily be combined to search for words that satisfy the combination. We show that most of the independent components represent such features. To quantify the interpretability of the components, we use the word intruder test, performed both by humans and by large language models. We propose to use the automated version of the word intruder test as a fast and inexpensive way of quantifying vector interpretability without the need for human effort.

Multilingual language models have recently gained attention as a promising solution for representing multiple languages in a single model. In this paper, we propose new criteria to evaluate the quality of lexical representation and vocabulary overlap observed in sub-word tokenizers.Our findings show that the overlap of vocabulary across languages can be actually detrimental to certain downstream tasks (POS, dependency tree labeling). In contrast, NER and sentence-level tasks (cross-lingual retrieval, NLI) benefit from sharing vocabulary. We also observe that the coverage of the language-specific tokens in the multilingual vocabulary significantly impacts the word-level tasks. Our study offers a deeper understanding of the role of tokenizers in multilingual language models and guidelines for future model developers to choose the most suitable tokenizer for their specific application before undertaking costly model pre-training.

Recent studies have shown that transformer models like BERT rely on number information encoded in their representations of sentences’ subjects and head verbs when performing subject-verb agreement. However, probing experiments suggest that subject number is also encoded in the representations of all words in such sentences. In this paper, we use causal interventions to show that BERT only uses the subject plurality information encoded in its representations of the subject and words that agree with it in number. We also demonstrate that current probing metrics are unable to determine which words’ representations contain functionally relevant information. This both provides a revised view of subject-verb agreement in language models, and suggests potential pitfalls for current probe usage and evaluation.

The representations in large language models contain multiple types of gender information. We focus on two types of such signals in English texts: factual gender information, which is a grammatical or semantic property, and gender bias, which is the correlation between a word and specific gender. We can disentangle the model’s embeddings and identify components encoding both types of information with probing. We aim to diminish the stereotypical bias in the representations while preserving the factual gender signal. Our filtering method shows that it is possible to decrease the bias of gender-neutral profession names without significant deterioration of language modeling capabilities. The findings can be applied to language generation to mitigate reliance on stereotypes while preserving gender agreement in coreferences.

We present a free online demo of THEaiTRobot, an open-source bilingual tool for interactively generating theatre play scripts, in two versions. THEaiTRobot 1.0 uses the GPT-2 language model with minimal adjustments. THEaiTRobot 2.0 uses two models created by fine-tuning GPT-2 on purposefully collected and processed datasets and several other components, generating play scripts in a hierarchical fashion (title → synopsis → script). The underlying tool is used in the THEaiTRE project to generate scripts for plays, which are then performed on stage by a professional theatre.

We experiment with adapting generative language models for the generation of long coherent narratives in the form of theatre plays. Since fully automatic generation of whole plays is not currently feasible, we created an interactive tool that allows a human user to steer the generation somewhat while minimizing intervention. We pursue two approaches to long-text generation: a flat generation with summarization of context, and a hierarchical text-to-text two-stage approach, where a synopsis is generated first and then used to condition generation of the final script. Our preliminary results and discussions with theatre professionals show improvements over vanilla language model generation, but also identify important limitations of our approach.

With the recent success of pre-trained models in NLP, a significant focus was put on interpreting their representations. One of the most prominent approaches is structural probing (Hewitt and Manning, 2019), where a linear projection of word embeddings is performed in order to approximate the topology of dependency structures. In this work, we introduce a new type of structural probing, where the linear projection is decomposed into 1. iso-morphic space rotation; 2. linear scaling that identifies and scales the most relevant dimensions. In addition to syntactic dependency, we evaluate our method on two novel tasks (lexical hypernymy and position in a sentence). We jointly train the probes for multiple tasks and experimentally show that lexical and syntactic information is separated in the representations. Moreover, the orthogonal constraint makes the Structural Probes less vulnerable to memorization.

The high performance of large pretrained language models (LLMs) such as BERT on NLP tasks has prompted questions about BERT’s linguistic capabilities, and how they differ from humans’. In this paper, we approach this question by examining BERT’s knowledge of lexical semantic relations. We focus on hypernymy, the “is-a” relation that relates a word to a superordinate category. We use a prompting methodology to simply ask BERT what the hypernym of a given word is. We find that, in a setting where all hypernyms are guessable via prompting, BERT knows hypernyms with up to 57% accuracy. Moreover, BERT with prompting outperforms other unsupervised models for hypernym discovery even in an unconstrained scenario. However, BERT’s predictions and performance on a dataset containing uncommon hyponyms and hypernyms indicate that its knowledge of hypernymy is still limited.

State-of-the-art contextual embeddings are obtained from large language models available only for a few languages. For others, we need to learn representations using a multilingual model. There is an ongoing debate on whether multilingual embeddings can be aligned in a space shared across many languages. The novel Orthogonal Structural Probe (Limisiewicz and Mareček, 2021) allows us to answer this question for specific linguistic features and learn a projection based only on mono-lingual annotated datasets. We evaluate syntactic (UD) and lexical (WordNet) structural information encoded inmBERT’s contextual representations for nine diverse languages. We observe that for languages closely related to English, no transformation is needed. The evaluated information is encoded in a shared cross-lingual embedding space. For other languages, it is beneficial to apply orthogonal transformation learned separately for each language. We successfully apply our findings to zero-shot and few-shot cross-lingual parsing.

This work focuses on analyzing the form and extent of syntactic abstraction captured by BERT by extracting labeled dependency trees from self-attentions. Previous work showed that individual BERT heads tend to encode particular dependency relation types. We extend these findings by explicitly comparing BERT relations to Universal Dependencies (UD) annotations, showing that they often do not match one-to-one. We suggest a method for relation identification and syntactic tree construction. Our approach produces significantly more consistent dependency trees than previous work, showing that it better explains the syntactic abstractions in BERT. At the same time, it can be successfully applied with only a minimal amount of supervision and generalizes well across languages.

Derivation is a type of a word-formation process which creates new words from existing ones by adding, changing or deleting affixes. In this paper, we explore the potential of word embeddings to identify properties of word derivations in the morphologically rich Czech language. We extract derivational relations between pairs of words from DeriNet, a Czech lexical network, which organizes almost one million Czech lemmas into derivational trees. For each such pair, we compute the difference of the embeddings of the two words, and perform unsupervised clustering of the resulting vectors. Our results show that these clusters largely match manually annotated semantic categories of the derivational relations (e.g. the relation ‘bake–baker’ belongs to category ‘actor’, and a correct clustering puts it into the same cluster as ‘govern–governor’).

We inspect the multi-head self-attention in Transformer NMT encoders for three source languages, looking for patterns that could have a syntactic interpretation. In many of the attention heads, we frequently find sequences of consecutive states attending to the same position, which resemble syntactic phrases. We propose a transparent deterministic method of quantifying the amount of syntactic information present in the self-attentions, based on automatically building and evaluating phrase-structure trees from the phrase-like sequences. We compare the resulting trees to existing constituency treebanks, both manually and by computing precision and recall.

This is a system description paper for the CUNI x-ling submission to the CoNLL 2018 UD Shared Task. We focused on parsing under-resourced languages, with no or little training data available. We employed a wide range of approaches, including simple word-based treebank translation, combination of delexicalized parsers, and exploitation of available morphological dictionaries, with a dedicated setup tailored to each of the languages. In the official evaluation, our submission was identified as the clear winner of the Low-resource languages category.

This is a work in progress about extracting the sentence tree structures from the encoder’s self-attention weights, when translating into another language using the Transformer neural network architecture. We visualize the structures and discuss their characteristics with respect to the existing syntactic theories and annotations.

In multi-source sequence-to-sequence tasks, the attention mechanism can be modeled in several ways. This topic has been thoroughly studied on recurrent architectures. In this paper, we extend the previous work to the encoder-decoder attention in the Transformer architecture. We propose four different input combination strategies for the encoder-decoder attention: serial, parallel, flat, and hierarchical. We evaluate our methods on tasks of multimodal translation and translation with multiple source languages. The experiments show that the models are able to use multiple sources and improve over single source baselines.

We once had a corp, or should we say, it once had us They showed us its tags, isn’t it great, unified tags They asked us to parse and they told us to use everything So we looked around and we noticed there was near nothing We took other langs, bitext aligned: words one-to-one We played for two weeks, and then they said, here is the test The parser kept training till morning, just until deadline So we had to wait and hope what we get would be just fine And, when we awoke, the results were done, we saw we’d won So, we wrote this paper, isn’t it good, Norwegian wood.

In this paper, we describe an improvement on the task of giving instructions to robots in a simulated block world using unrestricted natural language commands.

Part-of-speech (POS) induction is one of the most popular tasks in research on unsupervised NLP. Various unsupervised and semi-supervised methods have been proposed to tag an unseen language. However, many of them require some partial understanding of the target language because they rely on dictionaries or parallel corpora such as the Bible. In this paper, we propose a different method named delexicalized tagging, for which we only need a raw corpus of the target language. We transfer tagging models trained on annotated corpora of one or more resource-rich languages. We employ language-independent features such as word length, frequency, neighborhood entropy, character classes (alphabetic vs. numeric vs. punctuation) etc. We demonstrate that such features can, to certain extent, serve as predictors of the part of speech, represented by the universal POS tag.

We present HamleDT 2.0 (HArmonized Multi-LanguagE Dependency Treebank). HamleDT 2.0 is a collection of 30 existing treebanks harmonized into a common annotation style, the Prague Dependencies, and further transformed into Stanford Dependencies, a treebank annotation style that became popular in recent years. We use the newest basic Universal Stanford Dependencies, without added language-specific subtypes. We describe both of the annotation styles, including adjustments that were necessary to make, and provide details about the conversion process. We also discuss the differences between the two styles, evaluating their advantages and disadvantages, and note the effects of the differences on the conversion. We regard the stanfordization as generally successful, although we admit several shortcomings, especially in the distinction between direct and indirect objects, that have to be addressed in future. We release part of HamleDT 2.0 freely; we are not allowed to redistribute the whole dataset, but we do provide the conversion pipeline.

We propose HamleDT ― HArmonized Multi-LanguagE Dependency Treebank. HamleDT is a compilation of existing dependency treebanks (or dependency conversions of other treebanks), transformed so that they all conform to the same annotation style. While the license terms prevent us from directly redistributing the corpora, most of them are easily acquirable for research purposes. What we provide instead is the software that normalizes tree structures in the data obtained by the user from their original providers.

CzEng 1.0 is an updated release of our Czech-English parallel corpus, freely available for non-commercial research or educational purposes. In this release, we approximately doubled the corpus size, reaching 15 million sentence pairs (about 200 million tokens per language). More importantly, we carefully filtered the data to reduce the amount of non-matching sentence pairs. CzEng 1.0 is automatically aligned at the level of sentences as well as words. We provide not only the plain text representation, but also automatic morphological tags, surface syntactic as well as deep syntactic dependency parse trees and automatic co-reference links in both English and Czech. This paper describes key properties of the released resource including the distribution of text domains, the corpus data formats, and a toolkit to handle the provided rich annotation. We also summarize the procedure of the rich annotation (incl. co-reference resolution) and of the automatic filtering. Finally, we provide some suggestions on exploiting such an automatically annotated sentence-parallel corpus.